File: H5.format.html

package info (click to toggle)
hdf5 1.8.13%2Bdocs-15
links: PTS, VCS
area: main
in suites: jessie-kfreebsd
size: 171,520 kB
sloc: ansic: 387,158; f90: 35,195; sh: 20,035; xml: 17,780; cpp: 13,516; makefile: 1,487; perl: 1,299; yacc: 327; lex: 178; ruby: 37
file content (13722 lines) | stat: -rw-r--r-- 392,389 bytes
parent folder | download | duplicates (2)
<html>
  <head>
    <title>
      HDF5 File Format Specification Version 2.0
    </title>

<STYLE TYPE="text/css">

P { text-indent: 2em}
P.item { margin-left: 2em; text-indent: -2em}
P.item2 { margin-left: 2em; text-indent: 2em}

TABLE.format { border:solid; border-collapse:collapse; caption-side:top; text-align:center; width:80%;}
TABLE.format TH { border:ridge; padding:4px; width:25%;}
TABLE.format TD { border:ridge; padding:4px; }
TABLE.format CAPTION { font-weight:bold; font-size:larger;}

TABLE.note {border:none; text-align:right; width:80%;}

TABLE.desc { border:solid; border-collapse:collapse; caption-size:top; text-align:left; width:80%;}
TABLE.desc TR { vertical-align:top;}
TABLE.desc TH { border-style:ridge; font-size:larger; padding:4px; text-decoration:underline;}
TABLE.desc TD { border-style:ridge; padding:4px; }
TABLE.desc CAPTION { font-weight:bold; font-size:larger;}

TABLE.list { border:none; }
TABLE.list TR { vertical-align:top;}
TABLE.list TH { border:none; text-decoration:underline;}
TABLE.list TD { border:none; }

</STYLE>
       
<!-- #BeginLibraryItem "/ed_libs/styles_Format.lbi" -->
<!--
  * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
  * Copyright by The HDF Group.                                               *
  * Copyright by the Board of Trustees of the University of Illinois.         *
  * All rights reserved.                                                      *
  *                                                                           *
  * This file is part of HDF5.  The full HDF5 copyright notice, including     *
  * terms governing use, modification, and redistribution, is contained in    *
  * the files COPYING and Copyright.html.  COPYING can be found at the root   *
  * of the source code distribution tree; Copyright.html can be found at the  *
  * root level of an installed copy of the electronic HDF5 document set and   *
  * is linked from the top-level documents page.  It can also be found at     *
  * http://www.hdfgroup.org/HDF5/doc/Copyright.html.  If you do not have      *
  * access to either file, you may request a copy from help@hdfgroup.org.     *
  * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
 -->
<!-- #EndLibraryItem --><!-- #BeginLibraryItem "/ed_libs/NavBar_ADevG.lbi" -->
<hr>
<center>
<table border=0 width=98%>
<tr><td valign=top align=left>
    <a href="index.html">HDF5 documents and links</a>&nbsp;<br>
    <a href="H5.intro.html">Introduction to HDF5</a>&nbsp;<br>
    <!--
    <a href="Glossary.html">Glossary</a><br>
    -->
</td>
<td valign=top align=right>
    <a href="UG/index.html">HDF5 User's Guide</a>&nbsp;<br>
    <a href="RM/RM_H5Front.html">HDF5 Reference Manual</a>&nbsp;<br>
    <a href="ADGuide.html">HDF5 Application Developer's Guide</a>&nbsp;<br>
</td></tr>
</table>
</center>
<hr>
<!-- #EndLibraryItem --><center><h1>HDF5 File Format Specification Version 2.0</h1></center>

    <center>
    <table border=0 width=90%>
    <tr>
    <td valign=top>
    <ol type=I>
      <li><a href="#Intro">Introduction</a>
      <li><a href="#FileMetaData">Disk Format Level 0 - File Metadata</a>
        <font size=-2>
	<ol type=A>
	  <li><a href="#Superblock">Disk Format Level 0A - Format Signature and Superblock</a>
	  <li><a href="#DriverInfo">Disk Format Level 0B - File Driver Info</a>
	  <li><a href="#SuperblockExt">Disk Format Level 0C - Superblock Extension</a>
	</ol>
        </font>
      <li><a href="#FileInfra">Disk Format Level 1 - File Infrastructure</a>
        <font size=-2>
	<ol type=A>
	  <li><a href="#Btrees">Disk Format Level 1A - B-Trees</a>
	    <ol type=1>
              <li><a href="#V1Btrees">Version 1 B-Trees (B-link trees)</a>
              <li><a href="#V2Btrees">Version 2 B-Trees</a>
            </ol>
	  <li><a href="#SymbolTable">Disk Format Level 1B - Group Symbol Table</a>
	  <li><a href="#SymbolTableEntry">Disk Format Level 1C - Group Symbol Table Entry</a>
	  <li><a href="#LocalHeap">Disk Format Level 1D - Local Heaps</a>
	  <li><a href="#GlobalHeap">Disk Format Level 1E - Global Heap</a>
	  <li><a href="#FractalHeap">Disk Format Level 1F - Fractal Heap</a>
	  <li><a href="#FreeSpaceManager">Disk Format Level 1G - Free-Space Manager</a>
	  <li><a href="#SOHMTable">Disk Format Level 1H - Shared Object Header Message Table</a>
	</ol>
        </font>
      <li><a href="#DataObject">Disk Format Level 2 - Data Objects</a>
        <font size=-2>
	<ol type=A>
	  <li><a href="#ObjectHeader">Disk Format Level 2A - Data Object Headers</a>
            <ol type=1>
              <li><a href="#ObjectHeaderPrefix">Disk Format Level 2A1 - Data Object Header Prefix</a>
                <ol type=a>
                  <li><a href="#V1ObjectHeaderPrefix">Version 1 Data Object Header Prefix</a>
                  <li><a href="#V2ObjectHeaderPrefix">Version 2 Data Object Header Prefix</a>
                </ol>
              <li><a href="#ObjectHeaderMessages">Disk Format Level 2A2 - Data Object Header Messages</a>
                <ol type=a>
                  <li><a href="#NILMessage">NIL</a>                                                       <!-- 0x0000 -->
                  <li><a href="#DataspaceMessage">Dataspace</a>                                           <!-- 0x0001 -->
                </ol>
            </ol>
        </ol>
        </font>
    </ol>
    </td>
    
    <td>&nbsp;</td>

    <td valign=top>
    <ol type=I start=4>
      <li><a href="#DataObject">Disk Format Level 2 - Data Objects</a> 
        <font size=-2><i>(Continued)</i>
	<ol type=A>
	  <li><a href="#ObjectHeader">Disk Format Level 2A - Data Object Headers</a><i>(Continued)</i>
	    <ol type=1 start=2>
              <li><a href="#ObjectHeaderMessages">Disk Format Level 2A2 - Data Object Header Messages</a><i>(Continued)</i>
                <ol type=a start=3>
                  <li><a href="#LinkInfoMessage">Link Info</a>                                     	  <!-- 0x0002 -->
                  <li><a href="#DatatypeMessage">Datatype</a>                                             <!-- 0x0003 -->
                  <li><a href="#OldFillValueMessage">Data Storage - Fill Value (Old)</a>                  <!-- 0x0004 -->
                  <li><a href="#FillValueMessage">Data Storage - Fill Value</a>                           <!-- 0x0005 -->
                  <li><a href="#LinkMessage">Link Message</a>                                             <!-- 0x0006 -->
                  <li><a href="#ExternalFileListMessage">Data Storage - External Data Files</a>           <!-- 0x0007 -->
                  <li><a href="#LayoutMessage">Data Storage - Layout</a>                                  <!-- 0x0008 -->
                  <li><a href="#BogusMessage">Bogus</a>                                                   <!-- 0x0009 -->
                  <li><a href="#GroupInfoMessage">Group Info</a>                                          <!-- 0x000a -->
                  <li><a href="#FilterMessage">Data Storage - Filter Pipeline</a>                	  <!-- 0x000b -->
                  <li><a href="#AttributeMessage">Attribute</a>                                           <!-- 0x000c -->
                  <li><a href="#CommentMessage">Object Comment</a>                                        <!-- 0x000d -->
                  <li><a href="#OldModificationTimeMessage">Object Modification Time (Old)</a>            <!-- 0x000e -->
                  <li><a href="#SOHMTableMessage">Shared Message Table</a>                                <!-- 0x000f -->
                  <li><a href="#ContinuationMessage">Object Header Continuation</a>                       <!-- 0x0010 -->
                  <li><a href="#SymbolTableMessage">Symbol Table</a>                                      <!-- 0x0011 -->
                  <li><a href="#ModificationTimeMessage">Object Modification Time</a>                     <!-- 0x0012 -->
                  <li><a href="#BtreeKValuesMessage">B-tree 'K' Values</a>                    	          <!-- 0x0013 -->
                  <li><a href="#DrvInfoMessage">Driver Info</a>                    		          <!-- 0x0014 -->
                  <li><a href="#AinfoMessage">Attribute Info</a>                    		          <!-- 0x0015 -->
                  <li><a href="#RefCountMessage">Object Reference Count</a>                    		  <!-- 0x0016 -->
                </ol>
	    </ol>
	  <li><a href="#DataStorage">Disk Format: Level 2B - Data Object Data Storage</a>
	</ol>
        </font>
      <LI><A href="#Appendix">Appendix</A>
    </ol>
</td></tr>
</table>
</center>

    <BR>
    <HR>


    <a name="Intro"><h2>Introduction</h2></a>

    <table align=right width=100>
    <tr><td>&nbsp;</td><td align=center>
        <hr>
        <img src="FF-IH_FileGroup.gif" alt="HDF5 Groups" hspace=15 vspace=15>
    </td><td>&nbsp;</td></tr>
    <tr><td>&nbsp;</td><td align=center>
        <strong>Figure 1:</strong> Relationships among the HDF5 root group, other groups, and objects
        <hr>
    </td><td>&nbsp;</td></tr>

    <tr><td>&nbsp;</td><td align=center>
        <img src="FF-IH_FileObject.gif" alt="HDF5 Objects" hspace=15 vspace=15>
    </td><td>&nbsp;</td></tr>
    <tr><td>&nbsp;</td><td align=center>
        <strong>Figure 2:</strong> HDF5 objects -- datasets, datatypes, or dataspaces
        <hr>
    </td><td>&nbsp;</td></tr>
    </table>


    <P>The format of an HDF5 file on disk encompasses several
      key ideas of the HDF4 and AIO file formats as well as
      addressing some shortcomings therein.  The new format is
      more self-describing than the HDF4 format and is more
      uniformly applied to data objects in the file.
      
    <P>An HDF5 file appears to the user as a directed graph.  
      The nodes of this graph are the higher-level HDF5 objects
      that are exposed by the HDF5 APIs:

      <ul>
         <li>Groups
         <li>Datasets
         <li>Named datatypes
      </ul>

    <P>At the lowest level, as information is actually written to the disk, 
       an HDF5 file is made up of the following objects:
      <ul>
         <li>A superblock
         <li>B-tree nodes
         <li>Heap blocks
         <li>Object headers
         <li>Object data
         <li>Free space
      </ul>

    <P>The HDF5 library uses these low-level objects to represent the
      higher-level objects that are then presented to the user or 
      to applications through the APIs.
      For instance, a group is an object header that contains a message that
      points to a local heap (for storing the links to objects in the group)
      and to a B-tree (which indexes the links).
      A dataset is an object header that contains messages that describe
      datatype, dataspace, layout, filters, external files, fill value, etc
      with the layout message pointing to either a raw data chunk or to a
      B-tree that points to raw data chunks.


    <h3>This Document</h3>

    <p>This document describes the lower-level data objects;
      the higher-level objects and their properties are described 
      in the <a href="UG/index.html"><cite>HDF5 User's Guide</cite></a>.

    <P>Three levels of information comprise the file format.  
      Level 0 contains basic information for identifying and
      defining information about the file.  Level 1 information contains
      the information about the pieces of a file shared by many objects
      in the file (such as a B-trees and heaps).  Level 2 is the rest 
      of the file and contains all of the data objects, with each object
      partitioned into header information, also known as 
      <em>metadata</em>, and data.

    <p>The sizes of various fields in the following layout tables are
      determined by looking at the number of columns the field spans
      in the table.  There are three exceptions: (1) The size may be
      overridden by specifying a size in parentheses, (2) the size of
      addresses is determined by the <em>Size of Offsets</em> field
      in the superblock and is indicated in this document with a
      superscripted 'O', and (3) the size of length fields is determined
      by the <em>Size of Lengths</em> field in the superblock and is
      indicated in this document with a superscripted 'L'.

    <P>Values for all fields in this document should be treated as unsigned
      integers, unless otherwise noted in the description of a field.
      Additionally, all metadata fields are stored in little-endian byte
      order.  
    </P>
    
    <P>All checksums used in the format are computed with the
      <A href="http://www.burtleburtle.net/bob/hash/doobs.html">Jenkins'
      lookup3</A> algorithm.
    </P>

    <P>Various tables in this document aligned with "This space inserted
        only to align table nicely".  These entries in the table are just
        to make the table presentation nicer and do not represent any values
        or padding in the file.
    </P>

    <BR>
    <HR>

    <h2><a name="FileMetaData">
	Disk Format: Level 0 - File Metadata</a></h2>

    <H3><A name="Superblock">
	Disk Format: Level 0A - Format Signature and Superblock</A></H3>

    <P>The superblock may begin at certain predefined offsets within
      the HDF5 file, allowing a block of unspecified content for
      users to place additional information at the beginning (and
      end) of the HDF5 file without limiting the HDF5 library's
      ability to manage the objects within the file itself.  This
      feature was designed to accommodate wrapping an HDF5 file in
      another file format or adding descriptive information to an HDF5
      file without requiring the modification of the actual file's
      information.  The superblock is located by searching for the
      HDF5 format signature at byte offset 0, byte offset 512 and at
      successive locations in the file, each a multiple of two of
      the previous location, i.e.  0, 512, 1024, 2048, etc.

    <P>The superblock is composed of the format signature, followed by a
      superblock version number and information that is specific to each
      version of the superblock.
      Currently, there are three versions of the superblock format.
      Version 0 is the default format, while version 1 is basically the same
      as version 0 with additional information when a non-default B-tree 'K'
      value is stored.  Version 2 is the latest format, with some fields
      eliminated or compressed and with superblock extension and checksum
      support.

    <P>Version 0 and 1 of the superblock are described below:

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Superblock (Versions 0 and 1)
	</caption>

	<tr>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	</tr>  

	<tr>
	  <td colspan=4><br>Format Signature (8 bytes)<br><br></td>
	</tr>

	<tr>
	  <td>Version # of Superblock</td>
	  <td>Version # of File Free-space Storage</td>
	  <td>Version # of Root Group Symbol Table Entry</td>
	  <td>Reserved (zero)</td>
	</tr>

	<tr>
	  <td>Version # of Shared Header Message Format</td>
	  <td>Size of Offsets</td>
	  <td>Size of Lengths</td>
	  <td>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=2>Group Leaf Node K</td>
	  <td colspan=2>Group Internal Node K</td>
	</tr>

	<tr>
	  <td colspan=4>File Consistency Flags</td>
	</tr>

	<tr>
	  <td colspan=2 style="border:dotted;">Indexed Storage Internal Node K<sup>1</sup></td>
	  <td colspan=2 style="border:dotted;">Reserved (zero)<sup>1</sup></td>
	</tr>

	<tr>
	  <td colspan=4><br>Base Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Address of File Free-space Info<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>End of File Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Driver Information Block Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Root Group Symbol Table Entry</td>
	</tr>
      </table>

      <table class=note>
        <tr><td>
            (Items marked with an 'O' in the above table are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
        <tr><td>
            (Items marked with an '1' in the above table are
            <br>
            new in version 1 of the superblock)
        </td></tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Format Signature</td>
	  <td>
            <P>This field contains a constant value and can be used to
	    quickly identify a file as being an HDF5 file.  The
	    constant value is designed to allow easy identification of
	    an HDF5 file and to allow certain types of data corruption
	    to be detected.  The file signature of an HDF5 file always
	    contains the following values:
            </P>

            <center>
	      <table border align=center cellpadding=4>
		<tr align=center>
		  <td align=right>Decimal:</td>
                  <td width="8%">137</td>
		  <td width="8%">72</td>
		  <td width="8%">68</td>
		  <td width="8%">70</td>
		  <td width="8%">13</td>
		  <td width="8%">10</td>
		  <td width="8%">26</td>
		  <td width="8%">10</td>
		</tr>

		<tr align=center>
		  <td align=right>Hexadecimal:</td>
		  <td>89</td>
		  <td>48</td>
		  <td>44</td>
		  <td>46</td>
		  <td>0d</td>
		  <td>0a</td>
		  <td>1a</td>
		  <td>0a</td>
		</tr>

		<tr align=center>
		  <td align=right>ASCII C Notation:</td>
		  <td>\211</td>
		  <td>H</td>
		  <td>D</td>
		  <td>F</td>
		  <td>\r</td>
		  <td>\n</td>
		  <td>\032</td>
		  <td>\n</td>
		</tr>
	      </table>
	    </center>
	    <br>

	    <P>This signature both identifies the file as an HDF5 file
	    and provides for immediate detection of common
	    file-transfer problems. The first two bytes distinguish
	    HDF5 files on systems that expect the first two bytes to
	    identify the file type uniquely. The first byte is
	    chosen as a non-ASCII value to reduce the probability
	    that a text file may be misrecognized as an HDF5 file;
	    also, it catches bad file transfers that clear bit
	    7. Bytes two through four name the format. The CR-LF
	    sequence catches bad file transfers that alter newline
	    sequences. The control-Z character stops file display
	    under MS-DOS. The final line feed checks for the inverse
	    of the CR-LF translation problem.  (This is a direct
	    descendent of the <A href="http://www.libpng.org/pub/png/spec/iso/index-object.html#5PNG-file-signature">PNG</A> file
            signature.)
            </P>

            <P><EM>This field is present in version 0+ of the superblock.</EM>
            </P>
            </td>
	</tr>

	<tr>
	  <td>Version Number of the Superblock</td>
	  <td>
            <P>This value is used to determine the format of the
	    information in the superblock.  When the format of the
	    information in the superblock is changed, the version number
	    is incremented to the next integer and can be used to
	    determine how the information in the superblock is
	    formatted.
            </P>

            <P>Values of 0, 1 and 2 are defined for this field.  (The format
            of version 2 is described below, not here)
            </P>

            <P><EM>This field is present in version 0+ of the superblock.</EM>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Version Number of the File Free-Space Information</td>
	  <td>
            <P>This value is used to determine the format of the
	    information in the file's free-space Information.
            </P>
            <P>The only value currently valid in this field is '0', which
            indicates that the free space index is formatted as described 
            <A href="#FreeSpaceManager">below</A>.
            </P>

            <P><EM>This field is present in version 0 and 1 of the superblock.</EM>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Version Number of the Root Group Symbol Table Entry</td>
	  <td>
            <P>This value is used to determine the format of the
	    information in the Root Group Symbol Table Entry.  When the
            format of the information in that field is changed, the
	    version number is incremented to the next integer and can be
	    used to determine how the information in the field
	    is formatted.
            </P>
            <P>The only value currently valid in this field is '0', which
            indicates that the root group symbol table entry is formatted as
            described <A href="#SymbolTableEntry">below</A>.
            </P>

            <P><EM>This field is present in version 0 and 1 of the superblock.</EM>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Version Number of the Shared Header Message Format</td>
	  <td>
            <P>This value is used to determine the format of the
	    information in a shared object header message. Since the format
	    of the shared header messages differs from the other private
	    header messages, a version number is used to identify changes
	    in the format.
            </P>
            <P>The only value currently valid in this field is '0', which
            indicates that shared header messages are formatted as
            described <A href="#ObjectHeaderMessages">below</A>.
            </P>

            <P><EM>This field is present in version 0 and 1 of the superblock.</EM>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size of Offsets</td>
	  <td>
            <P>This value contains the number of bytes used to store
	    addresses in the file.  The values for the addresses of
	    objects in the file are offsets relative to a base address,
	    usually the address of the superblock signature.  This
	    allows a wrapper to be added after the file is created
	    without invalidating the internal offset locations.
            </P>

            <P><EM>This field is present in version 0+ of the superblock.</EM>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size of Lengths</td>
	  <td>
            <P>This value contains the number of bytes used to store
	    the size of an object.
            </P>

            <P><EM>This field is present in version 0+ of the superblock.</EM>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Group Leaf Node K</td>
	  <td>
            <P>Each leaf node of a group B-tree will have at
	    least this many entries but not more than twice this
	    many.  If a group has a single leaf node then it
	    may have fewer entries.
            </P>
            <P>This value must be greater than zero.
            </P>
            <P>See the <A href="#Btrees">description</A> of B-trees below.
            </P>

            <P><EM>This field is present in version 0 and 1 of the superblock.</EM>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Group Internal Node K</td>
	  <td>
            <P>Each internal node of a group B-tree will have at
	    least this many entries but not more than twice this
	    many.  If the group has only one internal
	    node then it might have fewer entries.
            </P>
            <P>This value must be greater than zero.
            </P>
            <P>See the <A href="#Btrees">description</A> of B-trees below.
            </P>

            <P><EM>This field is present in version 0 and 1 of the superblock.</EM>
            </P>
          </td>
	</tr>

        <tr>
          <td>File Consistency Flags</td>
          <td>
            <P>This value contains flags to indicate information
            about the consistency of the information contained
            within the file.  Currently, the following bit flags are
            defined: 
            <ul>
            <li>Bit 0 set indicates that the file is opened for
            write-access.
            <li>Bit 1 set indicates that the file has
            been verified for consistency and is guaranteed to be
            consistent with the format defined in this document.
            <li>Bits 2-31 are reserved for future use.
            </ul>
            Bit 0 should be
            set as the first action when a file is opened for write
            access and should be cleared only as the final action
            when closing a file.  Bit 1 should be cleared during
            normal access to a file and only set after the file's
            consistency is guaranteed by the library or a
            consistency utility.
            </P>

            <P><EM>This field is present in version 0+ of the superblock.</EM>
            </P>
          </td>
        </tr>

	<tr>
	  <td>Indexed Storage Internal Node K</td>
	  <td>
            <P>Each internal node of an indexed storage B-tree will have at
	    least this many entries but not more than twice this
	    many.  If the index storage B-tree has only one internal
	    node then it might have fewer entries.
            </P>
            <P>This value must be greater than zero.
            </P>
            <P>See the <A href="#Btrees">description</A> of B-trees below.
            </P>

            <P><EM>This field is present in version 1 of the superblock.</EM>
            </P>
          </td>
	</tr>

        <tr>
          <td>Base Address</td>
          <td>
            <P>This is the absolute file address of the first byte of
            the HDF5 data within the file.  The library currently 
            constrains this value to be the absolute file address 
            of the superblock itself when creating new files; 
            future versions of the library may provide greater 
            flexibility.  When opening an existing file and this address does
            not match the offset of the superblock, the library assumes
            that the entire contents of the HDF5 file have been adjusted in
            the file and adjusts the base address and end of file address to
            reflect their new positions in the file.  Unless otherwise noted,
            all other file addresses are relative to this base
            address.
            </P>

            <P><EM>This field is present in version 0+ of the superblock.</EM>
            </P>
          </td>
        </tr>

        <tr>
          <td>Address of Global Free-space Index</td>
          <td>
            <P>Free-space management is not yet defined in the HDF5
            file format and is not handled by the library.
            Currently this field always contains the 
            <A href="#UndefinedAddress">undefined address</A>. 
            </P>

            <P><EM>This field is present in version 0 and 1 of the superblock.</EM>
            </P>
          </td>
        </tr>

        <tr>
          <td>End of File Address</td>
          <td>
            <P>This is the absolute file address of the first byte past
            the end of all HDF5 data.  It is used to determine whether a
            file has been accidently truncated and as an address where
            file data allocation can occur if space from the free list is
            not used.
            </P>

            <P><EM>This field is present in version 0+ of the superblock.</EM>
            </P>
          </td>
        </tr>

        <tr>
          <td>Driver Information Block Address</td>
          <td>
            <P>This is the relative file address of the file driver
            information block which contains driver-specific
            information needed to reopen the file. If there is no
            driver information block then this entry should be the
            <A href="#UndefinedAddress">undefined address</A>.
            </P>

            <P><EM>This field is present in version 0 and 1 of the superblock.</EM>
            </P>
          </td>
        </tr>

        <tr>
          <td>Root Group Symbol Table Entry</td>
          <td>
            <P>This is the <A href="#SymbolTableEntry">symbol table entry</A>
            of the root group, which serves as the entry point into 
            the group graph for the file.
            </P>

            <P><EM>This field is present in version 0 and 1 of the superblock.</EM>
            </P>
          </td>
        </tr>
      </table>
    </div>

    <P>Version 2 of the superblock is described below:

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Superblock (Version 2)
	</caption>

	<tr>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	</tr>  

	<tr>
	  <td colspan=4><br>Format Signature (8 bytes)<br><br></td>
	</tr>

	<tr>
	  <td>Version # of Superblock</td>
	  <td>Size of Offsets</td>
	  <td>Size of Lengths</td>
	  <td>File Consistency Flags</td>
	</tr>

	<tr>
	  <td colspan=4><br>Base Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Superblock Extension Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>End of File Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Root Group Object Header Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Superblock Checksum</sup></td>
	</tr>
      </table>

      <table class=note>
        <tr><td>
            (Items marked with an 'O' in the above table are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
      </table>
    </div>
 
    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Format Signature</td>
	  <td>
            <P>This field is the same as described for versions 0 and 1 of the
                superblock.
            </P>
	</tr>

	<tr>
	  <td>Version Number of the Superblock</td>
	  <td>
            <P>This field has a value of 2 and has the same meaning as for
                versions 0 and 1.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size of Offsets</td>
	  <td>
            <P>This field is the same as described for versions 0 and 1 of the
                superblock.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size of Lengths</td>
	  <td>
            <P>This field is the same as described for versions 0 and 1 of the
                superblock.
            </P>
          </td>
	</tr>

        <tr>
          <td>File Consistency Flags</td>
          <td>
            <P>This field is the same as desribed for versions 0 and 1 except
            that it is smaller (the number of reserved bits has been reduced
            from 30 to 6).
            </P>
          </td>
        </tr>

        <tr>
          <td>Base Address</td>
          <td>
            <P>This field is the same as described for versions 0 and 1 of the
                superblock.
            </P>
          </td>
        </tr>

        <tr>
          <td>Superblock Extension Address</td>
          <td>
            <P>The field is the address of the object header for the
                <A href="#SuperblockExt">superblock extension</A>.
		If there is no extension then this entry should be the
		<A href="#UndefinedAddress">undefined address</A>.
            </P>
          </td>
        </tr>

        <tr>
          <td>End of File Address</td>
          <td>
            <P>This field is the same as described for versions 0 and 1 of the
                superblock.
            </P>
          </td>
        </tr>

        <tr>
          <td>Root Group Object Header Address</td>
          <td>
            <P>This is the address of 
		the <A href="#DataObject">root group object header</A>, 
		which serves as the entry point into the group graph for the file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Superblock Checksum</td>
          <td>
            <P>The checksum for the superblock.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <H3><A name="DriverInfo">
	Disk Format: Level 0B - File Driver Info</A></H3>

    <p>The <B>driver information block</B> is an optional region of the
      file which contains information needed by the file driver
      to reopen a file.  The format is described below:

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Driver Information Block
        </caption>
 
        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>  
 
        <tr>
          <td>Version</td>
          <td colspan=3>Reserved</td>
        </tr>

        <tr>
          <td colspan=4>Driver Information Size</td>
        </tr>

        <tr>
          <td colspan=4><br>Driver Identification (8 bytes)<br><br></td>
        </tr>

        <tr>
          <td colspan=4><br><br>Driver Information (<em>variable size</em>)<br><br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Version</td>
          <td>
            <P>The version number of the Driver Information Block.
	       This document describes version 0.
            </P>
          </td>
        </tr>

        <tr>
          <td>Driver Information Size</td>
          <td>
            <P>The size in bytes of the <em>Driver Information</em> field.
            </P>
          </td>
        </tr>

        <tr>
          <td>Driver Identification</td>
          <td>
            <P>This is an eight-byte ASCII string without null
            termination which identifies the driver and/or version number 
            of the Driver Information Block. The predefined driver encoded
            in this field by the HDF5 library is identified by the
            letters <code>NCSA</code> followed by the first four characters of
            the driver name. If the Driver Information block is not
            the original version then the last letter(s) of the
            identification will be replaced by a version number in
            ASCII, starting with 0.
            </P>
            <P>
            Identification for user-defined drivers is also eight-byte long.  
	    It can be arbitrary but should be unique to avoid 
            the four character prefix "NCSA".
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Driver Information</td>
          <td>Driver information is stored in a format defined by the
            file driver (see description below).</td>
        </tr>
      </table>
    </div>

    <BR>
    The two drivers encoded in the <em>Driver Identification</em> field are as follows:
    <UL>
    <LI>
    Multi driver:
    <P>
    The identifier for this driver is "NCSAmulti".  
    This driver provides a mechanism for segregating raw data and different types of meta data 
    into multiple files.
    These files are viewed by the library as a single virtual HDF5 file with a single file address.
    A maximum of 6 files will be created for the following data:
    superblock, B-tree, raw data, global heap, local heap, and object header.  
    More than one type of data can be written to the same file.
    </P>
    <LI>
    Family driver
    <P>
    The identifier for this driver is "NCSAfami" and is encoded in this field for library version 1.8 and after.
    This driver is designed for systems that do not support files larger than 2 gigabytes
    by splitting the HDF5 file address space across several smaller files. 
    It does nothing to segregate meta data and raw data; 
    they are mixed in the address space just as they would be in a single contiguous file.
    </P>
    </UL>
    <BR>
    The format of the <em>Driver Information</em> field for the above two drivers are described below:
    <BR>
    <BR>
    <div align=center>
      <table class=format>
	<caption>
	  Multi Driver Information
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Member Mapping</td>
	  <td>Member Mapping</td>
	  <td>Member Mapping</td>
	  <td>Member Mapping</td>
	</tr>

	<tr>
	  <td>Member Mapping</td>
	  <td>Member Mapping</td>
	  <td>Reserved</td>
	  <td>Reserved</td>
	</tr>

	<tr>
	  <td colspan=4><br>Address of Member File 1<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>End of Address for Member File 1<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Address of Member File 2<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>End of Address for Member File 2<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>... ...<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Address of Member File N<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>End of Address for Member File N<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Name of Member File 1 <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Name of Member File 2 <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>... ...<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Name of Member File N <em>(variable size)</em><br><br></td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
	  <th width="30%">Field Name</th>
	  <th>Description</th>
	</tr>

	<tr>
	  <td>Member Mapping</td>
	  <td>These fields are integer values from 1 to 6 
            indicating how the data can be mapped to or merged with another type of
            data.
	        <table class=list>
                    <tr>
                    <th width="30%">Member Mapping</th>
                    <th align=left>Description</th>
                    </tr>
                    <tr>
                        <td align=center>1</td>
                        <td>The superblock data.</td>
                    </tr>
                    <tr>
                        <td align=center>2</td>
                        <td>The B-tree data.</td>
                    </tr>
                    <tr>
                        <td align=center>3</td>
                        <td>The raw data.</td>
                    </tr>
                    <tr>
                        <td align=center>4</td>
                        <td>The global heap data.</td>
                    </tr>
                    <tr>
                        <td align=center>5</td>
                        <td>The local heap data.</td>
                    </tr>
                    <tr>
                        <td align=center>6</td>
                        <td>The object header data.</td>
                    </tr>
                </table></P>
            For example, if the third field has the value 3 and all the rest have the 
            value 1, it means there are two files: one for raw data, and one for superblock,
            B-tree, global heap, local heap, and object header.
          </td>
	</tr>

	<tr>
	  <td>Reserved</td>
	  <td><P>These fields are reserved and should always be zero.</P></td>
	</tr>

	<tr>
	  <td>Address of Member File N</td>
	  <td><P>This field Specifies the virtual address at which the member file starts.</P>
	      <P>N is the number of member files.</P>
	  </td>
	</tr>

	<tr>
	  <td>End of Address for Member File N</td>
	  <td><P>This field is the end of the allocated address for the member file.
            </P></td>
	</tr>

	<tr>
	  <td>Name of Member File N</td>
	  <td><P>This field is the null-terminated name of the member file and
		its length should be multiples of 8 bytes.  
		Additional bytes will be padded with <em>NULL</em>s. The default naming
		convention is <em>%s-X.h5</em>, where <em>X</em> is one of the letters 
            <em>s</em> (for superblock), <em>b</em> (for B-tree), <em>r</em> (for raw data),
            <em>g</em> (for global heap), <em>l</em> (for local heap), and <em>o</em> (for 
            object header). The name of the whole HDF5 file will substitute the <em>%s</em>
            in the string. 
          </P>
          </td> 
	</tr>
      </table>
    </div>

    <BR>
    <div align=center>
      <table class=format>
	<caption>
	  Family Driver Information
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td colspan=8><br>Size of member file<br><br></td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
	  <th width="30%">Field Name</th>
	  <th>Description</th>
	</tr>

	<tr>
	  <td>Size of member file</td>
	  <td><P>This field is the size of the member file in the family of files.</P></td>
	</tr>
      </table>
    </div>

    <br>
    <H3><A name="SuperblockExt">
	Disk Format: Level 0C - Superblock Extension</A></H3>

    <p>The <em>superblock extension</em> is used to store superblock metadata
      which is either optional, or added after the version of the superblock
      was defined.  Superblock extensions may only exist when version 2+ of
      superblock is used.  A superblock extension is an object header which may
      hold the following messages:
	<UL>
	<LI>
	<A href="#SOHMTableMessage">Shared Message Table message</A> containing
        information to locate the master table of shared object header message
        indices.
	<LI>
	<A href="#BtreeKValuesMessage">B-tree 'K' Values message</A> containing
        non-default B-tree 'K' values.
	<LI>
	<A href="#DrvInfoMessage">Driver Info message</A> containing information
        needed by the file driver in order to reopen a file.
	</UL>

    <BR>
    <HR>

    <h2><a name="FileInfra">
        Disk Format: Level 1 - File Infrastructure</a></h2>
    <h3><a name="Btrees">
        Disk Format: Level 1A - B-Trees and B-Tree Nodes</a></h3>

    <p>B-Trees allow flexible storage for objects which tend to grow
      in ways that cause the object to be stored discontiguously.  B-trees
      are described in various algorithms books including "Introduction to
      Algorithms" by Thomas H. Cormen, Charles E. Leiserson, and Ronald
      L. Rivest.  B-trees are used in several places in the HDF5 file format,
      when an index is needed for another data structure.

    <p>The version 1 B-tree structure described below is the original index
      structure, but are limited by some bugs in our implementation (mainly in
      how they handle deleting records).  The version 1 B-trees are being phased
      out in favor of the version 2 B-trees described below, although both
      types of structures may be found in the same file, depending on
      application settings when creating the file.

    <h4><a name="V1Btrees">
        Disk Format: Level 1A1 - Version 1 B-trees</a></h4>

    <p>Version 1 B-trees in HDF5 files an implementation of the B-link tree,
      in which the sibling nodes at a particular level in the tree are stored
      in a doubly-linked list, is described in the "Efficient Locking for
      Concurrent Operations on B-trees" paper by Phillip Lehman and S. Bing Yao
      as published in the <cite>ACM Transactions on Database Systems</cite>,
      Vol. 6, No. 4, December 1981.

    <p>The B-link trees implemented by the file format contain one more
      key than the number of children.  In other words, each child
      pointer out of a B-tree node has a left key and a right key.
      The pointers out of internal nodes point to sub-trees while
      the pointers out of leaf nodes point to symbol nodes and 
      raw data chunks.
      Aside from that difference, internal nodes and leaf nodes
      are identical.

    <br>
    <div align=center>
      <table class=format>
        <caption>
          B-link Tree Nodes
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>

        <tr>
          <td colspan=4>Signature</td>

        <tr>
          <td>Node Type</td>
          <td>Node Level</td>
          <td colspan=2>Entries Used</td>

        <tr>
          <td colspan=4><br>Address of Left Sibling<sup>O</sup><br><br></td>

        <tr>
          <td colspan=4><br>Address of Right Sibling<sup>O</sup><br><br></td>

        <tr>
          <td colspan=4>Key 0 (variable size)</td>

        <tr>
          <td colspan=4><br>Address of Child 0<sup>O</sup><br><br></td>

        <tr>
          <td colspan=4>Key 1 (variable size)</td>

        <tr>
          <td colspan=4><br>Address of Child 1<sup>O</sup><br><br></td>

        <tr>
          <td colspan=4>...</td>

        <tr>
          <td colspan=4>Key 2<em>K</em> (variable size)</td>

        <tr>
          <td colspan=4><br>Address of Child 2<em>K</em><sup>O</sup><br><br></td>

        <tr>
          <td colspan=4>Key 2<em>K</em>+1 (variable size)</td>
      </table>

      <table class=note>
        <tr><td>
            (Items marked with an 'O' the above table are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>TREE</code>" is 
            used to indicate the
            beginning of a B-link tree node.  This gives file
            consistency checking utilities a better chance of
            reconstructing a damaged file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Node Type</td>
          <td>
            <P>Each B-link tree points to a particular type of data.
            This field indicates the type of data as well as
            implying the maximum degree <em>K</em> of the tree and
            the size of each Key field.
            </P>

            <table class=list>
                <tr>
                <th width="30%">Node Type</th>
                <th align=left>Description</th>
                </tr>
                <tr>
                    <td align=center>0</td>
                    <td>This tree points to group nodes.</td>
                </tr>
                <tr>
                    <td align=center>1</td>
                    <td>This tree points to raw data chunk nodes.</td>
                </tr>
            </table>
          </td>
        </tr>

        <tr>
          <td>Node Level</td>
          <td>
            <P>The node level indicates the level at which this node
            appears in the tree (leaf nodes are at level zero).  Not
            only does the level indicate whether child pointers
            point to sub-trees or to data, but it can also be used
            to help file consistency checking utilities reconstruct
            damanged trees.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Entries Used</td>
          <td>
            <P>This determines the number of children to which this
            node points.  All nodes of a particular type of tree
            have the same maximum degree, but most nodes will point
            to less than that number of children.  The valid child
            pointers and keys appear at the beginning of the node
            and the unused pointers and keys appear at the end of
            the node.  The unused pointers and keys have undefined
            values.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Address of Left Sibling</td>
          <td>
            <P>This is the relative file address of the left sibling of
            the current node.  If the current
            node is the left-most node at this level then this field
            is the <A href="#UndefinedAddress">undefined address</A>.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Address of Right Sibling</td>
          <td>
            <P>This is the relative file address of the right sibling of
            the current node.  If the current
            node is the right-most node at this level then this
            field is the <A href="#UndefinedAddress">undefined address</A>.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Keys and Child Pointers</td>
          <td>
            <P>Each tree has 2<em>K</em>+1 keys with 2<em>K</em>
            child pointers interleaved between the keys.  The number
            of keys and child pointers actually containing valid
            values is determined by the node's <em>Entries Used</em> field.
            If that field is <em>N</em> then the B-link tree contains
            <em>N</em> child pointers and <em>N</em>+1 keys.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Key</td>
          <td>
            <P>The format and size of the key values is determined by
            the type of data to which this tree points.  The keys are
            ordered and are boundaries for the contents of the child
            pointer; that is, the key values represented by child
            <em>N</em> fall between Key <em>N</em> and Key
            <em>N</em>+1. Whether the interval is open or closed on
            each end is determined by the type of data to which the
            tree points.
            </P>

            <P>
            The format of the key depends on the node type.
            For nodes of node type 0 (group nodes), the key is formatted as
            follows:
            <center>
            <table class=list>
            <tr>
              <td width=30%>A single field of <i>Size of Lengths</i> 
                bytes:</td>
              <td>Indicates the byte offset into the local heap 
                for the first object name in the subtree which 
                that key describes.
              </td>
            </tr>
            </table>
            </center>
            </P>

            <P>
            For nodes of node type 1 (chunked raw data nodes), the key is
            formatted as follows:
            <center>
            <table class=list>
            <tr>
              <td width=30%>Bytes 1-4:</td>
              <td>Size of chunk in bytes.</td>
            </tr>
            <tr>
              <td>Bytes 4-8:</td>
              <td>Filter mask, a 32-bit bitfield indicating which 
                filters have been skipped for this chunk.  Each filter
                has an index number in the pipeline (starting at 0, with
                the first filter to apply) and if that filter is skipped,
                the bit corresponding to its index is set.</td>
            </tr>
            <tr>
              <td>(<em>D + 1</em>) 64-bit fields:</td>
              <td>The offset of the 
                chunk within the dataset where <i>D</i> is the number 
                of dimensions of the dataset, and the last value is the
                offset within the dataset's datatype and should always be
                zero.  For example, if 
                a chunk in a 3-dimensional dataset begins at the
                position <code>[5,5,5]</code>, there will be three
                such 64-bit values, each with the value of 
                <code>5</code>, followed by a <code>0</code> value.</td>
            </tr>
            </table>
            </center>
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Child Pointer</td>
          <td>
            <P>The tree node contains file addresses of subtrees or
            data depending on the node level.  Nodes at Level 0 point 
            to data addresses, either raw data chunks or group nodes.
            Nodes at non-zero levels point to other nodes of the 
            same B-tree.
            </P>
            <P>For raw data chunk nodes, the child pointer is the address
            of a single raw data chunk.  For group nodes, the child pointer
            points to a <A href="#SymbolTable">symbol table</A>, which contains
            information for multiple symbol table entries.
            </P>
          </td>
        </tr>
      </table>
    </div>

    <p>
     Conceptually, each B-tree node looks like this:
     <center>
     <table>
       <tr valign=top align=center>
         <td>key[0]</td><td>&nbsp;</td>
         <td>child[0]</td><td>&nbsp;</td>
         <td>key[1]</td><td>&nbsp;</td>
         <td>child[1]</td><td>&nbsp;</td>
         <td>key[2]</td><td>&nbsp;</td>
         <td>...</td><td>&nbsp;</td>
         <td>...</td><td>&nbsp;</td>
         <td>key[<i>N</i>-1]</td><td>&nbsp;</td>
         <td>child[<i>N</i>-1]</td><td>&nbsp;</td>
         <td>key[<i>N</i>]</td>
       </tr>
     </table>
     </center>
     <br>

     where child[<i>i</i>] is a pointer to a sub-tree (at a level 
     above Level 0) or to data (at Level 0).  
     Each key[<i>i</i>] describes an <i>item</i> stored by the B-tree 
     (a chunk or an object of a group node).  The range of values 
     represented by child[<i>i</i>] is indicated by key[<i>i</i>] 
     and key[<i>i</i>+1].


    <p>The following question must next be answered: 
     "Is the value described by key[<i>i</i>] contained in 
     child[<i>i</i>-1] or in child[<i>i</i>]?"  
     The answer depends on the type of tree. 
     In trees for groups (node type 0) the object described by 
     key[<i>i</i>] is the greatest object contained in 
     child[<i>i</i>-1] while in chunk trees (node type 1) the 
     chunk described by key[<i>i</i>] is the least chunk in 
     child[<i>i</i>].

    <p>That means that key[0] for group trees is sometimes unused; 
     it points to offset zero in the heap, which is always the 
     empty string and compares as "less-than" any valid object name.

    <p>And key[<i>N</i>] for chunk trees is sometimes unused; 
     it contains a chunk offset which compares as "greater-than" 
     any other chunk offset and has a chunk byte size of zero 
     to indicate that it is not actually allocated.

    <h4><a name="V2Btrees">
        Disk Format: Level 1A2 - Version 2 B-Trees</a></h4>

    <p>Version 2 B-trees are "traditional" B-trees, with one major difference.
        Instead of just using a simple pointer (or address in the file) to a
        child of an internal node, the pointer to the child node contains two
        additional pieces of information: the number of records in the child
        node itself, and the total number of records in the child node and
        all its descendents.  Storing this additional information allows fast
        array-like indexing to locate the n'th record in the B-tree.

    <P>The entry into a version 2 B-tree is a header which contains global
        information about the structure of the B-tree.  The <EM>root node
        address</EM>
        field in the header points to the B-tree root node, which is either an
        internal or leaf node, depending on the value in the header's
        <em>depth</em> field.  An internal node consists of records plus
        pointers to further leaf or internal nodes in the tree.  A leaf node
        consists of solely of records.  The format of the records depends on
        the B-tree type (stored in the header).

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree Header
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
        <tr>
          <td>Version</td>
          <td>Type</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>
	<tr>
          <td colspan=4>Node Size</td>
	</tr>
	<tr>
          <td colspan=2>Record Size</td>
          <td colspan=2>Depth</td>
	</tr>
	<tr>
          <td>Split Percent</td>
          <td>Merge Percent</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>
	<tr>
          <td colspan=4><br>Root Node Address<sup>O</sup><br><br></td>
	</tr>
	<tr>
          <td colspan=2>Number of Records in Root Node</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>
	<tr>
          <td colspan=4><br>Total Number of Records in B-tree<sup>L</sup><br><br></td>
	</tr>
	<tr>
          <td colspan=4>Checksum</td>
	</tr>
      </table>
    </div>


    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>BTHD</code>" is 
            used to indicate the header of a version 2 B-link tree node.
            </P>
          </td>
        </tr>

        <tr>
          <td>Version</td>
          <td>
            <P>The version number for this B-tree header.  This document
                describes version 0.
            </P>
          </td>
        </tr>

        <tr>
          <td>Type</td>
          <td>
            <P>This field indicates the type of B-tree:
            <table class=list>
                <tr>
                <th width="30%">Value</th>
                <th align=left>Description</th>
                </tr>
                <tr>
                    <td align=center>0</td>
                    <td>A "testing" B-tree, this value should <em>not</em> be
                        used for storing records in actual HDF5 files.
                    </td>
                </tr>
                <tr>
                    <td align=center>1</td>
                    <td>This B-tree is used for indexing indirectly accessed,
                        non-filtered 'huge' fractal heap objects.
                    </td>
                </tr>
                <tr>
                    <td align=center>2</td>
                    <td>This B-tree is used for indexing indirectly accessed,
                        filtered 'huge' fractal heap objects.
                    </td>
                </tr>
                <tr>
                    <td align=center>3</td>
                    <td>This B-tree is used for indexing directly accessed,
                        non-filtered 'huge' fractal heap objects.
                    </td>
                </tr>
                <tr>
                    <td align=center>4</td>
                    <td>This B-tree is used for indexing directly accessed,
                        filtered 'huge' fractal heap objects.
                    </td>
                </tr>
                <tr>
                    <td align=center>5</td>
                    <td>This B-tree is used for indexing the 'name' field for
                        links in indexed groups.
                    </td>
                </tr>
                <tr>
                    <td align=center>6</td>
                    <td>This B-tree is used for indexing the 'creation order'
                        field for links in indexed groups.
                    </td>
                </tr>
                <tr>
                    <td align=center>7</td>
                    <td>This B-tree is used for indexing shared object header
                        messages.
                    </td>
                </tr>
                <tr>
                    <td align=center>8</td>
                    <td>This B-tree is used for indexing the 'name' field for
                        indexed attributes.
                    </td>
                </tr>
                <tr>
                    <td align=center>9</td>
                    <td>This B-tree is used for indexing the 'creation order'
                        field for indexed attributes.
                    </td>
                </tr>
            </table>
            <P>The format of records for each type is described below.
          </td>
        </tr>

        <tr valign=top>
          <td>Node Size</td>
          <td>
            <P>This is the size in bytes of all B-tree nodes.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Record Size</td>
          <td>
            <P>This field is the size in bytes of the B-tree record.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Depth</td>
          <td>
            <P>This is the depth of the B-tree.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Split Percent</td>
          <td>
            <P>The percent full that a node needs to increase above before it
                is split.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Merge Percent</td>
          <td>
            <P>The percent full that a node needs to be decrease below before it
                is split.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Root Node Address</td>
          <td>
            <P>This is the address of the root B-tree node.  A B-tree with
                no records will have the <A href="#UndefinedAddress">undefined
                address</A> in this field.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Number of Records in Root Node</td>
          <td>
            <P>This is the number of records in the root node.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Total Number of Records in B-tree</td>
          <td>
            <P>This is the total number of records in the entire B-tree.
            </P>
          </td>
        </tr>

        <tr valign=top>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for the B-tree header.
            </P>
          </td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree Internal Node
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
        <tr>
          <td>Version</td>
          <td>Type</td>
          <td colspan=2>Records 0, 1, 2...N-1 <em>(variable size)</em></td>
	</tr>
	<tr>
	  <td colspan=4><br>Child Node Pointer 0<sup>O</sup><br><br></td>
	</tr>
	  <td colspan=4><br>Number of Records N<sub>0</sub> for Child Node 0 <em>(variable size)</em></td>
	</tr>
	<tr>
	  <td colspan=4><br>Total Number of Records for Child Node 0 <em>(optional, variable size)</em></td>
	</tr>
	<tr>
	  <td colspan=4><br>Child Node Pointer 1<sup>O</sup><br><br></td>
	</tr>
	  <td colspan=4><br>Number of Records N<sub>1</sub> for Child Node 1 <em>(variable size)</em></td>
	</tr>
	<tr>
	  <td colspan=4><br>Total Number of Records for Child Node 1 <em>(optional, variable size)</em></td>
	</tr>
	<tr>
	  <td colspan=4>...</td>
	</tr>
	<tr>
	  <td colspan=4><br>Child Node Pointer N<sup>O</sup><br><br></td>
	</tr>
	</tr>
	  <td colspan=4><br>Number of Records N<sub>n</sub>for Child Node N <em>(variable size)</em></td>
	</tr>
	<tr>
	  <td colspan=4><br>Total Number of Records for Child Node N <em>(optional, variable size)</em></td>
	</tr>
	<tr>
	  <td colspan=4>Checksum</td>
      </table>
    </div>


    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>BTIN</code>" is 
            used to indicate the internal node of a B-link tree.
            </P>
          </td>
        </tr>

        <tr>
          <td>Version</td>
          <td>
            <P>The version number for this B-tree internal node.  
	       This document describes version 0.
           </P>
          </td>
        </tr>

        <tr>
          <td>Type</td>
          <td>
            <P>This field is the type of the B-tree node.  It should always
                be the same as the B-tree type in the header.
            </P>
          </td>
        </tr>

        <tr>
          <td>Records</td>
          <td>
            <P>The size of this field is determined by the number of records
	       for this node and the record size (from the header).  The format
               of records depends on the type of B-tree.
            </P>
          </td>
        </tr>

        <tr>
          <td>Child Node Pointer</td>
          <td>
            <P>This field is the address of the child node pointed to by the
                internal node.
            </P>
          </td>
        </tr>

        <tr>
          <td>Number of Records in Child Node</td>
          <td>
            <P>This is the number of records in the child node pointed to by
                the corresponding <em>Node Pointer</em>.
            </P>
            <P>The number of bytes used to store this field is determined by
                the maximum possible number of records able to be stored in the
                child node.
            </P>
            <P>
                The maximum number of records in a child node is computed in
                the following way: Subtract the fixed size overhead for the
                child node
                (e.g. it's signature, version, checksum, etc. and <em>one</em>
                pointer triplet of information for the child node (because there
                is one more pointer triplet than records in each internal node))
                from the size of nodes for the B-tree and dividing that result
                by the size of a record plus the pointer triplet of information
                stored to reach each child node from this node.
            </P>
            <P>
                Note that leaf nodes don't encode any
                child pointer triplets, so the maximum number of records in a
                leaf node is just the node size minus the leaf node overhead,
                divided by the record size.
            </P>
            <P>
                Also note that the first level of internal nodes above the
                leaf nodes don't encode the <em>Total Number of Records in Child
                Node</em> value in the child pointer triplets (since it is the
                same as the <em>Number of Records in Child Node</em>), so the
                maximum number of records in these nodes is computed with the
                equation above, but using (<em>Child Pointer</em>, <em>Number of
                Records in Child Node</em>) pairs instead of triplets.
            </P>
            <P>
                The number of
                bytes used to encode this field is the least number of bytes
                required to encode the maximum number of records in a child
                node value for the child nodes below this level
                in the B-tree.
            </P>
            <P>
                For example, if the maximum number of child records is 123, one
                byte will be used to encode these values in this node, if the
                maximum number of child records is
                20000, two bytes will be used to encode these values in this
                node, etc.  The maximum number of bytes used to encode these
                values is 8 (i.e. an unsigned 64-bit integer).
            </P>
          </td>
        </tr>

        <tr>
          <td>Total Number of Records in Child Node</td>
          <td>
            <P>This is the total number of records for the node pointed to by
               the corresponding <em>Node Pointer</em> and all its children.
               This field exists only in nodes whose depth in the B-tree node
               is greater than 1 (i.e. the "twig" internal nodes, just above
               leaf nodes, don't store this field in their child node pointers).
            </P>
            <P>The number of bytes used to store this field is determined by
                the maximum possible number of records able to be stored in the
                child node and its descendents.
            </P>
            <P>
                The maximum possible number of records able to be stored in a
                child node and its descendents is computed iteratively, in the
                following way:  The maximum number of records in a leaf node
                is computed, then that value is used to compute the maximum
                possible number of records in the first level of internal nodes
                above the leaf nodes.  Multiplying these two values together
                determines the maximum possible number of records in child node
                pointers for the level of nodes two levels above leaf nodes.
                This process is continued up to any level in the B-tree.
            </P>
            <P>
                The number of bytes used to encode this value is computed in
                the same way as for the <em>Number of Records in Child Node</em>
                field.
            </P>
          </td>
        </tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for this node.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree Leaf Node
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
        <tr>
          <td>Version</td>
          <td>Type</td>
          <td colspan=2>Record 0, 1, 2...N-1 <em>(variable size)</em></td>
	</tr>
	<tr>
	  <td colspan=4>Checksum</td>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>BTLF</code>" is 
                used to indicate the leaf node of a version 2 B-link tree.
            </P>
          </td>
        </tr>

        <tr>
          <td>Version</td>
          <td>
            <P>The version number for this B-tree leaf node.  
	       This document describes version 0.
            </P>
          </td>
        </tr>

        <tr>
          <td>Type</td>
          <td>
            <P>This field is the type of the B-tree node.  It should always
                be the same as the B-tree type in the header.
            </P>
          </td>

        <tr>
          <td>Records</td>
          <td>
            <P>The size of this field is determined by the number of records
	       for this node and the record size (from the header).  The format
               of records depends on the type of B-tree.
            </P>
          </td>
        </tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for this node.
            </P>
          </td>
        </tr>

      </table>
    </div>


    <p>The record layout for each stored (i.e. non-testing) B-tree type is as
        follows:

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 1 Record Layout - Indirectly Accessed, Non-Filtered,
            'Huge' Fractal Heap Objects
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td colspan=4><br>Huge Object Address<sup>O</sup><br><br></td>
        </tr>
        <tr>
          <td colspan=4><br>Huge Object Length<sup>L</sup><br><br></td>
        </tr>
        <tr>
          <td colspan=4><br>Huge Object ID<sup>L</sup><br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Huge Object Address</td>
          <td>
            <P>The address of the huge object in the file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Huge Object Length</td>
          <td>
            <P>The length of the huge object in the file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Huge Object ID</td>
          <td>
            <P>The heap ID for the huge object.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 2 Record Layout - Indirectly Accessed, Filtered,
            'Huge' Fractal Heap Objects
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td colspan=4><br>Filtered Huge Object Address<sup>O</sup><br><br></td>
        </tr>
        <tr>
          <td colspan=4><br>Filtered Huge Object Length<sup>L</sup><br><br></td>
        </tr>
        <tr>
          <td colspan=4>Filter Mask</td>
        </tr>
        <tr>
          <td colspan=4><br>Filtered Huge Object Memory Size<sup>L</sup><br><br></td>
        </tr>
        <tr>
          <td colspan=4><br>Huge Object ID<sup>L</sup><br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Filtered Huge Object Address</td>
          <td>
            <P>The address of the filtered huge object in the file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Filtered Huge Object Length</td>
          <td>
            <P>The length of the filtered huge object in the file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Filter Mask</td>
          <td>
            <P>A 32-bit bitfield indicating which filters have been skipped for
                this chunk.  Each filter has an index number in the pipeline
                (starting at 0, with the first filter to apply) and if that
                filter is skipped, the bit corresponding to its index is set.
            </P>
          </td>
        </tr>

        <tr>
          <td>Filtered Huge Object Memory Size</td>
          <td>
            <P>The size of the de-filtered huge object in memory.
            </P>
          </td>
        </tr>

        <tr>
          <td>Huge Object ID</td>
          <td>
            <P>The heap ID for the huge object.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 3 Record Layout - Directly Accessed, Non-Filtered,
            'Huge' Fractal Heap Objects
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td colspan=4><br>Huge Object Address<sup>O</sup><br><br></td>
        </tr>
        <tr>
          <td colspan=4><br>Huge Object Length<sup>L</sup><br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Huge Object Address</td>
          <td>
            <P>The address of the huge object in the file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Huge Object Length</td>
          <td>
            <P>The length of the huge object in the file.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 4 Record Layout - Directly Accessed, Filtered,
            'Huge' Fractal Heap Objects
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td colspan=4><br>Filtered Huge Object Address<sup>O</sup><br><br></td>
        </tr>
        <tr>
          <td colspan=4><br>Filtered Huge Object Length<sup>L</sup><br><br></td>
        </tr>
        <tr>
          <td colspan=4>Filter Mask</td>
        </tr>
        <tr>
          <td colspan=4><br>Filtered Huge Object Memory Size<sup>L</sup><br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Filtered Huge Object Address</td>
          <td>
            <P>The address of the filtered huge object in the file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Filtered Huge Object Length</td>
          <td>
            <P>The length of the filtered huge object in the file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Filter Mask</td>
          <td>
            <P>A 32-bit bitfield indicating which filters have been skipped for
                this chunk.  Each filter has an index number in the pipeline
                (starting at 0, with the first filter to apply) and if that
                filter is skipped, the bit corresponding to its index is set.
            </P>
          </td>
        </tr>

        <tr>
          <td>Filtered Huge Object Memory Size</td>
          <td>
            <P>The size of the de-filtered huge object in memory.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 5 Record Layout - Link Name for Indexed Group
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td colspan=4>Hash of Name</td>
        </tr>
        <tr>
          <td colspan=4>ID <em>(bytes 1-4)</em></td>
	</tr>

        <tr>
          <td colspan=3>ID <em>(bytes 5-7)</em></td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Hash</td>
          <td>
            <P>This field is hash value of the name for the link.  The hash 
                value is the Jenkins' lookup3 checksum algorithm applied to
                the link's name.
            </P>
          </td>
        </tr>

        <tr>
          <td>ID</td>
          <td>
            <P>This is a 7-byte sequence of bytes and is the heap ID for the
                link record in the group's fractal heap.
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 6 Record Layout - Creation Order for Indexed Group
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td colspan=4><br>Creation Order <em>(8 bytes)</em><br><br></td>
        </tr>
        <tr>
          <td colspan=4>ID <em>(bytes 1-4)</em></td>
	</tr>
	<tr>
          <td colspan=3>ID <em>(bytes 5-7)</em></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Creation Order</td>
          <td>
            <P>This field is the creation order value for the link.
            </P>
          </td>
        </tr>

        <tr>
          <td>ID</td>
          <td>
            <P>This is a 7-byte sequence of bytes and is the heap ID for the
                link record in the group's fractal heap.
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 7 Record Layout - Shared Object Header Messages (Sub-Type 0 - Message in Heap)
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td colspan>Message Location</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>
        <tr>
          <td colspan=4>Hash</td>
	</tr>
	<tr>
          <td colspan=4>Reference Count</td>
	</tr>
	<tr>
          <td colspan=4><br>Heap ID <em>(8 bytes)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Message Location</td>
          <td>
            <P>This field Indicates the location where the message is stored:
            <table class=list>
                <tr>
                <th width="30%">Value</th>
                <th align=left>Description</th>
                </tr>
                <tr>
                    <td align=center>0</td>
                    <td>Shared message is stored in shared message index heap.
                    </td>
                </tr>
                <tr>
                    <td align=center>1</td>
                    <td>Shared message is stored in object header.
                    </td>
                </tr>
            </table>
          </td>
        </tr>

        <tr>
          <td>Hash</td>
          <td>
            <P>This field is hash value of the shared message.  The hash 
                value is the Jenkins' lookup3 checksum algorithm applied to
                the shared message.
          </td>
        </tr>

        <tr>
          <td>Reference Count</td>
          <td>
            <P>The number of objects which reference this message.
          </td>
        </tr>

        <tr>
          <td>Heap ID</td>
          <td>
            <P>This is a 8-byte sequence of bytes and is the heap ID for the
                shared message in the shared message index's fractal heap.
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 7 Record Layout - Shared Object Header Messages (Sub-Type 1 - Message in Object Header)
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td colspan>Message Location</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>
        <tr>
          <td colspan=4>Hash</td>
	</tr>
	<tr>
          <td>Reserved (zero)</td>
          <td>Message Type</td>
          <td colspan=2>Object Header Index</td>
	</tr>
	<tr>
          <td colspan=4><br>Object Header Address<sup>O</sup><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Message Location</td>
          <td>
            <P>This field Indicates the location where the message is stored:
            <table class=list>
                <tr>
                <th width="30%">Value</th>
                <th align=left>Description</th>
                </tr>
                <tr>
                    <td align=center>0</td>
                    <td>Shared message is stored in shared message index heap.
                    </td>
                </tr>
                <tr>
                    <td align=center>1</td>
                    <td>Shared message is stored in object header.
                    </td>
                </tr>
            </table>
          </td>
        </tr>

        <tr>
          <td>Hash</td>
          <td>
            <P>This field is hash value of the shared message.  The hash 
                value is the Jenkins' lookup3 checksum algorithm applied to
                the shared message.
          </td>
        </tr>

        <tr>
          <td>Message Type</td>
          <td>
            <P>The object header message type of the shared message.
          </td>
        </tr>

        <tr>
          <td>Object Header Index</td>
          <td>
            <P>This field indicates that the shared message is the n'th message
                of its type in the specified object header.
          </td>
        </tr>

        <tr>
          <td>Object Header Address</td>
          <td>
            <P>The address of the object header containing the shared message.
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 8 Record Layout - Attribute Name for Indexed Attributes
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

	<tr>
          <td colspan=4><br>Heap ID <em>(8 bytes)</em><br><br></td>
	</tr>
        <tr>
          <td colspan>Message Flags</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>
        <tr>
          <td colspan=4>Creation Order</td>
        </tr>
        <tr>
          <td colspan=4>Hash of Name</td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Heap ID</td>
          <td>
            <P>This is a 8-byte sequence of bytes and is the heap ID for the
                attribute in the object's attribute fractal heap.
          </td>
        </tr>

        <tr>
          <td>Message Flags</td>
          <td>
            <P>The object header message flags for the attribute message.
          </td>
        </tr>

        <tr>
          <td>Creation Order</td>
          <td>
            <P>This field is the creation order value for the attribute.
            </P>
          </td>
        </tr>

        <tr>
          <td>Hash</td>
          <td>
            <P>This field is hash value of the name for the attribute.  The hash 
                value is the Jenkins' lookup3 checksum algorithm applied to
                the attribute's name.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Version 2 B-tree, Type 9 Record Layout- Creation Order for Indexed Attributes
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

	<tr>
          <td colspan=4><br>Heap ID <em>(8 bytes)</em><br><br></td>
	</tr>
        <tr>
          <td colspan>Message Flags</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>
        <tr>
          <td colspan=4>Creation Order</td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Heap ID</td>
          <td>
            <P>This is a 8-byte sequence of bytes and is the heap ID for the
                attribute in the object's attribute fractal heap.
          </td>
        </tr>

        <tr>
          <td>Message Flags</td>
          <td>
            <P>The object header message flags for the attribute message.
          </td>
        </tr>

        <tr>
          <td>Creation Order</td>
          <td>
            <P>This field is the creation order value for the attribute.
            </P>
          </td>
        </tr>

      </table>
    </div>


    <h3><a name="SymbolTable">Disk Format: Level 1B - Group Symbol Table Nodes</a></h3>

    <p>A group is an object internal to the file that allows
      arbitrary nesting of objects within the file (including other groups).
      A group maps a set of link names in the group to a set of relative
      file addresses of objects in the file.  Certain metadata for an object to
      which the group points can be cached in the group's symbol table entry in
      addition to being in the object's header.

    <p>An HDF5 object name space can be stored hierarchically by
      partitioning the name into components and storing each
      component as a link in a group.  The link for a
      non-ultimate component points to the group containing
      the next component.  The link for the last
      component points to the object being named.

    <p>One implementation of a  group is a collection of symbol table nodes
      indexed by a B-link tree.  Each symbol table node contains entries
      for one or more links.  If an attempt is made to add a link to an already
      full symbol table node containing 2<em>K</em> entries, then the node is
      split and one node contains <em>K</em> symbols and the other contains
      <em>K</em>+1 symbols.

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Symbol Table Node (A Leaf of a B-link tree)
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>

        <tr>
          <td colspan=4>Signature</td>

        <tr>
          <td>Version Number</td>
          <td>Reserved (zero)</td>
          <td colspan=2>Number of Symbols</td>

        <tr>
          <td colspan=4><br><br>Group Entries<br><br><br></td>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>SNOD</code>" is 
            used to indicate the
            beginning of a symbol table node.  This gives file
            consistency checking utilities a better chance of
            reconstructing a damaged file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Version Number</td>
          <td>
            <P>The version number for the symbol table node.  This
            document describes version 1. (There is no version '0'
            of the symbol table node)
            </P>
          </td>
        </tr>

        <tr>
          <td>Number of Entries</td>
          <td>
            <P>Although all symbol table nodes have the same length,
            most contain fewer than the maximum possible number of
            link entries.  This field indicates how many entries
            contain valid data.  The valid entries are packed at the
            beginning of the symbol table node while the remaining
            entries contain undefined values.
            </P>
          </td>
        </tr>

        <tr>
          <td>Symbol Table Entries</td>
          <td>
            <P>Each link has an entry in the symbol table node.
            The format of the entry is described below.
            There are 2<EM>K</EM> entries in each group node, where
            <EM>K</EM> is the "Group Leaf Node K" value from the
            <A href="#Superblock">superblock</A>.
            </P>
          </td>
        </tr>
      </table>
    </div>

    <h3><a name="SymbolTableEntry">
        Disk Format: Level 1C - Symbol Table Entry </a></h3>

    <p>Each symbol table entry in a symbol table node is designed 
      to allow for very fast browsing of stored objects.
      Toward that design goal, the symbol table entries
      include space for caching certain constant metadata from the
      object header.

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Symbol Table Entry
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4><br>Link Name Offset<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Object Header Address<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4>Cache Type</td>
        </tr>

        <tr>
          <td colspan=4>Reserved (zero)</td>
        </tr>

        <tr>
          <td colspan=4><br><br>Scratch-pad Space (16 bytes)<br><br><br></td>
        </tr>
      </table>

      <table class=note>
        <tr><td>
            (Items marked with an 'O' the above table are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Link Name Offset</td>
          <td>
            <P>This is the byte offset into the group's local
            heap for the name of the link. The name is null
            terminated.
            </P>
          </td>
        </tr>

        <tr>
          <td>Object Header Address</td>
          <td>
            <P>Every object has an object header which serves as a
            permanent location for the object's metadata.  In addition
            to appearing in the object header, some of the object's metadata
            can be cached in the scratch-pad space.
            </P>
          </td>
        </tr>

        <tr>
          <td>Cache Type</td>
          <td>
            <P>The cache type is determined from the object header.
            It also determines the format for the scratch-pad space:
            <br>
            <table class=list>
                <tr align=left>
                    <th>Type:</th>
                    <th>Description:</th>
                </tr>
                <tr>
                    <td width="10%" align=center>0</td>
                    <td>No data is cached by the group entry.  This
                        is guaranteed to be the case when an object header
                        has a link count greater than one.
                    </td>
                </tr>
                <tr>
                    <td align=center>1</td>
                    <td>Group object header metadata is cached in the
                        scratch-pad space.  This implies that the symbol table
                        entry refers to another group.
                    </td>
                </tr>
                <tr>
                    <td align=center>2</td>
                    <td>The entry is a symbolic link.  The first four bytes
                        of the scratch-pad space are the offset into the local
                        heap for the link value.  The object header address
                        will be undefined.
                    </td>
                </tr>
            </table>
            </P>
          </td>
        </tr>

        <tr>
          <td>Reserved</td>
          <td>
            <P>These four bytes are present so that the scratch-pad
            space is aligned on an eight-byte boundary.  They are
            always set to zero.
            </P>
          </td>
        </tr>

        <tr>
          <td>Scratch-pad Space</td>
          <td>
            <P>This space is used for different purposes, depending
            on the value of the Cache Type field. Any metadata
            about an object represented in the scratch-pad
            space is duplicated in the object header for that
            object.
            </P>
            <P>
            Furthermore, no data is cached in the group
            entry scratch-pad space if the object header for
            the object has a link count greater than one.
            </P>
          </td>
        </tr>
      </table>
    </div>

    <h4>Format of the Scratch-pad Space</h4>

    <p>The symbol table entry scratch-pad space is formatted
      according to the value in the Cache Type field.  

    <p>If the Cache Type field contains the value zero 
      <code>(0)</code> then no information is
      stored in the scratch-pad space.

    <p>If the Cache Type field contains the value one
      <code>(1)</code>, then the scratch-pad space
      contains cached metadata for another object header 
      in the following format:

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Object Header Scratch-pad Format
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>

        <tr>
          <td colspan=4><br>Address of B-tree<sup>O</sup><br><br></td>

        <tr>
          <td colspan=4><br>Address of Name Heap<sup>O</sup><br><br></td>
      </table>

      <table class=note>
        <tr><td>
            (Items marked with an 'O' the above table are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Address of B-tree</td>
          <td>
            <P>This is the file address for the root of the 
                group's B-tree.
            </P>
          </td>
        </tr>

        <tr>
          <td>Address of Name Heap</td>
          <td>
            <P>This is the file address for the group's local
                heap, in which are stored the group's symbol names.
            </P>
          </td>
        </tr>
      </table>
    </div>


    <P>If the Cache Type field contains the value two
      <code>(2)</code>, then the scratch-pad space
      contains cached metadata for a symbolic link
      in the following format:

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Symbolic Link Scratch-pad Format
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Offset to Link Value</td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Offset to Link Value</td>
          <td>
            <P>The value of a symbolic link (that is, the name of the
                thing to which it points) is stored in the local heap.
                This field is the 4-byte offset into the local heap for
                the start of the link value, which is null terminated.
            </P>
          </td>
        </tr>
      </table>
    </div>

    <h3><a name="LocalHeap">Disk Format: Level 1D - Local Heaps</a></h3>

    <P>A local heap is a collection of small pieces of data that are particular
      to a single object in the HDF5 file.  Objects can be
      inserted and removed from the heap at any time.  
      The address of a heap does not change once the heap is created. 
      For example, a group stores addresses of objects in symbol table nodes
      with the names of links stored in the group's local heap.
    </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Local Heap
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Signature</td>
        </tr>

        <tr>
          <td>Version</td>
          <td colspan=3>Reserved (zero)</td>
        </tr>
          
        <tr>
          <td colspan=4><br>Data Segment Size<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Offset to Head of Free-list<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Address of Data Segment<sup>O</sup><br><br></td>
        </tr>
      </table>

      <table class=note>
        <tr><td>
            (Items marked with an 'L' the above table are
            <br>
            of the size specified in "Size of Lengths.")
        </td></tr>
        <tr><td>
            (Items marked with an 'O' the above table are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
      </table>
    </div>

    <p>
    <center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>HEAP</code>"
            is used to indicate the
            beginning of a heap.  This gives file consistency
            checking utilities a better chance of reconstructing a
            damaged file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Version</td>
          <td>
            <P>Each local heap has its own version number so that new
            heaps can be added to old files.  This document
            describes version zero (0) of the local heap.
            </P>
          </td>
        </tr>

        <tr>
          <td>Data Segment Size</td>
          <td>
            <P>The total amount of disk memory allocated for the heap
            data.  This may be larger than the amount of space
            required by the objects stored in the heap.  The extra
            unused space in the heap holds a linked list of free blocks.
            </P>
          </td>
        </tr>

        <tr>
          <td>Offset to Head of Free-list</td>
          <td>
            <P>This is the offset within the heap data segment of the
            first free block (or the
            <A href="#UndefinedAddress">undefined address</A> if there is no
            free block).  The free block contains "Size of Lengths" bytes that
            are the offset of the next free block (or the
            value '1' if this is the
            last free block) followed by "Size of Lengths" bytes that store
            the size of this free block.  The size of the free block includes
            the space used to store the offset of the next free block and
            the size of the current block, making the minimum size of a free 
            block 2 * "Size of Lengths".
            </P>
          </td>
        </tr>

        <tr>
          <td>Address of Data Segment</td>
          <td>
            <P>The data segment originally starts immediately after
            the heap header, but if the data segment must grow as a
            result of adding more objects, then the data segment may
            be relocated, in its entirety, to another part of the 
            file.
            </P>
          </td>
        </tr>
      </table>
    </center>

    <p>Objects within a local heap should be aligned on an 8-byte boundary.

    <h3><a name="GlobalHeap">Disk Format: Level 1E - Global Heap</a></h3>

    <P>Each HDF5 file has a global heap which stores various types of
      information which is typically shared between datasets.  The
      global heap was designed to satisfy these goals:

    <ol type="A">
      <li>Repeated access to a heap object must be efficient without
        resulting in repeated file I/O requests. Since global heap
        objects will typically be shared among several datasets, it is
        probable that the object will be accessed repeatedly.
      <li>Collections of related global heap objects should result in
        fewer and larger I/O requests.  For instance, a dataset of
        object references will have a global heap object for each
        reference.  Reading the entire set of object references
        should result in a few large I/O requests instead of one small
        I/O request for each reference.
      <li>It should be possible to remove objects from the global heap
        and the resulting file hole should be eligible to be reclaimed
        for other uses.
    </ol>
    </P>

    <P>The implementation of the heap makes use of the memory management
        already available at the file level and combines that with a new
        object called a <em>collection</em> to achieve goal B. The global heap
        is the set of all collections.  Each global heap object belongs to
        exactly one collection and each collection contains one or more global
        heap objects. For the purposes of disk I/O and caching, a collection is
        treated as an atomic object, addressing goal A.
    </P>

    <P>When a global heap object is deleted from a collection (which occurs
        when its reference count falls to zero), objects located after the
        deleted object in the collection are packed down toward the beginning
        of the collection and the collection's global heap object 0 is created
        (if possible) or its size is increased to account for the recently
        freed space.  There are no gaps between objects in each collection,
        with the possible exception of the final space in the collection, if
        it is not large enough to hold the header for the collection's global
        heap object 0.  These features address goal C.
    </P>

    <P>The HDF5 library creates global heap collections as needed, so there may
        be multiple collections throughout the file. The set of all of them is
        abstractly called the "global heap", although they don't actually link
        to each other, and there is no global place in the file where you can
        discover all of the collections.  The collections are found simply by
        finding a reference to one through another object in the file.  For
        example, data of variable-length datatype elements is stored in the
        global heap and is accessed via a global heap ID.  The format for
        global heap IDs is described at the end of this section.
    </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          A Global Heap Collection
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td colspan=3>Reserved (zero)</td>
        </tr>

        <tr>
          <td colspan=4><br>Collection Size<sup>L</sup><br><br></td>
        </tr>
          
        <tr>
          <td colspan=4><br>Global Heap Object 1<br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Global Heap Object 2<br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>...<br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Global Heap Object <em>N</em><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Global Heap Object 0 (free space)<br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>GCOL</code>"
            is used to indicate the
            beginning of a collection.  This gives file consistency
            checking utilities a better chance of reconstructing a
            damaged file.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td>
            <P>Each collection has its own version number so that new
            collections can be added to old files.  This document
            describes version one (1) of the collections (there is no
            version zero (0)).
            </P>
          </td>
        </tr>

        <tr>
          <td>Collection Size</td>
          <td>
            <P>This is the size in bytes of the entire collection
            including this field.  The default (and minimum)
            collection size is 4096 bytes which is a typical file
            system block size.  This allows for 127 16-byte heap
            objects plus their overhead (the collection header of 16 bytes
            and the 16 bytes of information about each heap object).
            </P>
          </td>
        </tr>

        <tr>
          <td>Global Heap Object 1 through <em>N</em></td> 
          <td>
            <P>The objects are stored in any order with no 
            intervening unused space.
            </P>
          </td>
        </tr>

        <tr>
          <td>Global Heap Object 0</td>
          <td>
            <P>Global Heap Object 0 (zero), when present, represents the free
            space in the collection.  Free space always appears at the end of
            the collection.  If the free space is too small to store the header
            for Object 0 (described below) then the header is implied and the
            collection contains no free space.
            </P>
          </td>
      </table>
    </div>
    
    <br>
    <div align=center>
      <table class=format>
        <caption>
          Global Heap Object
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>
          
        <tr>
          <td colspan=2>Heap Object Index</td>
          <td colspan=2>Reference Count</td>
        </tr>

        <tr>
          <td colspan=4>Reserved (zero)</td>
        </tr>

        <tr>
          <td colspan=4><br>Object Size<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Object Data<br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Heap Object Index</td>
          <td>
            <P>Each object has a unique identification number within a
            collection.  The identification numbers are chosen so that
            new objects have the smallest value possible with the
            exception that the identifier <code>0</code> always refers to the
            object which represents all free space within the
            collection.
            </P>
          </td>
        </tr>

        <tr>
          <td>Reference Count</td>
          <td>
            <P>All heap objects have a reference count field.  An
            object which is referenced from some other part of the
            file will have a positive reference count. The reference
            count for Object 0 is always zero.
            </P>
          </td>
        </tr>

        <tr>
          <td>Reserved</td>
          <td>
            <P>Zero padding to align next field on an 8-byte boundary.
            </P>
          </td>
        </tr>

        <tr>
          <td>Object Size</td>
          <td>
            <P>This is the size of the object data stored for the object.
            The actual storage space allocated for the object data is rounded
            up to a multiple of eight.
            </P>
          </td>
        </tr>

        <tr>
          <td>Object Data</td>
          <td>
            <P>The object data is treated as a one-dimensional array
            of bytes to be interpreted by the caller.
            </P>
          </td>
        </tr>
      </table>
      <table class=note>
        <tr><td>
            (Items marked with an 'O' in the above tables are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
        <tr><td>
            (Items marked with an 'L' the above tables are
            <br>
            of the size specified in "Size of Lengths.")
        </td></tr>
      </table>
    </div>

    <P>
    The format for the ID used to locate an object in the global heap is
    described here:

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Global Heap ID
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4><br>Collection Address<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4>Object Index</td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Collection Address</td>
          <td>
            <P>This field is the address of the global heap collection
		where the data object is stored.
            </P>
          </td>
        </tr>

        <tr>
          <td>ID</td>
          <td>
            <P>This field is the index of the data object within the
		global heap collection.
            </P>
          </td>
        </tr>

      </table>
    </div>
    </P>

    <br>
    <h3><a name="FractalHeap">Disk Format: Level 1F - Fractal Heap</a></h3>

        <P>
	    Each fractal heap consists of a header and zero or more direct and
            indirect blocks (described below).  The header contains general
            information as well as
            initialization parameters for the doubling table.  The <em>Root
            Block Address</em> in the header points to the first direct or
            indirect block in the heap.
        </P>

        <P>
            Fractal heaps are based on a data structure called a <em>doubling
            table</em>.  A doubling table provides a mechanism for quickly
            extending an array-like data structure that minimizes the number of
            empty blocks in the heap, while retaining very fast lookup of any
            element within the array.  More information on fractal heaps and
            doubling tables can be found in the RFC 
            &ldquo;<a href="Supplements/FractalHeap/PrivateHeap.pdf">Private
            Heaps in HDF5</a>.&rdquo;
        </P>

        <P>
            The fractal heap implements the doubling table structure with
            indirect and direct blocks.
            Indirect blocks in the heap do not actually contain data for
            objects in the heap, their "size" is abstract - 
            they represent the indexing structure for locating the
            direct blocks in the doubling table.
            Direct blocks
            contain the actual data for objects stored in the heap.
        </P>

        <P>
            All indirect blocks have a constant number of block entries in each
            row, called the <em>width</em> of the doubling table (stored in
            the heap header).

            The number
            of rows for each indirect block in the heap is determined by the
            size of the block that the indirect block represents in the
            doubling table (calculation of this is shown below) and is
            constant, except for the "root"
            indirect block, which expands and shrinks its number of rows as
            needed. 
        </P>

        <P>
            Blocks in the first <em>two</em> rows of an indirect block are
            <em>Starting Block Size</em> number of bytes in size, and the blocks
            in each subsequent row are twice the size of the blocks in the
            previous row (i.e. blocks in the third row are twice the
            <em>Starting Block Size</em>, blocks in the fourth row are
            four times the <em>Starting Block Size</em>, etc.).  Entries for
            blocks up to the <em>Maximum Direct Block Size</em> point to direct
            blocks and entries for blocks greater than that size point to
            further indirect blocks (which have their own entries for direct
            and indirect blocks, etc).
        </P>

        <P>
            The number of rows of blocks, <em>nrows</em>, in an
            indirect block of size <em>iblock_size</em> is given by the
            following expression:
            <br> <br>
                <em>nrows</em> = (log<sub>2</sub>(<em>iblock_size</em>) -
                    log<sub>2</sub>(<em>&lt;Starting Block Size&gt;</em> *
                        <em>&lt;Width&gt;</em>)) + 1
        </P>

        <P>
            The maximum number of rows of direct blocks, <em>max_dblock_rows</em>,
            in any indirect block of a fractal heap is given by the
            following expression:
            <br> <br>
                <em>max_dblock_rows</em> =
                    (log<sub>2</sub>(<em>&lt;Max. Direct Block Size&gt;</em>) -
                    log<sub>2</sub>(<em>&lt;Starting Block Size&gt;</em>)) + 2
        </P>

        <P>
            Using the computed values for <em>nrows</em> and
            <em>max_dblock_rows</em>, along with the <em>Width</em> of the
            doubling table, the number of direct and indirect block entries
            (<em>K</em> and <em>N</em> in the indirect block description, below)
            in an indirect block can be computed:
            <br> <br>
                <em>K</em> = MIN(<em>nrows</em>, <em>max_dblock_rows</em>) *
                    <em>Width</em>

            <br> <br>
            If <em>nrows</em> is less than or equal to <em>max_dblock_rows</em>,
            <em>N</em> is 0.  Otherwise, <em>N</em> is simply computed:
            <br> <br>
                <em>N</em> = <em>K</em> - (<em>max_dblock_rows</em> *
                    <em>Width</em>)
        </P>

        <P>
            The size indirect blocks on disk is determined by the number
            of rows in the indirect block (computed above).  The size of direct
            blocks on disk is exactly the size of the block in the doubling
            table.
        </P>


    <br>
    <div align=center>
      <table class=format>
        <caption>
          Fractal Heap Header
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=2>Heap ID Length</td>
          <td colspan=2>I/O Filters' Encoded Length</td>
        </tr>
          
        <tr>
          <td>Flags</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4>Maximum Size of Managed Objects</td>
        </tr>

        <tr>
          <td colspan=4><br>Next Huge Object ID<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>v2 B-tree Address of Huge Objects<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Amount of Free Space in Managed Blocks<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Address of Managed Block Free Space Manager<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Amount of Managed Space in Heap<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Amount of Allocated Managed Space in Heap<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Offset of Direct Block Allocation Iterator in Managed Space<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Number of Managed Objects in Heap<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Size of Huge Objects in Heap<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Number of Huge Objects in Heap<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Size of Tiny Objects in Heap<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Number of Tiny Objects in Heap<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=2>Table Width</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Starting Block Size<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Maximum Direct Block Size<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=2>Maximum Heap Size</td>
          <td colspan=2>Starting # of Rows in Root Indirect Block</td>
        </tr>

        <tr>
          <td colspan=4><br>Address of Root Block<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=2>Current # of Rows in Root Indirect Block</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Size of Filtered Root Direct Block <em>(optional)</em><sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4>I/O Filter Mask<em> (optional)</em></td>
        </tr>

        <tr>
          <td colspan=4>I/O Filter Information<em> (optional, variable size)</em></td>
        </tr>

        <tr>
          <td colspan=4>Checksum</td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="40%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>FRHP</code>"
            is used to indicate the
            beginning of a fractal heap header.  This gives file consistency
            checking utilities a better chance of reconstructing a
            damaged file.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td>
            <P>This document describes version 0.</P>
          </td>
        </tr>

	<tr>
          <td>Heap ID Length</td>
          <td>
            <P>This is the length in bytes of heap object IDs for this heap.</P>
          </td>
        </tr>

        <tr>
          <td>I/O Filters' Encoded Length</td>
	  <td>
	    <P>This is the size in bytes of the encoded <em>I/O Filter Information</em>.
            </P>
	  </td>
        </tr>

	<tr>
	  <td>Flags</td>
	  <td>
            <P>This field is the heap status flag and is a bit-field indicating additional 
	    information about the fractal heap.
            <table class=list>
                <tr>
                  <th width="30%">Bit(s)</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>If set, the ID value to use for huge object has wrapped
                    around.  If the value for the <em>Next Huge Object ID</em>
                    has wrapped around, each new huge object inserted into the
                    heap will require a search for an ID value.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>1</code></td>
                  <td>If set, the direct blocks in the heap are checksummed.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>2-7</code></td>
                  <td>Reserved</td>
                </tr>
	    </table>
            </P>
          </td>
	</tr>
	
	<tr>
	  <td>Maximum Size of Managed Objects</td>
	  <td>
	    <P>This is the maximum size of managed objects allowed in the heap.
                Objects greater than this this are 'huge' objects and will be
                stored in the file directly, rather than in a direct block for
                the heap.
	    </P>
	  </td>
	</tr>

	<tr>
	  <td>Next Huge Object ID</td>
	  <td>
	    <P>This is the next ID value to use for a huge object in the heap.
	    </P>
	  </td>

	<tr>
	  <td>v2 B-tree Address of Huge Objects</td>
	  <td>
	    <P>This is the address of the <a href="#V2Btrees">v2 B-tree</a>
                used to track huge objects in the heap.  The type of records
                stored in the <em>v2 B-tree</em> will
                be determined by whether the address & length of a huge object
                can fit into a heap ID (if yes, it's a "directly" accessed
                huge object) and whether there is a filter used on objects
                in the heap.
	    </P>
	  </td>
	</tr>

	<tr>
	  <td>Amount of Free Space in Managed Blocks</td>
	  <td>
	    <P>This is the total amount of free space in managed direct blocks
                (in bytes).
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Address of Managed Block Free Space Manager</td>
	  <td>
	    <P>This is the address of the
                <em><a href="#FreeSpaceManager">Free-Space Manager</a></em> for
                managed blocks.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Amount of Managed Space in Heap</td>
	  <td>
	    <P>This is the total amount of managed space in the heap (in bytes),
                essentially the upper bound of the heap's linear address space.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Amount of Allocated Managed Space in Heap</td>
	  <td>
	    <P>This is the total amount of managed space (in bytes) actually
                allocated in
                the heap.  This can be less than the <em>Amount of Managed Space
                in Heap</em> field, if some direct blocks in the heap's linear
                address space are not allocated.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Offset of Direct Block Allocation Iterator in Managed Space</td>
	  <td>
	    <P>This is the linear heap offset where the next direct
                block should be allocated at (in bytes).  This may be less than
                the <em>Amount of Managed Space in Heap</em> value because the
                heap's address space is increased by a "row" of direct blocks
                at a time, rather than by single direct block increments.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Number of Managed Objects in Heap</td>
	  <td>
	    <P>This is the number of managed objects in the heap.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Size of Huge Objects in Heap</td>
	  <td>
	    <P>This is the total size of huge objects in the heap (in bytes).
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Number of Huge Objects in Heap</td>
	  <td>
	    <P>This is the number of huge objects in the heap.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Size of Tiny Objects in Heap</td>
	  <td>
	    <P>This is the total size of tiny objects that are packed in heap
                IDs (in bytes).
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Number of Tiny Objects in Heap</td>
	  <td>
	    <P>This is the number of tiny objects that are packed in heap IDs.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Table Width</td>
	  <td>
	    <P>This is the number of columns in the doubling table for managed
                blocks.  This value must be a power of two.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Starting Block Size</td>
	  <td>
	    <P>This is the starting block size to use in the doubling table for 
	       managed blocks (in bytes).  This value must be a power of two.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Maximum Direct Block Size</td>
	  <td>
	    <P>This is the maximum size allowed for a managed direct block.
                Objects inserted into the heap that are larger than this value
                (less the # of bytes of direct block prefix/suffix)
                are stored as 'huge' objects.  This value must be a power of
                two.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Maximum Heap Size</td>
	  <td>
	    <P>This is the maximum size of the heap's linear address space for
                managed objects (in bytes).  The value stored is the log2 of
                the actual value, that is: the # of bits of the address space.
                'Huge' and 'tiny' objects aren't counted in this value, since
                they don't store objects in the linear address space of the
                heap.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Starting # of Rows in Root Indirect Block</td>
	  <td>
	    <P>This is the starting number of rows for the root indirect block.
	       A value of 0 indicates that the root indirect block will have
               the maximum number of rows needed to address the heap's <em>Maximum
               Heap Size</em>.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Address of Root Block</td>
	  <td>
	    <P>This is the address of the root block for the heap.  It can
                be the <A href="#UndefinedAddress">undefined address</A> if
                there is no data in the heap.  It either points to a direct
                block (if the <em>Current # of Rows in the Root Indirect Block</em>
                value is 0), or an indirect block.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Current # of Rows in Root Indirect Block</td>
	  <td>
	    <P>This is the current number of rows in the root indirect block.
		A value of 0 indicates that <em>Address of Root Block</em>
                points to direct block instead of indirect block.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>Size of Filtered Root Direct Block</td>
	  <td>
	    <P>This is the size of the root direct block, if filters are
                applied to heap objects (in bytes).  This field is only
                stored in the header if the <em>I/O Filters' Encoded Length</em>
                is greater than 0.
	    </P>
	  </td>
	</tr>

	<tr>
          <td>I/O Filter Mask</td>
	  <td>
	    <P>This is the filter mask for the root direct block, if filters
                are applied to heap objects.  This mask has the same format as
                that used for the filter mask in chunked raw data records in a
                <a href="#V1Btrees">v1 B-tree</a>.
                This field is only
                stored in the header if the <em>I/O Filters' Encoded Length</em>
                is greater than 0.
	    </P>
	  </td>
	</tr>

        <tr>
          <td>I/O Filter Information</td>
	  <td>
	    <P>This is the I/O filter information encoding direct blocks and
                huge objects, if filters are applied to heap objects.  This
                field is encoded as a <a href="#FilterMessage">Filter Pipeline</a>
                message.
                The size of this field is determined by <em>I/O Filters'
                Encoded Length</em>.
	    </P>
	  </td>
        </tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for the header.</P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Fractal Heap Direct Block
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Heap Header Address<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4>Block Offset <em>(variable size)</em></td>
        </tr>
          
        <tr>
          <td colspan=4><br>Object Data <em>(variable size)</em><br><br></td>
        </tr>
          
        <tr>
          <td colspan=4>Checksum <em>(optional)</em></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>FHDB</code>"
            is used to indicate the
            beginning of a fractal heap direct block.  This gives file consistency
            checking utilities a better chance of reconstructing a
            damaged file.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td>
            <P>This document describes version 0.</P>
          </td>
        </tr>

        <tr>
          <td>Heap Header Address</td>
          <td>
            <P>This is the address for the fractal heap header that this
                block belongs to.  This field is principally used for file
                integrity checking.
            </P>
          </td>
        </tr>

        <tr>
          <td>Block Offset</td>
          <td>
            <P>This is the offset of the block within the fractal heap's
                address space (in bytes).  The number of bytes used to encode
                this field is the <em>Maximum Heap Size</em> (in the heap's
                header) divided by 8 and rounded up to the next highest integer,
                for values that aren't a multiple of 8.  This value is
                principally used for file integrity checking.
            </P>
          </td>
        </tr>

        <tr>
          <td>Object Data</td>
          <td>
            <P>This section of the direct block stores the actual data for
                objects in the heap.  The size of this section is determined by
                the direct block's size, minus the size of the other fields
                stored in the direct block (the <em>Signature</em>,
                <em>Version</em>, etc., including the <em>Checksum</em> if it's
                present).
            </P>
          </td>
        </tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for the direct block.</P>
	    <P>This field is only present if bit 1 of <em>Flags</em> in the
                heap's header is set.</P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Fractal Heap Indirect Block
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Heap Header Address<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4>Block Offset <em>(variable size)</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Child Direct Block #0 Address<sup>O</sup><br><br></td>
	</tr>
	<tr>
	  <td colspan=4><br>Size of Filtered Direct Block #0 <em>(optional)</em> <sup>L</sup><br><br></td>
	</tr>
	<tr>
	  <td colspan=4>Filter Mask for Direct Block #0 <em>(optional)</em></td>
	</tr>

        <tr>
          <td colspan=4><br>Child Direct Block #1 Address<sup>O</sup><br><br></td>
	</tr>
	<tr>
	  <td colspan=4><br>Size of Filtered Direct Block #1 <em>(optional)</em><sup>L</sup><br><br></td>
	</tr>
	<tr>
	  <td colspan=4>Filter Mask for Direct Block #1 <em>(optional)</em></td>
	</tr>

	<tr>
	    <td colspan=4>...</td>
	</tr>

        <tr>
          <td colspan=4><br>Child Direct Block #K-1 Address<sup>O</sup><br><br></td>
	</tr>
	<tr>
	  <td colspan=4><br>Size of Filtered Direct Block #K-1 <em>(optional)</em><sup>L</sup><br><br></td>
	</tr>
	<tr>
	  <td colspan=4>Filter Mask for Direct Block #K-1 <em>(optional)</em></td>
	</tr>

        <tr>
          <td colspan=4><br>Child Indirect Block #0 Address<sup>O</sup><br><br></td>
	</tr>

        <tr>
          <td colspan=4><br>Child Indirect Block #1 Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4>...</td>
	</tr>

        <tr>
          <td colspan=4><br>Child Indirect Block #N-1 Address<sup>O</sup><br><br></td>
	</tr>
          
        <tr>
	  <td colspan=4>Checksum</td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>FHIB</code>" is used to
                indicate the beginning of a fractal heap indirect block.  This
                gives file consistency checking utilities a better chance of
                reconstructing a damaged file.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td>
            <P>This document describes version 0.</P>
          </td>
        </tr>

        <tr>
          <td>Heap Header Address</td>
          <td>
            <P>This is the address for the fractal heap header that this
                block belongs to.  This field is principally used for file
                integrity checking.
            </P>
          </td>
        </tr>

        <tr>
          <td>Block Offset</td>
          <td>
            <P>This is the offset of the block within the fractal heap's
                address space (in bytes).  The number of bytes used to encode
                this field is the <em>Maximum Heap Size</em> (in the heap's
                header) divided by 8 and rounded up to the next highest integer,
                for values that aren't a multiple of 8.  This value is
                principally used for file integrity checking.
            </P>
          </td>
        </tr>

        <tr>
          <td>Child Direct Block #K Address</td>
          <td>
	    <P>This field is the address of the child direct block.
                The size of the [uncompressed] direct block can be computed by
                its offset in the heap's linear address space.  
            </P>
          </td>
        </tr>

	<tr>
	  <td>Size of Filtered Direct Block #K</td>
	  <td>
	    <P>This is the size of the child direct block after passing through
                the I/O filters defined for this heap (in bytes).  If no I/O
                filters are present for this heap, this field is not present.
            </P>
	  </td>
	</tr>
	<tr>
	  <td>Filter Mask for Direct Block #K</td>
	  <td>
            <P>This is the I/O filter mask for the filtered direct block.
                This mask has the same format as that used for the filter mask
                in chunked raw data records in a <a href="#V1Btrees">v1 B-tree</a>.
                If no I/O filters are present for this heap, this field is not
                present.
            </P>
	  </td>
        </tr>

        <tr>
          <td>Child Indirect Block #N Address</td>
          <td>
	    <P>This field is the address of the child indirect block.
                The size of the indirect block can be computed by
                its offset in the heap's linear address space.  
            </P>
          </td>
        </tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for the indirect block.</P>
          </td>
        </tr>

      </table>

      <table class=note>
        <tr><td>
            (Items marked with an 'O' in the above tables are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
        <tr><td>
            (Items marked with an 'L' the above table are
            <br>
            of the size specified in "Size of Lengths.")
        </td></tr>
      </table>
    </div>

    <br>
    <P>An object in the fractal heap is identified by means of a fractal heap ID, 
	which encodes information to locate the object in the heap.
	Currently, the fractal heap stores an object in one of three ways,
        depending on the object's size:

    <br>
    <div align=center>
      <table class=list>
        <tr>
          <th width="15%">Type</th>
          <th align=left>Description</th>
        </tr>

        <tr>
          <td align=center>Tiny</td>
          <td>
            <P>When an object is small enough to be encoded in the heap ID, the
            object's data is embedded in the fractal heap ID itself.  There are
            2 sub-types for this type of object: normal and extended.  The
            sub-type for tiny heap IDs depends on whether the heap ID is large
            enough to store objects greater than 16 bytes or not.  If the
            heap ID length is 18 bytes or smaller, the "normal" tiny heap ID
            form is used.  If the heap ID length is greater than 18 bytes in
            length, the "extented" form is used.  See format description below
            for both sub-types.
            </P>
          </td>
        </tr>

        <tr>
          <td align=center>Huge</td>
          <td>
            <P>When the size of an object is larger than <em>Maximum Size of
            Managed Objects</em> in the <em>Fractal Heap Header</em>, the
            object's data is stored on its own in the file and the object
            is tracked/indexed via a version 2 B-tree.  All huge objects
            for a particular fractal heap use the same v2 B-tree.  All huge
            objects for a particular fractal heap use the same format for
            their huge object IDs.
            </P>

            <P>Depending on whether the IDs for a heap are large enough to hold
            the object's retrieval information and whether I/O pipeline filters
            are applied to the heap's objects, 4 sub-types are derived for
            for huge object IDs for this heap:

            <div align=center>
              <table class=list>
                <tr>
                  <th width="35%">Sub-type</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=left>Directly accessed, non-filtered</td>
                  <td>
                    <P>The object's address and length are embedded in the
                    fractal heap ID itself and the object is directly accessed
                    from them.  This allows the object to be accessed without
                    resorting to the B-tree.
                    </P>
                  </td>
                </tr>

                <tr>
                  <td align=left>Directly accessed, filtered</td>
                  <td>
                    <P>The filtered object's address, length, filter mask and
                    de-filtered size are embedded in the fractal heap ID itself
                    and the object is accessed directly with them.  This allows
                    the object to be accessed without resorting to the B-tree.
                    </P>
                  </td>
                </tr>

                <tr>
                  <td align=left>Indirectly accessed, non-filtered</td>
                  <td>
                    <P>The object is located by using a B-tree key embedded in
                    the fractal heap ID to retrieve the address and length from
                    the version 2 B-tree for huge objects.  Then, the address
                    and length are used to access the object.
                    </P>
                  </td>
                </tr>

                <tr>
                  <td align=left>Indirectly accessed, filtered</td>
                  <td>
                    <P>The object is located by using a B-tree key embedded in
                    the fractal heap ID to retrieve the filtered object's
                    address, length, filter mask and de-filtered size from the
                    version 2 B-ree for huge objects.  Then, this information
                    is used to access the object.
                    </P>
                  </td>
                </tr>
              </table>
            </div>

          </td>
        </tr>

        <tr>
          <td align=center>Managed</td>
          <td>
            <P>When the size of an object does not meet the above two
            conditions, the object is stored and managed via the direct and
            indirect blocks based on the doubling table.
            </P>
          </td>
        </tr>
      </table>
    </div>
    </P>

    <P>The specific format for each type of heap ID is described below:
    </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>Fractal Heap ID for Tiny Objects (sub-type 1 - "Normal")
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td>Version, Type & Length</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Data <em>(variable size)</em></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Version, Type & Length</td>
          <td>
            <P>This is a bit field with the following definition:
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>6-7</code></td>
                  <td>The current version of ID format.  This document
                    describes version 0.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>4-5</code></td>
                  <td>The ID type.  Tiny objects have a value of <code>2</code>.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>0-3</code></td>
                  <td>The length of the tiny object.  The value stored is one
                    less than the actual length (since zero-length objects are
                    not allowed to be stored in the heap).
		    For example, an object of actual length 1 has an encoded
		    length of 0, an object of actual length 2 has an encoded
		    length of 1, etc.
                  </td>
                </tr>
	    </table>
            </P>
          </td>
        </tr>

        <tr>
          <td>Data</td>
          <td>
            <P>This is the data for the object.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>Fractal Heap ID for Tiny Objects (sub-type 2 - "Extended")
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td>Version, Type & Length</td>
          <td>Extended Length</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4>Data <em>(variable size)</em></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Version, Type & Length</td>
          <td>
            <P>This is a bit field with the following definition:
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>6-7</code></td>
                  <td>The current version of ID format.  This document
                    describes version 0.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>4-5</code></td>
                  <td>The ID type.  Tiny objects have a value of <code>2</code>.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>0-3</code></td>
                  <td>These 4 bits, together with the next byte, form an
                    unsigned 12-bit integer for holding the length of the
                    object.  These 4-bits are bits 8-11 of the 12-bit integer.
                    See description for the <em>Extended Length</em> field below.
                  </td>
                </tr>
	    </table>
            </P>
          </td>
        </tr>

        <tr>
          <td>Extended Length</td>
          <td>
            <P>This byte, together with the 4 bits in the previous byte, 
		forms an unsigned 12-bit integer for holding the length of the
                tiny object.  These 8 bits are bits 0-7 of the 12-bit integer
                formed.  The value stored is one less than the actual length
                (since zero-length objects are not allowed to be stored in the
                heap).  For example, an object of actual length 1 has an encoded
		length of 0, an object of actual length 2 has an encoded
		length of 1, etc.
            </P>
          </td>
        </tr>

        <tr>
          <td>Data</td>
          <td>
            <P>This is the data for the object.
            </P>
          </td>
        </tr>

      </table>
    </div>


    <br>
    <div align=center>
      <table class=format>
        <caption>Fractal Heap ID for Huge Objects (sub-type 1 & 2): indirectly accessed, non-filtered/filtered
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td>Version & Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>v2 B-tree Key<sup>L</sup><em> (variable size)</em><br><br></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
        <tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Version & Type</td>
          <td>
            <P>This is a bit field with the following definition:
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>6-7</code></td>
                  <td>The current version of ID format.  This document
                    describes version 0.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>4-5</code></td>
                  <td>The ID type.  Huge objects have a value of <code>1</code>.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>0-3</code></td>
                  <td>Reserved.
                  </td>
                </tr>
            </table>
            </P>
          </td>
        </tr>

        <tr>
            <td>v2 B-tree Key</td>
            <td><P>This field is the B-tree key for retrieving the information
            from the version 2 B-tree for huge objects needed to access the
            object.  See the description of <a href="#V2Btrees">v2 B-tree</a>
            records sub-type 1 & 2 for a description of the fields.  New key
            values are derived from <em>Next Huge Object ID</em> in the
            <em>Fractal Heap Header</em>.</P>
            </td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>Fractal Heap ID for Huge Objects (sub-type 3): directly accessed, non-filtered
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td>Version & Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Address <sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Length <sup>L</sup><br><br></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
        <tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Version & Type</td>
          <td>
            <P>This is a bit field with the following definition:
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>6-7</code></td>
                  <td>The current version of ID format.  This document
                    describes version 0.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>4-5</code></td>
                  <td>The ID type.  Huge objects have a value of <code>1</code>.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>0-3</code></td>
                  <td>Reserved.
                  </td>
                </tr>
            </table>
            </P>
          </td>
        </tr>

        <tr>
            <td>Address</td>
            <td><P>This field is the address of the object in the file.</P>
            </td>
        </tr>

        <tr>
            <td>Length</td>
            <td><P>This field is the length of the object in the file.</P>
            </td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>Fractal Heap ID for Huge Objects (sub-type 4): directly accessed, filtered
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td>Version & Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Address <sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Length <sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4>Filter Mask</td>
        </tr>

        <tr>
          <td colspan=4><br>De-filtered Size <sup>L</sup><br><br></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
        <tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Version & Type</td>
          <td>
            <P>This is a bit field with the following definition:
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>6-7</code></td>
                  <td>The current version of ID format.  This document
                    describes version 0.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>4-5</code></td>
                  <td>The ID type.  Huge objects have a value of <code>1</code>.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>0-3</code></td>
                  <td>Reserved.
                  </td>
                </tr>
            </table>
            </P>
          </td>
        </tr>

        <tr>
            <td>Address</td>
            <td><P>This field is the address of the filtered object in the file.</P>
        </td>
	</tr>

	<tr>
	    <td>Length</td>
	    <td><P>This field is the length of the filtered object in the file.</P>
	    </td>
	</tr>

	<tr>
	    <td>Filter Mask</td>
	    <td><P>This field is the I/O pipeline filter mask for the 
		filtered object in the file.</P>
	    </td>
	</tr>

	<tr>
	    <td>Filtered Size</td>
	    <td><P>This field is the size of the de-flitered object in the file.</P>
	    </td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>Fractal Heap ID for Managed Objects
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
	</tr>

        <tr>
          <td>Version & Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>
        <tr>
          <td colspan=4>Offset <em>(variable size)</em></td>
        </tr>

        <tr>
          <td colspan=4>Length <em>(variable size)</em></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Version & Type</td>
          <td>
            <P>This is a bit field with the following definition:
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>6-7</code></td>
                  <td>The current version of ID format.  This document
                    describes version 0.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>4-5</code></td>
                  <td>The ID type.  Managed objects have a value of <code>0</code>.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>0-3</code></td>
                  <td>Reserved.
                  </td>
                </tr>
	    </table>
            </P>
          </td>
        </tr>

        <tr>
          <td>Offset</td>
          <td>
            <P>This field is the offset of the object in the heap.  This field's
                size is the minimum number of bytes necessary to encode the
                <em>Maximum Heap Size</em> value (from the <em>Fractal Heap
                Header</em>).  For example, if the value of the <em>Maximum
                Heap Size</em> is less than 256 bytes, this field is 1 byte
                in length, a <em>Maximum Heap Size</em> of of 256-65535 bytes
                uses a 2 byte length, etc.
            </P>
          </td>
        </tr>

        <tr>
          <td>Length</td>
          <td>
            <P>This field is the length of the object in the heap.
		It is determined by taking the minimum value of 
		<em>Maximum Direct Block Size</em>
		and <em>Maximum Size of Managed Objects</em> in the
		<em>Fractal Heap Header</em>.  Again, the minimum number of
                bytes needed to encode that value is used for the size of
                this field.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <h3><a name="FreeSpaceManager">Disk Format: Level 1G - Free-Space Manager</a></h3>

        <P>
            Free-space managers are used to describe space within a heap or
            the entire HDF5 file that is not currently used for that heap or
            file.
        </P>

        <P>
            The superblock contains a pointer to a free-space managers for the
            whole file, but that pointer is currently required to be the
            <A href="#UndefinedAddress">undefined address</A>.
        </P>


        <P>
            The <em>free-space manager header</em> contains metadata information
            about the space being tracked, along with the address of the list
            of <em>free space sections</em> which actually describes the free
            space.

	    The header records information about free-space sections being
            tracked, creation parameters for handling free-space sections of a
            client, and section information used to locate the collection of
            free-space sections.
        </P>
            
        <P>
            The <em>free-space section list</em> stores a collection of
            free-space sections that is specific to each <em>client</em> of the
            free-space manager.

            For example, the fractal heap is a client of the free space manager
            and uses it to track unused space within the heap.  There are 4
            types of section records for the fractal heap, each of which has
            its own format, listed below.
        </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Free-Space Manager Header
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td>Client ID</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Total Space Tracked<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Total Number of Sections<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Number of Serialized Sections<sup>L</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Number of Un-Serialized Sections<sup>L</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=2>Number of Section Classes</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

        <tr>
          <td colspan=2>Shrink Percent</td>
          <td colspan=2>Expand Percent</td>
        </tr>

        <tr>
          <td colspan=2>Size of Address Space</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

        <tr>
          <td colspan=4><br>Maximum Section Size <sup>L</sup><br><br></td>
	</tr>

        <tr>
          <td colspan=4><br>Address of Serialized Section List<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Size of Serialized Section List Used<sup>L</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Allocated Size of Serialized Section List<sup>L</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Checksum</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="35%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>FSHD</code>" is used to
                indicate the beginning of the Free-Space Manager Header.
                This gives file consistency checking utilities a better chance of
                reconstructing a damaged file.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td>
            <P>This is the version number for the Free-Space Manager Header
		and this document describes version 0.</P>
          </td>
        </tr>

        <tr>
          <td>Client ID</td>
          <td>
            <P>This is the client ID for identifying the user of this
                free-space manager:

                <table class=list>
                    <tr>
                      <th width="30%">ID</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>Fractal heap
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>1+</code></td>
                      <td>Reserved.
                      </td>
                    </tr>
                </table>
            </P>
          </td>
        </tr>

	<tr>
          <td>Total Space Tracked</td>
	  <td>
            <P>This is the total amount of free space being tracked, in bytes.
            </P>
          </td>
	</tr>

        <tr>
          <td>Total Number of Sections</td>
          <td>
	    <P>This is the total number of free-space sections being tracked.
            </P>
          </td>
        </tr>

	<tr>
	  <td>Number of Serialized Sections</td>
	  <td>
	    <P>This is the number of serialized free-space sections being
                tracked.
            </P>
	  </td>
	</tr>
	<tr>
	  <td>Number of Un-Serialized Sections</td>
	  <td>
            <P>This is the number of un-serialized free-space sections being
                managed.  Un-serialized sections are created by the free-space
                client when the list of sections is read in.
            </P>
	  </td>
        </tr>

        <tr>
          <td>Number of Section Classes</td>
          <td>
	    <P>This is the number of section classes handled by this free space
                manager for the free-space client.
            </P>
          </td>
        </tr>

        <tr>
          <td>Shrink Percent</td>
          <td>
	    <P>This is the percent of current size to shrink the allocated
                serialized free-space section list.
            </P>
          </td>
        </tr>

        <tr>
          <td>Expand Percent</td>
          <td>
	    <P>This is the percent of current size to expand the allocated
                serialized free-space section list.
            </P>
          </td>
        </tr>

        <tr>
          <td>Size of Address Space</td>
          <td>
            <P>This is the size of the address space that free-space sections
                are within.  This is stored as the log<sub>2</sub> of the
                actual value (i.e. the number of bits required to store values
                within that address space).
            </P>
          </td>
        </tr>

        <tr>
          <td>Maximum Section Size</td>
          <td>
            <P>This is the maximum size of a section to be tracked.
            </P>
          </td>
        </tr>

        <tr>
          <td>Address of Serialized Section List</td>
          <td>
            <P>This is the address where the serialized free-space section
                list is stored.
            </P>
          </td>
        </tr>

        <tr>
          <td>Size of Serialized Section List Used</td>
          <td>
            <P>This is the size of the serialized free-space section
                list used (in bytes).  This value must be less than
                or equal to the <em>allocated size of serialized section
                list</em>, below.
	    </P>
          </td>
        </tr>

        <tr>
          <td>Allocated Size of Serialized Section List</td>
          <td>
            <P>This is the size of serialized free-space section list
                actually allocated (in bytes).
            </P>
          </td>
        </tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for the free-space manager header.</P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <P>
        The free-space sections being managed are stored in a <em>free-space
        section list</em>, described below.  The sections in the <em>free-space
        section list</em> are stored in the following way: a count of the
        number of
        sections describing a particular size of free space and the size of the
        free-space desribed (in bytes), followed by a list of section
        description records; then another section count & size, followed by
        the list of section descriptions for that size; etc.
    </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Free-Space Section List
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
        
        <tr>
          <td>Version</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>

        <tr>
          <td colspan=4><br>Free-Space Manager Header Address<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4>Number of Section Records in Set #0 <em>(variable size)</em></td>
        </tr>

        <tr>
          <td colspan=4>Size of Free-Space Section Described in Record Set #0 <em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #0 Section Record #0 Offset<em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=1>Record Set #0 Section Record #0 Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #0 Section Record #0 Data <em>(variable size)</td>
	</tr>

	<tr>
	  <td colspan=4>...</td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #0 Section Record #K-1 Offset<em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=1>Record Set #0 Section Record #K-1 Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #0 Section Record #K-1 Data <em>(variable size)</td>
	</tr>

        <tr>
          <td colspan=4>Number of Section Records in Set #1 <em>(variable size)</em></td>
        </tr>

        <tr>
          <td colspan=4>Size of Free-Space Section Described in Record Set #1 <em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #1 Section Record #0 Offset<em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=1>Record Set #1 Section Record #0 Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #1 Section Record #0 Data <em>(variable size)</td>
	</tr>

	<tr>
	  <td colspan=4>...</td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #1 Section Record #K-1 Offset<em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=1>Record Set #1 Section Record #K-1 Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #1 Section Record #K-1 Data <em>(variable size)</td>
	</tr>

	<tr>
	  <td colspan=4><strong>...</strong></td>
	</tr>

	<tr>
	  <td colspan=4><strong>...</strong></td>
	</tr>

        <tr>
          <td colspan=4>Number of Section Records in Set #N-1 <em>(variable size)</em></td>
        </tr>

        <tr>
          <td colspan=4>Size of Free-Space Section Described in Record Set #N-1 <em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #N-1 Section Record #0 Offset<em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=1>Record Set #N-1 Section Record #0 Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #N-1 Section Record #0 Data <em>(variable size)</td>
	</tr>

	<tr>
	  <td colspan=4>...</td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #N-1 Section Record #K-1 Offset<em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=1>Record Set #N-1 Section Record #K-1 Type</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Record Set #N-1 Section Record #K-1 Data <em>(variable size)</td>
	</tr>

	<tr>
	  <td colspan=4>Checksum</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="35%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>FSSE</code>" is used to
                indicate the beginning of the Free-Space Section Information.
                This gives file consistency checking utilities a better chance of
                reconstructing a damaged file.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Version</td>
          <td>
            <P>This is the version number for the Free-Space Section List
		and this document describes version 0.</P>
          </td>
        </tr>

        <tr>
          <td>Free-Space Manager Header Address</td>
          <td>
            <P>This is the address of the <em>Free-Space Manager Header</em>.
                This field is principally used for file
                integrity checking.
            </P>
          </td>
        </tr>

	<tr>
          <td>Number of Section Records for Set #N</td>
	  <td>
            <P>This is the number of free-space section records for set #N.
		The length of this field is the minimum number of bytes needed
                to store the <em>number of serialized sections</em> (from the
                <em>free-space manager header</em>).
            </P>

            <P>
                The number of sets of free-space section records is
                determined by the <em>size of serialized section list</em> in
                the <em>free-space manager header</em>.
            </P>
          </td>
	</tr>

        <tr>
          <td>Section Size for Record Set #N</td>
          <td>
	    <P>This is the size (in bytes) of the free-space section described
                for <em>all</em> the section records in set #N.
            </P>

            <P>
		The length of this field is the minimum number of bytes needed
                to store the <em>maximum section size</em> (from the
                <em>free-space manager header</em>).
            </P>
          </td>
        </tr>

        <tr>
          <td>Record Set #N Section #K Offset</td>
          <td>
	    <P>This is the offset (in bytes) of the free-space section within
                the client for the free-space manager.
            </P>

            <P>
		The length of this field is the minimum number of bytes needed
                to store the <em>size of address space</em> (from the
                <em>free-space manager header</em>).
            </P>
          </td>
        </tr>

        <tr>
          <td>Record Set #N Section #K Type</td>
          <td>
	    <P>This is the type of the section record, used to decode the
                <em>record set #N section #K data</em> information.
                Currently defined record types are:

                <table class=list>
                    <tr>
                      <th width="30%">Type</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>Fractal heap "single" section
                      </td>
                    </tr>

                    <tr>
                      <td align=center><code>1</code></td>
                      <td>Fractal heap "first row" section
                      </td>
                    </tr>

                    <tr>
                      <td align=center><code>2</code></td>
                      <td>Fractal heap "normal row" section
                      </td>
                    </tr>

                    <tr>
                      <td align=center><code>3</code></td>
                      <td>Fractal heap "indirect" section
                      </td>
                    </tr>

                    <tr>
                      <td align=center><code>4+</code></td>
                      <td>Reserved.
                      </td>
                    </tr>
                </table>
            </P>
          </td>
        </tr>

        <tr>
          <td>Record Set #N Section #K Data</td>
          <td>
	    <P>This is the section-type specific information for each record
                in the record set, described below.
            </P>
          </td>
        </tr>

        <tr>
          <td>Checksum</td>
          <td>
	    <P>This is the checksum for the <em>Free-Space Section List</em>.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <div>
      <table class=note>
        <tr><td>
            (Items marked with an 'O' in the above tables are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
        <tr><td>
            (Items marked with an 'L' the above tables are
            <br>
            of the size specified in "Size of Lengths.")
        </td></tr>
      </table>
    </div>

    <br>
    <P>
        The section-type specific data for each free-space section record is
        described below:
    </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Fractal Heap "Single" Section Data Record
        </caption>

        <tr>
          <td colspan=4><em>No additional record data stored</em></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Fractal Heap "First Row" Section Data Record
        </caption>

        <tr>
          <td colspan=4><em>Same format as "indirect" section data</em></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Fractal Heap "Normal Row" Section Data Record
        </caption>

        <tr>
          <td colspan=4><em>No additional record data stored</em></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Fractal Heap "Indirect" Section Data Record
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Fractal Heap Indirect Block Offset <em>(variable size)</em></td>
        </tr>

        <tr>
          <td colspan=2>Block Start Row</td>
          <td colspan=2>Block Start Column</td>
        </tr>

        <tr>
          <td colspan=2>Number of Blocks</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Fractal Heap Block Offset</td>
          <td>
            <P>The offset of the indirect block in the fractal heap's address
                space containing the empty blocks.
            </P>
            <P>
                The number of bytes used to encode this field is the minimum
                number of bytes needed to encode values for the <em>Maximum
                Heap Size</em> (in the fractal heap's header).
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Block Start Row</td>
          <td>
            <P>This is the row that the empty blocks start in.
	    </P>
          </td>
        </tr>

        <tr>
          <td>Block Start Column</td>
          <td>
            <P>This is the column that the empty blocks start in.
	    </P>
          </td>
        </tr>

        <tr>
          <td>Number of Blocks</td>
          <td>
            <P>This is the number of empty blocks covered by the section.
	    </P>
          </td>
        </tr>
      </table>
    </div>

    <br>
    <h3><a name="SOHMTable">Disk Format: Level 1H - Shared Object Header Message Table</a></h3>

        <P>
            The <em>shared object header message table</em> is used to locate
            object
            header messages that are shared between two or more object headers
            in the file.  Shared object header messages are stored and indexed
            in the file in one of two ways: indexed sequentially in a
            <em>shared header message list</em> or indexed with a v2 B-tree.
            The shared messages themselves are either stored in a fractal
            heap (when two or more objects share the message), or remain in an
            object's header (when only one object uses the message currently,
            but the message can be shared in the future).
        </P>

        <P>
            The <em>shared object header message table</em>
            contains a list of shared message index headers.  Each index header
            records information about the version of the index format, the index
            storage type, flags for the message types indexed, the number of
            messages in the index, the address where the index resides,
            and the fractal heap address if shared messages are stored there.
        </P>

        <P>
            Each index can be either a list or a v2 B-tree and may transition
            between those two forms as the number of messages in the index
            varies.  Each shared message record contains information used to
            locate the shared message from either a fractal heap or an object
            header.  The types of messages that can be shared are: <em>Dataspace,
            Datatype, Fill Value, Filter Pipeline and Attribute</em>.
        </P>

        <P>
            The <em>shared object header message table</em> is pointed to
            from a <a href="#SOHMTableMessage">shared message table</a> message
            in the superblock extension for a file.  This message stores the
            version of the table format, along with the number of index headers
            in the table.
        </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Shared Object Header Message Table
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
          
        <tr>
          <td>Version for index #0</td>
          <td>Index Type for index #0</td>
          <td colspan=2>Message Type Flags for index #0</td>
        </tr>

        <tr>
          <td colspan=4>Minimum Message Size for index #0</td>
        </tr>

        <tr>
          <td colspan=2>List Cutoff for index #0</td>
          <td colspan=2>v2 B-tree Cutoff for index #0</td>
        </tr>

        <tr>
          <td colspan=2>Number of Messages for index #0</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Index Address<sup>O</sup> for index #0<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Fractal Heap Address<sup>O</sup> for index #0<br><br></td>
	</tr>

        <tr>
          <td colspan=4>...</td>
	</tr>

        <tr>
          <td colspan=4>...</td>
	</tr>

        <tr>
          <td>Version for index #N-1</td>
          <td>Index Type for index #N-1</td>
          <td colspan=2>Message Type Flags for index #N-1</td>
        </tr>

        <tr>
          <td colspan=4>Minimum Message Size for index #N-1</td>
        </tr>

        <tr>
          <td colspan=2>List Cutoff for index #N-1</td>
          <td colspan=2>v2 B-tree Cutoff for index #N-1</td>
        </tr>

        <tr>
          <td colspan=2>Number of Messages for index #N-1</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Index Address<sup>O</sup> for index #N-1<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Fractal Heap Address<sup>O</sup> for index #N-1<br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Checksum</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="35%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>SMTB</code>" is used to
                indicate the beginning of the Shared Object Header Message table.
                This gives file consistency checking utilities a better chance of
                reconstructing a damaged file.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Version for index #N</td>
          <td>
            <P>This is the version number for the list of shared object header message
		indexes and this document describes version 0.</P>
          </td>
        </tr>

        <tr>
          <td>Index Type for index #N</td>
          <td>
            <P>The type of index can be an unsorted list or a v2 B-tree.
            </P>
          </td>
        </tr>

	<tr>
          <td>Message Type Flags for index #N</td>
	  <td>
            <P>This field indicates the type of messages tracked in the index,
                as follows:
                <table class=list>
                    <tr>
                      <th width="30%">Bits</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>If set, the index tracks <em>Dataspace Messages</em>.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>1</code></td>
                      <td>If set, the message tracks <em>Datatype Messages</em>.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2</code></td>
                      <td>If set, the message tracks <em>Fill Value Messages</em>.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>3</code></td>
                      <td>If set, the message tracks <em>Filter Pipeline Messages</em>.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>4</code></td>
                      <td>If set, the message tracksn <em>Attribute Messages</em>.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>5-15</code></td>
                      <td>Reserved (zero).
                      </td>
                    </tr>
                </table>
            </P>

            <P>
                An index can track more than one type of message, but each type
                of message can only by in one index.
            </P>
          </td>
	</tr>

        <tr>
          <td>Minimum Message Size for index #N</td>
          <td>
	    <P>This is the message size sharing threshold for the index.
		If the encoded size of the message is less than this value, the
                message is not shared.
            </P>
          </td>
        </tr>

	<tr>
	  <td>List Cutoff for index #N</td>
	  <td>
	    <P>This is is the cutoff value for the indexing of messages to
                switch from a list to a v2 B-tree.  If the number of messages
                is greater than this value, the index should be a v2 B-tree.
            </P>
	  </td>
	</tr>
	<tr>
	  <td>v2 B-tree Cutoff for index #N</td>
	  <td>
            <P>This is is the cutoff value for the indexing of messages to 
		switch from a v2 B-tree back to a list.  If the number of
                messages is less than this value, the index should be a list.
            </P>
	  </td>
        </tr>

        <tr>
          <td>Number of Messages for index #N</td>
          <td>
	    <P>The number of shared messages being tracked for the index.
            </P>
          </td>
        </tr>

        <tr>
          <td>Index Address for index #N</td>
          <td>
	    <P>This field is the address of the list or v2 B-tree where the
                index nodes reside.
            </P>
          </td>
        </tr>

        <tr>
          <td>Fractal Heap Address for index #N</td>
          <td>
	    <P>This field is the address of the fractal heap if shared messages
                are stored there.
            </P>
          </td>
        </tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for the table.</P>
          </td>
        </tr>

      </table>
    </div>

    <div>
      <table class=note>
        <tr><td>
            (Items marked with an 'O' in the above tables are
            <br>
            of the size specified in "Size of Offsets.")
        </td></tr>
      </table>
    </div>


    <br>
    <P>
        Shared messages are indexed either with a <em>shared message record
        list</em>, described below, or using a v2 B-tree (using record type 7).
        The number of records in the <em>shared message record list</em> is
        determined in the index's entry in the <em>shared object header message
        table</em>.
    </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Shared Message Record List
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td colspan=4>Signature</td>
        </tr>
          
        <tr>
          <td colspan=4>Shared Message Record #0</td>
        </tr>

        <tr>
          <td colspan=4>Shared Message Record #1</td>
        </tr>

        <tr>
          <td colspan=4>...</td>
        </tr>

        <tr>
          <td colspan=4>Shared Message Record #N-1</td>
        </tr>

        <tr>
          <td colspan=4>Checksum</td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Signature</td>
          <td>
            <P>The ASCII character string "<code>SMLI</code>" is used to
                indicate the beginning of a list of index nodes.
                This gives file consistency checking utilities a better chance of
                reconstructing a damaged file.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Shared Message Record #N</td>
          <td>
            <P>The record for locating the shared message, either in the
                fractal heap for the index, or an object header (see format for
                <em>index nodes</em> below).
	    </P>
          </td>
        </tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for the list.
            </P>
          </td>
        </tr>

      </table>
    </div>

    <br>
    <P>
        The record for each shared message in an index is stored in one of the
        following forms:
    </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Shared Message Record, for messages stored in a fractal heap
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td>Message Location</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>
          
        <tr>
          <td colspan=4>Hash Value</td>
        </tr>

        <tr>
          <td colspan=4>Reference Count</td>
        </tr>

        <tr>
          <td colspan=4><br>Fractal Heap ID<br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Message Location</td>
          <td>
            <P>This has a value of 0 indicating that the message is stored in
                the heap.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Hash Value</td>
          <td>
            <P>This is the hash value for the message.
	    </P>
          </td>
        </tr>

        <tr>
          <td>Reference Count</td>
          <td>
            <P>This is the number of times the message is used in the file.
            </P>
          </td>
        </tr>

        <tr>
          <td>Fractal Heap ID</td>
          <td>
	    <P>This is an 8-byte fractal heap ID for the message as stored in
                the fractal heap for the index.
            </P>
          </td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Shared Message Record, for messages stored in an object header
        </caption>

        <tr>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
          <th>byte</th>
        </tr>

        <tr>
          <td>Message Location</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
        </tr>
          
        <tr>
          <td colspan=4>Hash Value</td>
        </tr>

        <tr>
          <td>Reserved</td>
          <td>Message Type</td>
          <td colspan=2>Creation Index</td>
        </tr>

        <tr>
          <td colspan=4><br>Object Header Address<sup>O</sup><br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Message Location</td>
          <td>
            <P>This has a value of 1 indicating that the message is stored in
                an object header.
            </P>
          </td>
        </tr>
          
        <tr>
          <td>Hash Value</td>
          <td>
            <P>This is the hash value for the message.
	    </P>
          </td>
        </tr>

        <tr>
          <td>Message Type</td>
          <td>
            <P>This is the message type in the object header.
            </P>
          </td>
        </tr>

        <tr>
          <td>Creation Index</td>
          <td>
	    <P>This is the creation index of the message within the object
                header.
            </P>
          </td>
        </tr>

        <tr>
          <td>Object Header Address</td>
          <td>
	    <P>This is the address of the object header where the message is
                located.
            </P>
          </td>
        </tr>
      </table>
    </div>

    <BR>
    <HR>

    <h2><a name="DataObject">Disk Format: Level 2 - Data Objects </a></h2>

    <P>Data objects contain the "real" user-visible information in the file.
      These objects compose the scientific data and other information which
      are generally thought of as "data" by the end-user.  All the
      other information in the file is provided as a framework for
      storing and accessing these data objects.
    </P>

    <P>A data object is composed of header and data
      information.  The header information contains the information
      needed to interpret the data information for the object as
      well as additional "metadata" or pointers to additional
      "metadata" used to describe or annotate each object.
    </P>

    <h3><a name="ObjectHeader">
	Disk Format: Level 2A - Data Object Headers</a></h3>

    <P>The header information of an object is designed to encompass
      all the information about an object, except for the data itself.
      This information includes
      the dataspace, datatype, information about how the data
      is stored on disk (in external files, compressed, broken up in
      blocks, etc.), as well as other information used by the library
      to speed up access to the data objects or maintain a file's
      integrity.  Information stored by user applications as attributes
      is also stored in the object's header.  The header of each object is
      not necessarily located immediately prior to the object's data in the
      file and in fact may be located in any position in the file.  The order
      of the messages in an object header is not significant.
    </P>

    <P>Object headers are composed of a prefix and a set of messages.  The
      prefix contains the information needed to interpret the messages and
      a small amount of metadata about the object, and the messages contain
      the majority of the metadata about the object.
    </P>

    <h3><a name="ObjectHeaderPrefix">
	Disk Format: Level 2A1 - Data Object Header Prefix</a></h3>

    <h4><a name="V1ObjectHeaderPrefix">
	Version 1 Data Object Header Prefix</a></h4>

    <P>Header messages are aligned on 8-byte boundaries for version 1 
        object headers.
    </P>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Version 1 Object Header
	</caption>

	<tr>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Reserved (zero)</td>
	  <td colspan=2>Total Number of Header Messages</td>
	</tr>

	<tr>
	  <td colspan=4>Object Reference Count</td>
	</tr>

	<tr>
	  <td colspan=4>Object Header Size</td>
	</tr>

	<tr>
	  <td colspan=2>Header Message Type #1</td>
	  <td colspan=2>Size of Header Message Data #1</td>
	</tr>

	<tr>
	  <td>Header Message #1 Flags</td>
	  <td colspan=3>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4><br>Header Message Data #1<br><br></td>
	</tr>

	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	</tr>

	<tr>
	  <td colspan=2>Header Message Type #n</td>
	  <td colspan=2>Size of Header Message Data #n</td>
	</tr>

	<tr>
	  <td>Header Message #n Flags</td>
	  <td colspan=3>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4><br>Header Message Data #n<br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>
            <P>This value is used to determine the format of the
                information in the object header.  When the format of the
                object header is changed, the version number
                is incremented and can be used to determine how the
                information in the object header is formatted.  This
                is version one (1) (there was no version zero (0)) of the
                object header.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Total Number of Header Messages</td>
	  <td>
            <P>This value determines the total number of messages listed in
                object headers for this object.  This value includes the messages
                in continuation messages for this object.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Object Reference Count</td>
	  <td>
            <P>This value specifies the number of "hard links" to this object
                within the current file.  References to the object from external
                files, "soft links" in this file and object references in this
                file are not tracked.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Object Header Size</td>
	  <td>
            <P>This value specifies the number of bytes of header message data
                following this length field that contain object header messages
                for this object header.  This value does not include the size of
                object header continuation blocks for this object elsewhere in the
                file.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Header Message #n Type</td>
	  <td>
            <P>This value specifies the type of information included in the
                following header message data.  The message types for 
                header messages are defined in sections below.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size of Header Message #n Data</td>
	  <td>
            <P>This value specifies the number of bytes of header
                message data following the header message type and length
                information for the current message. The size includes
                padding bytes to make the message a multiple of eight
                bytes.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Header Message #n Flags</td>
	  <td>
            <P>This is a bit field with the following definition:
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>If set, the message data is constant.  This is used
                    for messages like the datatype message of a dataset.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>1</code></td>
                  <td>If set, the message is <em>shared</em> and stored 
                    in another location than the object header.  The Header
                    Message Data field contains a Shared Message
                    (described in the <A href="#ObjectHeaderMessages">Data Object Header Messages</a>
                    section below)
                    and the Size of Header Message Data field
                    contains the size of that Shared Message.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>2</code></td>
                  <td>If set, the message should not be shared.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>3</code></td>
                  <td>If set, the HDF5 decoder should fail to open this object
                    if it doesn't understand the message's type.  (Normally,
                    unknown messages should can be ignored by HDF5 decoders)
                  </td>
                </tr>
                <tr>
                  <td align=center><code>4</code></td>
                  <td>If set, the HDF5 decoder should set bit 5 of this
                    message's flags (this bit-field) if it doesn't understand
                    the message's type and the object is modified in any way.
                    (Normally, unknown messages can just be ignored by HDF5
                    decoders)
                  </td>
                </tr>
                <tr>
                  <td align=center><code>5</code></td>
                  <td>If set, this object was modified by software that didn't
                    understand this message.
                    (Normally, unknown messages should just be ignored by HDF5
                    decoders)  (Can be used to invalidate an index or a similar
                    feature)
                  </td>
                </tr>
                <tr>
                  <td align=center><code>6</code></td>
                  <td>If set, this message is shareable.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>7</code></td>
                  <td><em>Reserved</em></td>
                </tr>
	    </table>
            </P>
          </td>
        </tr>

	<tr>
	  <td>Header Message #n Data</td>
	  <td>
            <P>The format and length of this field is determined by the
                header message type and size respectively.  Some header
                message types do not require any data and this information
                can be eliminated by setting the length of the message to
                zero. The data is padded with enough zeros to make the
                size a multiple of eight.
            </P>
          </td>
	</tr>
      </table>
    </div>

    <h4><a name="V2ObjectHeaderPrefix">
	Version 2 Data Object Headers</a></h4>

    <P>Note that the "total number of messages" field has been dropped from
        the data object header prefix in this version.  The number is messages
        in the data object header is just determined by the messages encountered
        in all the object header blocks.

    <P>Note also that the fields and messages in this version of data object
        headers have <em>no</em> alignment or padding bytes inserted - they are 
        stored packed together.

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Version 2 Object Header
	</caption>

	<tr>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	</tr>

	<tr>
	  <td colspan=4>Signature</td>
	</tr>
	<tr>
	  <td>Version</td>
	  <td>Flags</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Access time <em>(optional)</em></td>
	</tr>

	<tr>
	  <td colspan=4>Modification Time <em>(optional)</em></td>
	</tr>

	<tr>
	  <td colspan=4>Change Time <em>(optional)</em></td>
	</tr>

	<tr>
	  <td colspan=4>Birth Time <em>(optional)</em></td>
	</tr>

	<tr>
	  <td colspan=2>Maximum # of compact attributes <em>(optional)</em></td>
	  <td colspan=2>Minimum # of dense attributes <em>(optional)</em></td>
	</tr>

	<tr>
	  <td>Size of Chunk #0 <em>(variable size)</em></td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td>Header Message Type #1</td>
	  <td colspan=2>Size of Header Message Data #1</td>
	  <td>Header Message #1 Flags</td>
	</tr>

	<tr>
	  <td colspan=2>Header Message #1 Creation Order <em>(optional)</em></td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Header Message Data #1<br><br></td>
	</tr>

	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	</tr>

	<tr>
	  <td>Header Message Type #n</td>
	  <td colspan=2>Size of Header Message Data #n</td>
	  <td>Header Message #n Flags</td>
	</tr>

	<tr>
	  <td colspan=2>Header Message #n Creation Order <em>(optional)</em></td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Header Message Data #n<br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Gap <em>(optional, variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=4>Checksum</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Signature</td>
	  <td>
            <P>The ASCII character string "<code>OHDR</code>"
                is used to indicate the
                beginning of an object header.  This gives file consistency
                checking utilities a better chance of reconstructing a
                damaged file.
	    </P>
	  </td>
	</tr>
	
	<tr>
	  <td>Version</td>
	  <td>
            <P>This field has a value of 2 indicating version 2 of the object header.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Flags</td>
	  <td>
            <P>This field is a bit-field indicating additional information
                about the object header.
            <table class=list>
                <tr>
                  <th width="30%">Bit(s)</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0-1</code></td>
                  <td>This two bit field determines the size of the
                    <em>Size of Chunk #0</em> field.  The values are:
                    <table class=list>
                        <tr>
                          <th width="30%">Value</th>
                          <th align=left>Description</th>
                        </tr>

                        <tr>
                          <td align=center><code>0</code></td>
                          <td>The <em>Size of Chunk #0</em> field is 1 byte.
                          </td>
                        </tr>
                        <tr>
                          <td align=center><code>1</code></td>
                          <td>The <em>Size of Chunk #0</em> field is 2 bytes.
                          </td>
                        </tr>
                        <tr>
                          <td align=center><code>2</code></td>
                          <td>The <em>Size of Chunk #0</em> field is 4 bytes.
                          </td>
                        </tr>
                        <tr>
                          <td align=center><code>3</code></td>
                          <td>The <em>Size of Chunk #0</em> field is 8 bytes.
                          </td>
                        </tr>
                    </table>
                  </td>
                </tr>
                <tr>
                  <td align=center><code>2</code></td>
                  <td>If set, attribute creation order is tracked.</td>
                </tr>
                <tr>
                  <td align=center><code>3</code></td>
                  <td>If set, attribute creation order is indexed.</td>
                </tr>
                <tr>
                  <td align=center><code>4</code></td>
                  <td>If set, non-default attribute storage phase change
                      values are stored.</td>
                </tr>
                <tr>
                  <td align=center><code>5</code></td>
                  <td>If set, access, modification, change and birth times
                      are stored.</td>
                </tr>
                <tr>
                  <td align=center><code>6-7</code></td>
                  <td>Reserved</td>
                </tr>
	    </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Access Time</td>
	  <td>
            <P>This 32-bit value represents the number of seconds after the
                UNIX epoch when the object's raw data was last accessed (i.e.
                read or written).
            </P>
	    <P>This field is present if bit 5 of <em>flags</em> is set.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Modification Time</td>
	  <td>
            <P>This 32-bit value represents the number of seconds after the
                UNIX epoch when the object's raw data was last modified (i.e.
                written).
            </P>
	    <P>This field is present if bit 5 of <em>flags</em> is set.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Change Time</td>
	  <td>
            <P>This 32-bit value represents the number of seconds after the
                UNIX epoch when the object's metadata was last changed.
            </P>
	    <P>This field is present if bit 5 of <em>flags</em> is set.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Birth Time</td>
	  <td>
            <P>This 32-bit value represents the number of seconds after the
                UNIX epoch when the object was created.
            </P>
	    <P>This field is present if bit 5 of <em>flags</em> is set.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Maximum # of compact attributes</td>
	  <td>
            <P>This is the maximum number of attributes to store in the compact
                format before switching to the indexed format.
            </P>
	    <P>This field is present if bit 4 of <em>flags</em> is set.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Minimum # of dense attributes</td>
	  <td>
            <P>This is the minimum number of attributes to store in the indexed
                format before switching to the compact format.
            </P>
	    <P>This field is present if bit 4 of <em>flags</em> is set.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size of Chunk #0</td>
	  <td>
            <P>
               This unsigned value specifies the number of bytes of header
               message data following this field that contain object header
               information.
            </P>
            <P>
               This value does not include the size of object header
               continuation blocks for this object elsewhere in the file.
            </P>
            <P>
	       The length of this field varies depending on bits 0 and 1 of
	       the <em>flags</em> field.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Header Message #n Type</td>
	  <td>
            <P>Same format as version 1 of the object header, described above.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size of Header Message #n Data</td>
	  <td>
            <P>This value specifies the number of bytes of header
                message data following the header message type and length
                information for the current message. The size of messages
                in this version does <em>not</em> include any padding bytes.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Header Message #n Flags</td>
	  <td>
            <P>Same format as version 1 of the object header, described above.
            </P>
          </td>
        </tr>

	<tr>
	  <td>Header Message #n Creation Order</td>
	  <td>
            <P>This field stores the order that a message of a given type
                was created in.
            </P>
	    <P>This field is present if bit 2 of <em>flags</em> is set.
            </P>
          </td>
        </tr>

	<tr>
	  <td>Header Message #n Data</td>
	  <td>
            <P>Same format as version 1 of the object header, described above.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Gap</td>
	  <td>
            <P>A gap in an object header chunk is inferred by the end of the
                messages for the chunk before the beginning of the chunk's
                checksum.  Gaps are always smaller than the size of an
                object header message prefix (message type + message size +
                message flags).
            </P>
            <P>Gaps are formed when a message (typically an attribute message)
                in an earlier chunk is deleted and a message from a later
                chunk that doesn't quite fit into the free space is moved
                into the earlier chunk.
            </P>
          </td>
	</tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for the object header chunk.
            </P>
          </td>
        </tr>
      </table>
    </div>

    <P>The header message types and the message data associated with
      them compose the critical "metadata" about each object.  Some
      header messages are required for each object while others are
      optional.  Some optional header messages may also be repeated
      several times in the header itself, the requirements and number
      of times allowed in the header will be noted in each header
      message description below.
    </P>

    <hr>

    <h3><a name="ObjectHeaderMessages">
	Disk Format: Level 2A2 - Data Object Header Messages</a></h3>

    <P>Data object header messages are small pieces of metadata that are
      stored in the data object header for each object in an HDF5 file.
      Data object header messages provide the metadata required to describe
      an object and its contents, as well as optional pieces of metadata
      that annotate the meaning or purpose of the object.
    </P>

    <P>Data object header messages are either stored directly in the data
      object header for the object or are shared between multiple objects
      in the file.  When a message is shared, a flag in the <em>Message Flags</em>
      indicates that the actual <em>Message Data</em>
      portion of that message is stored in another location (such as another
      data object header, or a heap in the file) and the <em>Message Data</em>
      field contains the information needed to locate the actual information
      for the message.
    </P>

    <P class=item>
    The format of shared message data is described here:
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Shared Message (Version 1)
	</caption>

	<tr>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Type</td>
	  <td colspan=2>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4><br>Address<sup>O</sup><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number is used when there are changes in the format
            of a shared object message and is described here:</P>
	    <table class=list>
	      <tr>
		<th width="30%">Version</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
		<td>Never used.</td>
	      </tr>

	      <tr>
		<td align=center><code>1</code></td>
		<td>Used by the library before version 1.6.1.
                </td>
	      </tr>
	    </table>
	</tr>

	<tr>
	  <td>Type</td>
	  <td><P>The type of shared message location:
	    <table class=list>
	      <tr>
		<th width="30%">Value</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
		<td>Message stored in another object's header (a <em>committed</em>
                    message).
                </td>
	      </tr>
	    </table>
          </td>
	</tr>

	<tr>
	  <td>Address</td>
	  <td><P>The address of the object header
            containing the message to be shared.</P>
          </td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Shared Message (Version 2)
	</caption>

	<tr>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Type</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Address<sup>O</sup><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number is used when there are changes in the format
            of a shared object message and is described here:</P>
	    <table class=list>
	      <tr>
		<th width="30%">Version</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>2</code></td>
		<td>Used by the library of version 1.6.1 and after.
                </td>
	      </tr>
	    </table>
          </td>
	</tr>

	<tr>
	  <td>Type</td>
	  <td><P>The type of shared message location:
	    <table class=list>
	      <tr>
		<th width="30%">Value</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
		<td>Message stored in another object's header (a <em>committed</em>
                    message).
                </td>
	      </tr>
	    </table>
          </td>
	</tr>

	<tr>
	  <td>Address</td>
	  <td><P>The address of the object header
            containing the message to be shared.</P></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Shared Message (Version 3)
	</caption>

	<tr>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Type</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Location <em>(variable size)</em></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number indicates changes in the format of shared 
	    object message and is described here:</P>
	    <table class=list>
	      <tr>
		<th width="30%">Version</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>3</code></td>
		<td>Used by the library of version 1.8 and after.  In this
                    version, the <em>Type</em> field can indicate that
                    the message is stored in the fractal heap.
                </td>
	      </tr>
	    </table>
	</tr>

	<tr>
	  <td>Type</td>
	  <td><P>The type of shared message location:
	    <table class=list>
	      <tr>
		<th width="30%">Value</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
		<td>Message is not shared and is not sharable..
                </td>
	      </tr>

	      <tr>
		<td align=center><code>1</code></td>
		<td>Message stored in file's <em>shared object header message</em>
                    heap (a <em>shared</em> message).
                </td>
	      </tr>

	      <tr>
		<td align=center><code>2</code></td>
		<td>Message stored in another object's header (a <em>committed</em>
                    message).
                </td>
	      </tr>

	      <tr>
		<td align=center><code>3</code></td>
		<td>Message stored is not shared, but is sharable.
                </td>
	      </tr>

	    </table>
          </td>
	</tr>

	<tr>
	  <td>Location</td>
	  <td><P>This field contains either a <em>Size of Offsets</em>-bytes
            address of the object header
            containing the message to be shared, or an 8-byte fractal heap ID
            for the message in the file's <em>shared object header message</em>
            heap.
	    </P>
          </td>
	</tr>
      </table>
    </div>
    </P>

    <P>The following is a list of currently defined header messages:
    </P>

    <hr>
    <h4><a name="NILMessage">Name: NIL</a></h4>

    <P class=item><B>Header Message Type: </B>0x0000
    </P>
    <P class=item><B>Length:</B> varies
    </P>
    <P class=item><B>Status:</B> Optional, may be repeated.
    </P>
    <P class=item><B>Purpose and Description:</B> The NIL message is used to indicate a
        message which is to be ignored when reading the header messages for a
        data object.  [Possibly one which has been deleted for some reason.]
    </P>
    <P class=item><B>Format of Data:</B> Unspecified.
    </P>

    <hr>
    <h4><a name="DataspaceMessage">Name: Dataspace</a></h4>

    <P class=item><B>Header Message Type: </B>0x0001
    </P>
    <P class=item><B>Length:</B> Varies according to the number of dimensions, 
      as described in the following table.
    </P>
    <P class=item><B>Status:</B> Required for dataset objects, may not be
        repeated.
    </P>
    <P class=item><B>Description:</B> The dataspace message describes the
        number of dimensions (i.e. "rank") and size of each dimension that the
        data object has.  This message is only used for datasets which have a
        simple, rectilinear, array-like layout; datasets requiring a more
        complex layout are not yet supported.
    </P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Dataspace Message - Version 1
	</caption>

	<tr>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Dimensionality</td>
	  <td>Flags</td>
	  <td>Reserved</td>
	</tr>

	<tr>
	  <td colspan=4>Reserved</td>
	</tr>

	<tr>
	  <td colspan=4><br>Dimension #1 Size<sup>L</sup><br><br></td>
	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	<tr>
	  <td colspan=4><br>Dimension #n Size<sup>L</sup><br><br></td>
	<tr>
	  <td colspan=4><br>Dimension #1 Maximum Size<sup>L</sup> <em>(optional)</em><br><br></td>
	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	<tr>
	  <td colspan=4><br>Dimension #n Maximum Size<sup>L</sup> <em>(optional)</em><br><br></td>
	<tr>
	  <td colspan=4><br>Permutation Index #1<sup>L</sup> <em>(optional)</em><br><br></td>
	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	<tr>
	  <td colspan=4><br>Permutation Index #n<sup>L</sup> <em>(optional)</em><br><br></td>
      </table>

      <table class=note>
        <tr><td>
            (Items marked with an 'L' the above table are
            <br>
            of the size specified in "Size of Lengths.")
        </td></tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>
            <P>This value is used to determine the format of the
	    Dataspace Message.  When the format of the
	    information in the message is changed, the version number
	    is incremented and can be used to determine how the
	    information in the object header is formatted.  This
            document describes version one (1) (there was no version
            zero (0)).
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimensionality</td>
	  <td>
            <P>This value is the number of dimensions that the data
	    object has.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Flags</td>
	  <td>
            <P>This field is used to store flags to indicate the
	    presence of parts of this message.  Bit 0 (the least
	    significant bit) is used to indicate that maximum
	    dimensions are present.  Bit 1 is used to indicate that
	    permutation indices are present.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimension #n Size</td>
	  <td>
            <P>This value is the current size of the dimension of the
	    data as stored in the file.  The first dimension stored in
	    the list of dimensions is the slowest changing dimension
	    and the last dimension stored is the fastest changing
	    dimension.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimension #n Maximum Size</td>
	  <td>
            <P>This value is the maximum size of the dimension of the
	    data as stored in the file.  This value may be the special
            "<A href="#UnlimitedDim">unlimited</A>" size which indicates
	    that the data may expand along this dimension indefinitely.
            If these values are not stored, the maximum size of each
            dimension is assumed to be the dimension's current size.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Permutation Index #n</td>
	  <td>
            <P>This value is the index permutation used to map
	    each dimension from the canonical representation to an
	    alternate axis for each dimension.  If these values are
	    not stored, the first dimension stored in the list of
	    dimensions is the slowest changing dimension and the last
	    dimension stored is the fastest changing dimension.
            </P>
          </td>
	</tr>
      </table>
    </div>

    </P>

    <P>Version 2 of the dataspace message dropped the optional permutation
        index value support, as it was never implemented in the HDF5 library:

    <div align=center>
      <table class=format>
	<caption>
	  Dataspace Message - Version 2
	</caption>

	<tr>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Dimensionality</td>
	  <td>Flags</td>
	  <td>Type</td>
	</tr>

	<tr>
	  <td colspan=4><br>Dimension #1 Size<sup>L</sup><br><br></td>
	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	<tr>
	  <td colspan=4><br>Dimension #n Size<sup>L</sup><br><br></td>
	<tr>
	  <td colspan=4><br>Dimension #1 Maximum Size<sup>L</sup> <em>(optional)</em><br><br></td>
	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	<tr>
	  <td colspan=4><br>Dimension #n Maximum Size<sup>L</sup> <em>(optional)</em><br><br></td>
      </table>

      <table class=note>
        <tr><td>
            (Items marked with an 'L' the above table are
            <br>
            of the size specified in "Size of Lengths.")
        </td></tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>
            <P>This value is used to determine the format of the
                Dataspace Message.  This field should be '2' for version 2
                format messages.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimensionality</td>
	  <td>
            <P>This value is the number of dimensions that the data object has.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Flags</td>
	  <td>
            <P>This field is used to store flags to indicate the
                presence of parts of this message.  Bit 0 (the least
                significant bit) is used to indicate that maximum
                dimensions are present.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Type</td>
	  <td>
            <P>This field indicates the type of the dataspace:
            <table class=list>
                <tr>
                  <th width="30%">Value</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>A <em>scalar</em> dataspace, i.e. a dataspace with a
                    single, dimensionless element.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>1</code></td>
                  <td>A <em>simple</em> dataspace, i.e. a dataspace with a
                    a rank > 0 and an appropriate # of dimensions.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>2</code></td>
                  <td>A <em>null</em> dataspace, i.e. a dataspace with no
                    elements.
                  </td>
                </tr>
            </table>
          </td>
	</tr>

	<tr>
	  <td>Dimension #n Size</td>
	  <td>
            <P>This value is the current size of the dimension of the
                data as stored in the file.  The first dimension stored in
                the list of dimensions is the slowest changing dimension
                and the last dimension stored is the fastest changing
                dimension.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimension #n Maximum Size</td>
	  <td>
            <P>This value is the maximum size of the dimension of the
                data as stored in the file.  This value may be the special
                "<A href="#UnlimitedDim">unlimited</A>" size which indicates
                that the data may expand along this dimension indefinitely.
                If these values are not stored, the maximum size of each
                dimension is assumed to be the dimension's current size.
            </P>
          </td>
	</tr>

      </table>
    </div>

    </P>

<!--
    <hr>
    <h4><a name="DataSpaceMessage">Name: Complex Dataspace (Fiber Bundle?)</a></h4>
    <b>Header Message Type: </b>0x0002<br>
    <b>Length:</b> varies<br>

    <b>Status:</b> One of the <em>Simple Dataspace</em> or
    <em>Complex Dataspace</em> messages is required (but not both) and may
    not be repeated.<br> <b>Purpose and Description:</b> The
    <em>Dataspace</em> message describes space that the dataset is
    mapped onto in a more comprehensive way than the <em>Simple
    Dimensionality</em> message is capable of handling.  The
    dataspace of a dataset encompasses the type of coordinate system
    used to locate the dataset's elements as well as the structure and
    regularity of the coordinate system.  The dataspace also
    describes the number of dimensions which the dataset inhabits as
    well as a possible higher dimensional space in which the dataset
    is located within.

    <br>
    <b>Format of Data:</b>

    <center>
      <table border cellpadding=4 width="80%">
	<caption align=bottom>
	  <B>HDF5 Dataspace Message Layout</B>
	</caption>

	<tr align=center>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>

	<tr align=center>
	  <td colspan=4>Mesh Type</td>
	<tr align=center>
	  <td colspan=4>Logical Dimensionality</td>
      </table>
    </center>

    <p>
    <dl>
      <dt>The elements of the dimensionality message are described below: 
      <dd>
	<dl>
	  <dt>Mesh Type: (unsigned 32-bit integer)
	  <dd>This value indicates whether the grid is
	    polar/spherical/cartesion, 
	    structured/unstructured and regular/irregular. <br>
	    The mesh type value is broken up as follows: <br>

	    <P>
	    <center>
	      <table border cellpadding=4 width="80%">
		<caption align=bottom>
		  <B>HDF5 Mesh-type Layout</B>
		</caption>

		<tr align=center>
		  <th width="25%">byte</th>
		  <th width="25%">byte</th>
		  <th width="25%">byte</th>
		  <th width="25%">byte</th>

		<tr align=center>
		  <td colspan=1>Mesh Embedding</td>
		  <td colspan=1>Coordinate System</td>
		  <td colspan=1>Structure</td>
		  <td colspan=1>Regularity</td>
	      </table>
	    </center>
	    The following are the definitions of mesh-type bytes:
	    <dl>
	      <dt>Mesh Embedding
	      <dd>This value indicates whether the dataset dataspace
		is located within 
		another dataspace or not:
		<dl> <dl>
		    <dt>&lt;STANDALONE&gt;
		    <dd>The dataset mesh is self-contained and is not
		      embedded in another mesh. 
		    <dt>&lt;EMBEDDED&gt;
		    <dd>The dataset's dataspace is located within
		      another dataspace, as 
		      described in information below.
		  </dl> </dl>
	      <dt>Coordinate System
	      <dd>This value defines the type of coordinate system
		used for the mesh: 
		<dl> <dl>
		    <dt>&lt;POLAR&gt;
		    <dd>The last two dimensions are in polar
		      coordinates, higher dimensions are 
		      cartesian.
		    <dt>&lt;SPHERICAL&gt;
		    <dd>The last three dimensions are in spherical
		      coordinates, higher dimensions 
		      are cartesian.
		    <dt>&lt;CARTESIAN&gt;
		    <dd>All dimensions are in cartesian coordinates.
		  </dl> </dl>
	      <dt>Structure
	      <dd>This value defines the locations of the grid-points
		on the axes: 
		<dl> <dl>
		    <dt>&lt;STRUCTURED&gt;
		    <dd>All grid-points are on integral, sequential
		      locations, starting from 0. 
		    <dt>&lt;UNSTRUCTURED&gt;
		    <dd>Grid-points locations in each dimension are
		      explicitly defined and 
		      may be of any numeric datatype.
		  </dl> </dl>
	      <dt>Regularity
	      <dd>This value defines the locations of the dataset
		points on the grid: 
		<dl> <dl>
		    <dt>&lt;REGULAR&gt;
		    <dd>All dataset elements are located at the
		      grid-points defined. 
		    <dt>&lt;IRREGULAR&gt;
		    <dd>Each dataset element has a particular
		      grid-location defined. 
		  </dl> </dl>
	    </dl>
	    <p>The following grid combinations are currently allowed:
	    <dl> <dl>
		<dt>&lt;POLAR-STRUCTURED-REGULAR&gt;
		<dt>&lt;SPHERICAL-STRUCTURED-REGULAR&gt;
		<dt>&lt;CARTESIAN-STRUCTURED-REGULAR&gt;
		<dt>&lt;POLAR-UNSTRUCTURED-REGULAR&gt;
		<dt>&lt;SPHERICAL-UNSTRUCTURED-REGULAR&gt;
		<dt>&lt;CARTESIAN-UNSTRUCTURED-REGULAR&gt;
		<dt>&lt;CARTESIAN-UNSTRUCTURED-IRREGULAR&gt;
	      </dl> </dl>
	    All of the above grid types can be embedded within another
	    dataspace.
	    <br> <br>
	  <dt>Logical Dimensionality: (unsigned 32-bit integer)
	  <dd>This value is the number of dimensions that the dataset occupies.

	    <P>
	    <center>
	      <table border cellpadding=4 width="80%">
		<caption align=bottom>
		  <B>HDF5 Dataspace Embedded Dimensionality Information</B>
		</caption>

		<tr align=center>
		  <th width="25%">byte</th>
		  <th width="25%">byte</th>
		  <th width="25%">byte</th>
		  <th width="25%">byte</th>

		<tr align=center>
		  <td colspan=4>Embedded Dimensionality</td>
		<tr align=center>
		  <td colspan=4>Embedded Dimension Size #1</td>
		<tr align=center>
		  <td colspan=4>.<br>.<br>.<br></td>
		<tr align=center>
		  <td colspan=4>Embedded Dimension Size #n</td>
		<tr align=center>
		  <td colspan=4>Embedded Origin Location #1</td>
		<tr align=center>
		  <td colspan=4>.<br>.<br>.<br></td>
		<tr align=center>
		  <td colspan=4>Embedded Origin Location #n</td>
	      </table>
	    </center>

	  <dt>Embedded Dimensionality: (unsigned 32-bit integer)
	  <dd>This value is the number of dimensions of the space the
	    dataset is located 
	    within.  i.e. a planar dataset located within a 3-D space,
	    or a 3-D dataset 
	    which is a subset of another 3-D space, etc.
	  <dt>Embedded Dimension Size: (unsigned 32-bit integer)
	  <dd>These values are the sizes of the dimensions of the
	    embedded dataspace 
	    that the dataset is located within.
	  <dt>Embedded Origin Location: (unsigned 32-bit integer)
	  <dd>These values comprise the location of the dataset's
	    origin within the embedded dataspace. 
	</dl>
    </dl>
    [Comment: need some way to handle different orientations of the
    dataset dataspace 
    within the embedded dataspace]<br>

    <P>
    <center>
      <table border cellpadding=4 width="80%">
	<caption align=bottom>
	  <B>HDF5 Dataspace Structured/Regular Grid Information</B>
	</caption>

	<tr align=center>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>

	<tr align=center>
	  <td colspan=4>Logical Dimension Size #1</td>
	<tr align=center>
	  <td colspan=4>Logical Dimension Maximum #1</td>
	<tr align=center>
	  <td colspan=4>.<br>.<br>.<br></td>
	<tr align=center>
	  <td colspan=4>Logical Dimension Size #n</td>
	<tr align=center>
	  <td colspan=4>Logical Dimension Maximum #n</td>
      </table>
    </center>

    <p>
    <dl>
      <dt>The elements of the dimensionality message are described below: 
      <dd>
	<dl>
	  <dt>Logical Dimension Size #n: (unsigned 32-bit integer)
	  <dd>This value is the current size of the dimension of the
	    data as stored in 
	    the file.  The first dimension stored in the list of
	    dimensions is the slowest 
	    changing dimension and the last dimension stored is the
	    fastest changing 
	    dimension. 
	  <dt>Logical Dimension Maximum #n: (unsigned 32-bit integer)
	  <dd>This value is the maximum size of the dimension of the
	    data as stored in 
	    the file.  This value may be the special value
	    &lt;UNLIMITED&gt; which 
	    indicates that the data may expand along this dimension
	    indefinitely. 
	</dl>
    </dl>
    <P>
    <center>
      <table border cellpadding=4 width="80%">
	<caption align=bottom>
	  <B>HDF5 Dataspace Structured/Irregular Grid Information</B>
	</caption>

	<tr align=center>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>

	<tr align=center>
	  <td colspan=4># of Grid Points in Dimension #1</td>
	<tr align=center>
	  <td colspan=4>.<br>.<br>.<br></td>
	<tr align=center>
	  <td colspan=4># of Grid Points in Dimension #n</td>
	<tr align=center>
	  <td colspan=4>Datatype of Grid Point Locations</td>
	<tr align=center>
	  <td colspan=4>Location of Grid Points in Dimension #1</td>
	<tr align=center>
	  <td colspan=4>.<br>.<br>.<br></td>
	<tr align=center>
	  <td colspan=4>Location of Grid Points in Dimension #n</td>
      </table>
    </center>

    <P>
    <center>
      <table border cellpadding=4 width="80%">
	<caption align=bottom>
	  <B>HDF5 Dataspace Unstructured Grid Information</B>
	</caption>

	<tr align=center>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>

	<tr align=center>
	  <td colspan=4># of Grid Points</td>
	<tr align=center>
	  <td colspan=4>Datatype of Grid Point Locations</td>
	<tr align=center>
	  <td colspan=4>Grid Point Locations<br>.<br>.<br></td>
      </table>
    </center>
-->

    <hr>
    <h4><a name="LinkInfoMessage">Name: Link Info</a></h4>

    <P class=item><B>Header Message Type:</B> 0x002 </P>
    <P class=item><B>Length:</B> varies </P>
    <P class=item><B>Status:</B> Optional, may not be repeated. </P>

    <P class=item><B>Description:</B>This message tracks variable information
        about the current state of the links for a "new style" group's behavior.
        Variable information
        will be stored in this message and constant information will be stored
        in the <a href="#GroupInfoMessage">Group Info</a> message.</P>
    </P>

    <P class=item><B>Format of Data:</B>
    <div align=center>
      <table class=format>
	<caption>
	  Link Info
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Flags</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Maximum Creation Index <em>(8 bytes, optional)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Fractal Heap Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Address of v2 B-tree for Name Index<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Address of v2 B-tree for Creation Order Index<sup>O</sup> <em>(optional)</em><br><br></td>
	</tr>
	
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>
            <P>The version number for this message. This document describes
                version 0.
	  </td>
	</tr>

	<tr>
	<td>Flags</td>
	  <td><P>This field determines various optional aspects of the link
                info message:
	      </P>
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>If set, creation order for the links is tracked.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>1</code></td>
                  <td>If set, creation order for the links is indexed.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>2-7</code></td>
                  <td>Reserved</td>
                </tr>
	    </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Maximum Creation Index</td>
	  <td><P>This 64-bit value is the maximum creation order index value
                stored for a link in this group.</P>
	    <P>This field is present if bit 0 of <em>flags</em> is set.</P>
	  </td>
	</tr>

	<tr>
	  <td>Fractal Heap Address</td>
	  <td>
            <P>
                This is the address of the fractal heap to store dense links.
                Each link stored in the fractal heap is stored as a
                <a href="#LinkMessage">Link Message</a>.
            </P>
            <P>
                If there are no links in the group, or the group's links
                are stored "compactly" (as object header messages), this
                value will be the <A href="#UndefinedAddress">undefined
                address</A>. 
            </P>
	  </td>
	</tr>

	<tr>
	  <td>Address of v2 B-tree for Name Index</td>
	  <td><P>This is the address of the version 2 B-tree to index names of links.</P>
                <P>If there are no links in the group, or the group's links
                    are stored "compactly" (as object header messages), this
                    value will be the <A href="#UndefinedAddress">undefined
                    address</A>. 
                </P>
	  </td>
	</tr>

	<tr>
	  <td>Address of v2 B-tree for Creation Order Index</td>
	  <td><P>This is the address of the version 2 B-tree to index creation order of links.</P>
                <P>If there are no links in the group, or the group's links
                    are stored "compactly" (as object header messages), this
                    value will be the <A href="#UndefinedAddress">undefined
                    address</A>. 
                </P>
	    <P>This field exists if bit 1 of <em>flags</em> is set.
	  </td>
	</tr>

      </table>
    </div>


    <hr>
    <h4><a name="DatatypeMessage">Name: Datatype</a></h4>

    <P class=item><B>Header Message Type:</B> 0x0003
    </P>
    <P class=item><B>Length:</B> variable
    </P>
    <P class=item><B>Status:</B> Required for dataset or named datatype objects,
        may not be repeated.
    </P>

    <P class=item><B>Description:</B> The datatype message defines the datatype
        for each element of a dataset or a common datatype for sharing between
        multiple datasets.  A datatype can describe an atomic type
        like a fixed- or floating-point type or more complex types like a C
        struct (compound datatype), array (array datatype) or C++ vector
        (variable-length datatype).
    </P>

    <P class=item2>Datatype messages that are part of a dataset object, 
        do not describe how elements are related to one another, the dataspace
        message is used for that purpose.  Datatype messages that are part of
        a named datatype message describe a common datatype that can be
        shared by multiple datasets in the file.
    </P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Datatype Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Class and Version</td>
	  <td>Class Bit Field, Bits 0-7</td>
	  <td>Class Bit Field, Bits 8-15</td>
	  <td>Class Bit Field, Bits 16-23</td>
	</tr>

	<tr>
	  <td colspan=4>Size</td>
	</tr>

	<tr>
	  <td colspan=4><br><br>Properties<br><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Class and Version</td>
	  <td>
            <P>The version of the datatype message and the datatype's class
                information are packed together in this field.  The version
                number is packed in the top 4 bits of the field and the class
                is contained in the bottom 4 bits.
            </P>
            <P>The version number information is used for changes in the
                format of the datatype message and is described here:
                <table class=list>
                    <tr>
                      <th width="30%">Version</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>Never used
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>1</code></td>
                      <td>Used by early versions of the library to encode
                            compound datatypes with explicit array fields.
                            See the compound datatype description below for
                            further details.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2</code></td>
                      <td>Used when an array datatype needs to be encoded.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>3</code></td>
                      <td>Used when a VAX byte-ordered type needs to be
                        encoded.  Packs various other datatype classes more
                        efficiently also.
                      </td>
                    </tr>
                </table>
            </P>
            <P>The class of the datatype determines the format for the class
                bit field and properties portion of the datatype message, which
                are described below.  The
                following classes are currently defined:
                <table width=100% class=list>
                    <tr>
                      <th width="30%">Value</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>Fixed-Point</td>
                    </tr>

                    <tr>
                      <td align=center><code>1</code></td>
                      <td>Floating-Point</td>
                    </tr>

                    <tr>
                      <td align=center><code>2</code></td>
                      <td>Time</td>
                    </tr>

                    <tr>
                      <td align=center><code>3</code></td>
                      <td>String</td>
                    </tr>

                    <tr>
                      <td align=center><code>4</code></td>
                      <td>Bitfield</td>
                    </tr>

                    <tr>
                      <td align=center><code>5</code></td>
                      <td>Opaque</td>
                    </tr>

                    <tr>
                      <td align=center><code>6</code></td>
                      <td>Compound</td>
                    </tr>

                    <tr>
                      <td align=center><code>7</code></td>
                      <td>Reference</td>
                    </tr>

                    <tr>
                      <td align=center><code>8</code></td>
                      <td>Enumerated</td>
                    </tr>

                    <tr>
                      <td align=center><code>9</code></td>
                      <td>Variable-Length</td>
                    </tr>

                    <tr>
                      <td align=center><code>10</code></td>
                      <td>Array</td>
                    </tr>
                </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Class Bit Fields</td>
	  <td>
            <P>The information in these bit fields is specific to each datatype
                class and is described below.  All bits not defined for a
                datatype class are set to zero.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size</td>
	  <td>
            <P>The size of a datatype element in bytes.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Properties</td>
	  <td>
            <P>This variable-sized sequence of bytes encodes information
                specific to each datatype class and is described for each class
                below.  If there is no property information specified for a
                datatype class, the size of this field is zero bytes.
            </P>
          </td>
	</tr>

      </table>
    </div>
    </P>

    <P>Class specific information for Fixed-Point Numbers (Class 0):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  Fixed-Point Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0</td>
	  <td><b>Byte Order.</b> If zero, byte order is little-endian;
	    otherwise, byte order is big endian.</td>
	</tr>

	<tr>
	  <td>1, 2</td>
	  <td><b>Padding type.</b>  Bit 1 is the lo_pad bit and bit 2
	    is the hi_pad bit.  If a datum has unused bits at either
	    end, then the lo_pad or hi_pad bit is copied to those
	    locations.</td>
	</tr>

	<tr>
	  <td>3</td>
	  <td><b>Signed.</b> If this bit is set then the fixed-point
	    number is in 2's complement form.</td>
	</tr>

	<tr>
	  <td>4-23</td>
	  <td>Reserved (zero).</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Fixed-Point Property Description
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=2>Bit Offset</td>
	  <td colspan=2>Bit Precision</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Bit Offset</td>
	  <td>
            <P>The bit offset of the first significant bit of the fixed-point
                value within the datatype.  The bit offset specifies the number
                of bits "to the right of" the value (which are set to the
                lo_pad bit value).
            </P>
          </td>
	</tr>

	<tr>
	  <td>Bit Precision</td>
	  <td>
            <P>The number of bits of precision of the fixed-point value
                within the datatype.  This value, combined with the datatype
                element's size and the Bit Offset field specifies the number
                of bits "to the left of" the value (which are set to the
                hi_pad bit value).
            </P>
          </td>
	</tr>

      </table>
    </div>
    </P>

    <P>Class specific information for Floating-Point Numbers (Class 1):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  Floating-Point Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0, 6</td>
	  <td><b>Byte Order.</b> These two non-contiguous bits specify the 
            "endianess" of the bytes in the datatype element.
            <table class=list>
                <tr>
                  <th width="20%">Bit 6</th>
                  <th width="20%">Bit 0</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td align=center><code>0</code></td>
                  <td>Byte order is little-endian
                  </td>
                </tr>
                <tr>
                  <td align=center><code>0</code></td>
                  <td align=center><code>1</code></td>
                  <td>Byte order is big-endian
                  </td>
                </tr>
                <tr>
                  <td align=center><code>1</code></td>
                  <td align=center><code>0</code></td>
                  <td>Reserved
                  </td>
                </tr>
                <tr>
                  <td align=center><code>1</code></td>
                  <td align=center><code>1</code></td>
                  <td>Byte order is VAX-endian
                  </td>
                </tr>
            </table>
          </td>
	</tr>

	<tr>
	  <td>1, 2, 3</td>
	  <td><b>Padding type.</b>  Bit 1 is the low bits pad type, bit 2
	    is the high bits pad type, and bit 3 is the internal bits
	    pad type.  If a datum has unused bits at either end or between
	    the sign bit, exponent, or mantissa, then the value of bit
	    1, 2, or 3 is copied to those locations.</td>
	</tr>

	<tr>
	  <td>4-5</td>
	  <td><b>Mantissa Normalization.</b> This 2-bit bit-field specifies
            how the most significant bit of the mantissa is managed.
                <table class=list>
                    <tr>
                      <th width="30%">Value</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>No normalization
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>1</code></td>
                      <td>The most significant bit of the mantissa is always set
                        (except for 0.0).
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2</code></td>
                      <td>The most significant bit of the mantissa is not stored,
                        but is implied to be set.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>3</code></td>
                      <td>Reserved.
                      </td>
                    </tr>
                </table>
            </td>
	</tr>

	<tr>
	  <td>7</td>
	  <td>Reserved (zero).</td>
	</tr>

	<tr>
	  <td>8-15</td>
	  <td><b>Sign Location.</b> This is the bit position of the sign
	    bit.  Bits are numbered with the least significant bit zero.</td>
	</tr>

	<tr>
	  <td>16-23</td>
	  <td>Reserved (zero).</td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Floating-Point Property Description
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=2>Bit Offset</td>
	  <td colspan=2>Bit Precision</td>
	</tr>

	<tr>
	  <td>Exponent Location</td>
	  <td>Exponent Size</td>
	  <td>Mantissa Location</td>
	  <td>Mantissa Size</td>
	</tr>

	<tr>
	  <td colspan=4>Exponent Bias</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Bit Offset</td>
	  <td>
            <P>The bit offset of the first significant bit of the floating-point
                value within the datatype.  The bit offset specifies the number
                of bits "to the right of" the value.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Bit Precision</td>
	  <td>
            <P>The number of bits of precision of the floating-point value
                within the datatype.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Exponent Location</td>
	  <td>
            <P>The bit position of the exponent field.  Bits are numbered with
                the least significant bit number zero.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Exponent Size</td>
	  <td>
            <P>The size of the exponent field in bits.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Mantissa Location</td>
	  <td>
            <P>The bit position of the mantissa field.  Bits are numbered with
                the least significant bit number zero.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Mantissa Size</td>
	  <td>
            <P>The size of the mantissa field in bits.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Exponent Bias</td>
	  <td>
            <P>The bias of the exponent field.
            </P>
          </td>
	</tr>

      </table>
    </div>
    </P>

    <P>Class specific information for Time (Class 2):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  Time Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0</td>
	  <td><b>Byte Order.</b> If zero, byte order is little-endian;
	    otherwise, byte order is big endian.</td>
	</tr>

	<tr>
	  <td>1-23</td>
	  <td>Reserved (zero).</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Time Property Description
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=2>Bit Precision</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Bit Precision</td>
	  <td>
            <P>The number of bits of precision of the time value.
            </P>
          </td>
	</tr>

      </table>
    </div>
    </P>

    <P>Class specific information for Strings (Class 3):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  String Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0-3</td>
	  <td><b>Padding type.</b>  This four-bit value determines the
	    type of padding to use for the string.  The values are:

            <table width=100% class=list>
                <tr>
                  <th width="30%">Value</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>Null Terminate: A zero byte marks the end of the
                    string and is guaranteed to be present after
                    converting a long string to a short string.  When
                    converting a short string to a long string the value is
                    padded with additional null characters as necessary.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>1</code></td>
                  <td>Null Pad: Null characters are added to the end of
                    the value during conversions from short values to long
                    values but conversion in the opposite direction simply
                    truncates the value.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>2</code></td>
                  <td>Space Pad: Space characters are added to the end of
                    the value during conversions from short values to long
                    values but conversion in the opposite direction simply
                    truncates the value.  This is the Fortran
                    representation of the string.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>3-15</code></td>
                  <td>Reserved
                  </td>
                </tr>
            </table>
	</tr>

	<tr>
	  <td>4-7</td>
	  <td><b>Character Set.</b>  The character set used to
	    encode the string.
            <table width=100% class=list>
                <tr>
                  <th width="30%">Value</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>ASCII character set encoding
                  </td>
                </tr>

                <tr>
                  <td align=center><code>1</code></td>
                  <td>UTF-8 character set encoding
                  </td>
                </tr>

                <tr>
                  <td align=center><code>2-15</code></td>
                  <td>Reserved
                  </td>
                </tr>
            </table>
          </td>
	</tr>

	<tr>
	  <td>8-23</td>
	  <td>Reserved (zero).</td>
	</tr>
      </table>
    </div>

    <P>There are no properties defined for the string class.
    </P>
    </P>

    <P>Class specific information for Bitfields (Class 4):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  Bitfield Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0</td>
	  <td><b>Byte Order.</b> If zero, byte order is little-endian;
	    otherwise, byte order is big endian.</td>
	</tr>

	<tr>
	  <td>1, 2</td>
	  <td><b>Padding type.</b>  Bit 1 is the lo_pad type and bit 2
	    is the hi_pad type.  If a datum has unused bits at either
	    end, then the lo_pad or hi_pad bit is copied to those
	    locations.</td>
	</tr>

	<tr>
	  <td>3-23</td>
	  <td>Reserved (zero).</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Bitfield Property Description
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=2>Bit Offset</td>
	  <td colspan=2>Bit Precision</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Bit Offset</td>
	  <td>
            <P>The bit offset of the first significant bit of the bitfield
                within the datatype.  The bit offset specifies the number
                of bits "to the right of" the value.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Bit Precision</td>
	  <td>
            <P>The number of bits of precision of the bitfield
                within the datatype.
            </P>
          </td>
	</tr>
      </table>
    </div>
    </P>

    <P>Class specific information for Opaque (Class 5):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  Opaque Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0-7</td>
	  <td>Length of ASCII tag in bytes.</td>
	</tr>

	<tr>
	  <td>8-23</td>
	  <td>Reserved (zero).</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Opaque Property Description
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=4><br>ASCII Tag<br>
	    <br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>ASCII Tag</td>
	  <td>
            <P>This NUL-terminated string provides a description for the
                opaque type.  It is NUL-padded to a multiple of 8 bytes.
            </P>
          </td>
	</tr>
      </table>
    </div>
    </P>

    <P>Class specific information for Compound (Class 6):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  Compound Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0-15</td>
	  <td><b>Number of Members.</b> This field contains the number
	    of members defined for the compound datatype.  The member
	    definitions are listed in the Properties field of the data
	    type message.
	</tr>

	<tr>
	  <td>16-23</td>
	  <td>Reserved (zero).</td>
	</tr>
      </table>
    </div>
    </P>

    <P>The Properties field of a compound datatype is a list of the
      member definitions of the compound datatype.  The member
      definitions appear one after another with no intervening bytes.
      The member types are described with a (recursively) encoded datatype
      message.

    <P>Note that the property descriptions are different for different
      versions of the datatype version.  Additionally note that the version
      0 datatype encoding is deprecated and has been replaced with later
      encodings in versions of the HDF5 library from the 1.4 release onward.

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Compound Properties Description for Datatype Version 1
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=4><br>Name<br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Byte Offset of Member</td>
	</tr>

	<tr>
	  <td>Dimensionality</td>
	  <td colspan=3>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension Permutation</td>
	</tr>

	<tr>
	  <td colspan=4>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension #1 Size (required)</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension #2 Size (required)</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension #3 Size (required)</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension #4 Size (required)</td>
	</tr>

	<tr>
	  <td colspan=4><br>Member Type Message<br><br></td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Name</td>
	  <td>
            <P>This NUL-terminated string provides a description for the
                opaque type.  It is NUL-padded to a multiple of 8 bytes.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Byte Offset of Member</td>
	  <td>
            <P>This is the byte offset of the member within the datatype.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimensionality</td>
	  <td>
            <P>If set to zero, this field indicates a scalar member.  If set
                to a value greater than zero, this field indicates that the
                member is an array of values.  For array members, the size of
                the array is indicated by the 'Size of Dimension n' field in
                this message.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimension Permutation</td>
	  <td>
            <P>This field was intended to allow an array field to have 
                it's dimensions permuted, but this was never implemented.
                This field should always be set to zero.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimension #n Size</td>
	  <td>
            <P>This field is the size of a dimension of the array field as
            stored in the file.  The first dimension stored in the list of
            dimensions is the slowest changing dimension and the last
            dimension stored is the fastest changing dimension.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Member Type Message</td>
	  <td>
            <P>This field is a datatype message describing the datatype of
            the member.
            </P>
          </td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Compound Properties Description for Datatype Version 2
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=4><br>Name<br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Byte Offset of Member</td>
	</tr>

	<tr>
	  <td colspan=4><br>Member Type Message<br><br></td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Name</td>
	  <td>
            <P>This NUL-terminated string provides a description for the
                opaque type.  It is NUL-padded to a multiple of 8 bytes.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Byte Offset of Member</td>
	  <td>
            <P>This is the byte offset of the member within the datatype.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Member Type Message</td>
	  <td>
            <P>This field is a datatype message describing the datatype of
            the member.
            </P>
          </td>
	</tr>

      </table>
    </div>
    </P>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Compound Properties Description for Datatype Version 3
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=4><br>Name<br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Byte Offset of Member <em>(variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Member Type Message<br><br></td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Name</td>
	  <td>
            <P>This NUL-terminated string provides a description for the
                opaque type.  It is <em>not</em> NUL-padded to a multiple of
                8 bytes.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Byte Offset of Member</td>
	  <td>
            <P>This is the byte offset of the member within the datatype.  The
                field size is the minimum # of bytes necessary, based on size
                of the datatype element.  For example, a datatype element size
                of less than 256 bytes uses a 1 byte length, a datatype element
                size of 256-65535 bytes uses a 2 byte length, etc.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Member Type Message</td>
	  <td>
            <P>This field is a datatype message describing the datatype of
            the member.
            </P>
          </td>
	</tr>

      </table>
    </div>
    </P>

    <P>Class specific information for Reference (Class 7):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  Reference Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0-3</td>
	  <td><b>Type.</b> This four-bit value contains the type of reference
            described.  The values defined are:

            <table width=100% class=list>
                <tr>
                  <th width="30%">Value</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>Object Reference: A reference to another object in this
                    HDF5 file.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>1</code></td>
                  <td>Dataset Region Reference: A reference to a region within
                    a dataset in this HDF5 file.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>2-15</code></td>
                  <td>Reserved
                  </td>
                </tr>
            </table>

            </td>
	</tr>

	<tr>
	  <td>4-23</td>
	  <td>Reserved (zero).</td>
	</tr>
      </table>
    </div>

    <P>There are no properties defined for the reference class.
    </P>
    </P>

    <P>Class specific information for Enumeration (Class 8):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  Enumeration Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0-15</td>
	  <td><b>Number of Members.</b> The number of name/value
	    pairs defined for the enumeration type.</td>
	</tr>

	<tr>
	  <td>16-23</td>
	  <td>Reserved (zero).</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Enumeration Property Description for Datatype Versions 1 & 2
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=4><br>Base Type<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Names<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Values<br><br></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Base Type</td>
	  <td>
            <P>Each enumeration type is based on some parent type, usually an
                integer. The information for that parent type is described
                recursively by this field.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Names</td>
	  <td>
            <P>The name for each name/value pair. Each name is stored as a null
                terminated ASCII string in a multiple of eight bytes. The names
                are in no particular order.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Values</td>
	  <td>
            <P>The list of values in the same order as the names.  The values
                are packed (no inter-value padding) and the size of each value
                is determined by the parent type.
            </P>
          </td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Enumeration Property Description for Datatype Version 3
	</caption>

	<tr>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	  <th width="25%">Byte</th>
	</tr>

	<tr>
	  <td colspan=4><br>Base Type<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Names<br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Values<br><br></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Base Type</td>
	  <td>
            <P>Each enumeration type is based on some parent type, usually an
                integer. The information for that parent type is described
                recursively by this field.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Names</td>
	  <td>
            <P>The name for each name/value pair. Each name is stored as a null
                terminated ASCII string, <em>not</em> padded to a multiple of
                eight bytes. The names are in no particular order.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Values</td>
	  <td>
            <P>The list of values in the same order as the names.  The values
                are packed (no inter-value padding) and the size of each value
                is determined by the parent type.
            </P>
          </td>
	</tr>

      </table>
    </div>
    </P>


    <P>Class specific information for Variable-Length (Class 9):

    <br>
    <div align=center>
      <table class=desc>
	<caption>
	  Variable-Length Bit Field Description
	</caption>

	<tr>
	  <th width="10%">Bits</th>
	  <th>Meaning</th>
	</tr>

	<tr>
	  <td>0-3</td>
	  <td><b>Type.</b> This four-bit value contains the type of
            variable-length datatype described.  The values defined are:

            <table width=100% class=list>
                <tr>
                  <th width="30%">Value</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>Sequence: A variable-length sequence of any datatype.
                    Variable-length sequences do not have padding or
                    character set information.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>1</code></td>
                  <td>String: A variable-length sequence of characters.
                    Variable-length strings have padding and character set
                    information.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>2-15</code></td>
                  <td>Reserved
                  </td>
                </tr>
            </table>

            </td>
	</tr>

	<tr>
	  <td>4-7</td>
	  <td><b>Padding type.</b> (variable-length string only)
                  This four-bit value determines the type of padding 
                  used for variable-length strings.  The values are the same 
                  as for the string padding type, as follows:
            <table width=100% class=list>
                <tr>
                  <th width="30%">Value</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>Null terminate: A zero byte marks the end of a string
                    and is guaranteed to be present after converting a long
                    string to a short string.  When converting a short string
                    to a long string, the value is padded with additional null
                    characters as necessary.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>1</code></td>
                  <td>Null pad: Null characters are added to the end of the
                    value during conversion from a short string to a longer
                    string.  Conversion from a long string to a shorter string
                    simply truncates the value.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>2</code></td>
                  <td>Space pad: Space characters are added to the end of the
                    value during conversion from a short string to a longer
                    string.  Conversion from a long string to a shorter string
                    simply truncates the value.  This is the Fortran
                    representation of the string.
                  </td>
                </tr>

                <tr>
                  <td align=center><code>3-15</code></td>
                  <td>Reserved
                  </td>
                </tr>
            </table>

            This value is set to zero for variable-length sequences.

          </td>
	</tr>

	<tr>
	  <td>8-11</td>
	  <td><b>Character Set.</b> (variable-length string only)
                  This four-bit value specifies the character set 
                  to be used for encoding the string:
            <table width=100% class=list>
                <tr>
                  <th width="30%">Value</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>ASCII character set encoding
                  </td>
                </tr>

                <tr>
                  <td align=center><code>1</code></td>
                  <td>UTF-8 character set encoding
                  </td>
                </tr>

                <tr>
                  <td align=center><code>2-15</code></td>
                  <td>Reserved
                  </td>
                </tr>
            </table>

            This value is set to zero for variable-length sequences.

          </td>
	</tr>

	<tr>
	  <td>12-23</td>
	  <td>Reserved (zero).</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Variable-Length Property Description
        </caption>

        <tr>
          <th width="25%">Byte</th>
          <th width="25%">Byte</th>
          <th width="25%">Byte</th>
          <th width="25%">Byte</th>
        </tr>

        <tr>
          <td colspan=4><br>Base Type<br><br></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Base Type</td>
	  <td>
            <P>Each variable-length type is based on some parent type.  The
                information for that parent type is described recursively by
                this field.
            </P>
          </td>
	</tr>

      </table>
    </div>
    </P>

    <P>Class specific information for Array (Class 10):

    <P>There are no bit fields defined for the array class.
    </P>

    <P>Note that the dimension information defined in the property for this
      datatype class is independent of dataspace information for a dataset.
      The dimension information here describes the dimensionality of the
      information within a data element (or a component of an element, if the
      array datatype is nested within another datatype) and the dataspace for a
      dataset describes the size and locations of the elements in a dataset.
    </P>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Array Property Description for Datatype Version 2
        </caption>

        <tr>
          <th width="25%">Byte</th>
          <th width="25%">Byte</th>
          <th width="25%">Byte</th>
          <th width="25%">Byte</th>
        </tr>

	<tr>
	  <td>Dimensionality</td>
	  <td colspan=3>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension #1 Size</td>
	</tr>
	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	</tr>
	<tr>
	  <td colspan=4>Dimension #n Size</td>
	</tr>

	<tr>
	  <td colspan=4>Permutation Index #1</td>
	</tr>
	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	</tr>
	<tr>
	  <td colspan=4>Permutation Index #n</td>
	</tr>

        <tr>
          <td colspan=4><br>Base Type<br><br></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Dimensionality</td>
	  <td>
            <P>This value is the number of dimensions that the array has.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimension #n Size</td>
	  <td>
            <P>This value is the size of the dimension of the array
	    as stored in the file.  The first dimension stored in
	    the list of dimensions is the slowest changing dimension
	    and the last dimension stored is the fastest changing
	    dimension.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Permutation Index #n</td>
	  <td>
            <P>This value is the index permutation used to map
	    each dimension from the canonical representation to an
	    alternate axis for each dimension.  Currently, dimension
            permutations are not supported and these indices should be set
            to the index position minus one (i.e. the first dimension should
            be set to 0, the second dimension should be set to 1, etc.)
            </P>
          </td>
	</tr>

	<tr>
	  <td>Base Type</td>
	  <td>
            <P>Each array type is based on some parent type.  The
                information for that parent type is described recursively by
                this field.
            </P>
          </td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          Array Property Description for Datatype Version 3
        </caption>

        <tr>
          <th width="25%">Byte</th>
          <th width="25%">Byte</th>
          <th width="25%">Byte</th>
          <th width="25%">Byte</th>
        </tr>

	<tr>
	  <td>Dimensionality</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Dimension #1 Size</td>
	</tr>
	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	</tr>
	<tr>
	  <td colspan=4>Dimension #n Size</td>
	</tr>

        <tr>
          <td colspan=4><br>Base Type<br><br></td>
        </tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Dimensionality</td>
	  <td>
            <P>This value is the number of dimensions that the array has.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dimension #n Size</td>
	  <td>
            <P>This value is the size of the dimension of the array
	    as stored in the file.  The first dimension stored in
	    the list of dimensions is the slowest changing dimension
	    and the last dimension stored is the fastest changing
	    dimension.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Base Type</td>
	  <td>
            <P>Each array type is based on some parent type.  The
                information for that parent type is described recursively by
                this field.
            </P>
          </td>
	</tr>

      </table>
    </div>

    </P>

    <hr>
    <h4><a name="OldFillValueMessage">Name: Data Storage - Fill Value (Old)</a></h4>

    <P class=item><B>Header Message Type:</B> 0x0004
    </P>
    <P class=item><B>Length:</B> varies
    </P>
    <P class=item><B>Status:</B> Optional, may not be repeated.
    </P>

    <P class=item><B>Description:</B> The fill value message stores a single
        data value which is returned to the application when an uninitialized
        data element is read from a dataset.  The fill value is interpreted
        with the same datatype as the dataset.  If no fill value message is
        present then a fill value of all zero bytes is assumed.
    </P>

    <P class=item2>This fill value message is deprecated in favor of the "new"
        fill value message (Message Type 0x0005) and is only written to the
        file for forward compatibility with versions of the HDF5 library before
        the 1.6.0 version.  Additionally, it only appears for datasets with a
        user defined fill value (as opposed to the library default fill value
        or an explicitly set "undefined" fill value).
    </P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Fill Value Message (Old)
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td colspan=4>Size</td>
	</tr>

	<tr>
	  <td colspan=4><br>Fill Value <em>(optional, variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Size</td>
	  <td>
            <P>This is the size of the Fill Value field in bytes.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Fill Value</td>
	  <td>
            <P>The fill value.  The bytes of the fill value are interpreted
                using the same datatype as for the dataset.
            </P>
          </td>
	</tr>
      </table>
    </div>
    </P>

    <hr>
    <h4><a name="FillValueMessage">Name: Data Storage - Fill Value </a></h4>

    <P class=item><B>Header Message Type:</B> 0x0005
    </P>
    <P class=item><B>Length:</B> varies
    </P>
    <P class=item><B>Status:</B> Required for dataset objects, may not be repeated.
    </P>

    <P class=item><B>Description:</B> The fill value message stores a single
        data value which is returned to the application when an uninitialized
        data element is read from a dataset.  The fill value is interpreted
        with the same datatype as the dataset.
    </P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Fill Value Message - Versions 1 & 2
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Space Allocation Time</td>
	  <td>Fill Value Write Time</td>
	  <td>Fill Value Defined</td>
	</tr>

	<tr>
	  <td colspan=4>Size <em>(optional)</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Fill Value <em>(optional, variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>
            <P>The version number information is used for changes in the
                format of the fill value message and is described here:
                <table class=list>
                    <tr>
                      <th width="30%">Version</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>Never used
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>1</code></td>
                      <td>Initial version of this message.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2</code></td>
                      <td>In this version, the Size and Fill Value fields are
                        only present if the Fill Value Defined field is set
                        to 1.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>3</code></td>
                      <td>This version packs the other fields in the message
                        more efficiently than version 2.
                      </td>
                    </tr>
                </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Space Allocation Time</td>
	  <td>
            <P>When the storage space for the dataset's raw data will be
                allocated.  The allowed values are:
                <table class=list>
                    <tr>
                      <th width="30%">Value</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>Not used.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>1</code></td>
                      <td>Early allocation.  Storage space for the entire dataset
                        should be allocated in the file when the dataset is
                        created.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2</code></td>
                      <td>Late allocation.  Storage space for the entire dataset
                        should not be allocated until the dataset is written
                        to.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>3</code></td>
                      <td>Incremental allocation.  Storage space for the
                        dataset should not be allocated until the portion
                        of the dataset is written to.  This is currently
                        used in conjunction with chunked data storage for
                        datasets.
                      </td>
                    </tr>
                </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Fill Value Write Time</td>
	  <td>
            <P>At the time that storage space for the dataset's raw data is
                allocated, this value indicates whether the fill value should
                be written to the raw data storage elements.  The allowed values
                are:
                <table class=list>
                    <tr>
                      <th width="30%">Value</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>On allocation.  The fill value is always written to
                        the raw data storage when the storage space is allocated.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>1</code></td>
                      <td>Never.  The fill value should never be written to
                        the raw data storage.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2</code></td>
                      <td>Fill value written if set by user.  The fill value
                        will be written to the raw data storage when the storage
                        space is allocated only if the user explicitly set
                        the fill value.  If the fill value is the library
                        default or is undefined, it will not be written to
                        the raw data storage.
                      </td>
                    </tr>
                </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Fill Value Defined</td>
	  <td>
            <P>This value indicates if a fill value is defined for this
                dataset.  If this value is 0, the fill value is undefined.
                If this value is 1, a fill value is defined for this dataset.
                For version 2 or later of the fill value message, this value
                controls the presence of the Size and Fill Value fields.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size</td>
	  <td>
            <P>This is the size of the Fill Value field in bytes.  This field
                is not present if the Version field is >1 and the Fill Value
                Defined field is set to 0.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Fill Value</td>
	  <td>
            <P>The fill value.  The bytes of the fill value are interpreted
                using the same datatype as for the dataset.  This field is
                not present if the Version field is >1 and the Fill Value
                Defined field is set to 0.
            </P>
          </td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Fill Value Message - Version 3
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Flags</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Size <em>(optional)</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Fill Value <em>(optional, variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>
            <P>The version number information is used for changes in the
                format of the fill value message and is described here:
                <table class=list>
                    <tr>
                      <th width="30%">Version</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>Never used
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>1</code></td>
                      <td>Initial version of this message.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2</code></td>
                      <td>In this version, the Size and Fill Value fields are
                        only present if the Fill Value Defined field is set
                        to 1.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>3</code></td>
                      <td>This version packs the other fields in the message
                        more efficiently than version 2.
                      </td>
                    </tr>
                </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Flags</td>
	  <td>
            <P>When the storage space for the dataset's raw data will be
                allocated.  The allowed values are:
                <table class=list>
                    <tr>
                      <th width="30%">Bits</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0-1</code></td>
                      <td>Space Allocation Time, with the same
                        values as versions 1 and 2 of the message.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2-3</code></td>
                      <td>Fill Value Write Time, with the same
                        values as versions 1 and 2 of the message.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>4</code></td>
                      <td>Fill Value Undefined, indicating that the fill
                        value has been marked as "undefined" for this dataset.
                        Bits 4 and 5 can not both be set.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>5</code></td>
                      <td>Fill Value Defined, with the same values as
                        versions 1 and 2 of the message.
                        Bits 4 and 5 can not both be set.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>6-7</code></td>
                      <td>Reserved (zero).
                      </td>
                    </tr>
                </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Size</td>
	  <td>
            <P>This is the size of the Fill Value field in bytes.  This field
                is not present if the Version field is >1 and the Fill Value
                Defined flag is set to 0.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Fill Value</td>
	  <td>
            <P>The fill value.  The bytes of the fill value are interpreted
                using the same datatype as for the dataset.  This field is
                not present if the Version field is >1 and the Fill Value
                Defined flag is set to 0.
            </P>
          </td>
	</tr>
      </table>
    </div>
    </P>

    <hr>
    <h4><a name="LinkMessage">Name: Link Message</a></h4>
    <P class=item><B>Header Message Type:</B> 0x0006</P>
    <P class=item><B>Length:</B> varies </P>
    <P class=item><B>Status:</B> Optional, may be repeated. </P>

    <P class=item><B>Description:</B>This message encodes the information
        for a link in a group's object header, when the group is storing its
        links "compactly", or in the group's fractal heap, when the group is
        storing its links "densely".
    </P>
    <P>
        A group is storing its links compactly when the fractal heap address in
        the <em><a href="#LinkInfoMessage">Link Info Message</a></em>
        is set to the "undefined address" value.
    </P>

    <P class=item><B>Format of Data:</B>
    <div align=center>
      <table class=format>
	<caption>
	  Link Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Flags</td>
	  <td>Link type <em>(optional)</em></td>
          <td bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>
	<tr>
	  <td colspan=4><br>Creation Order <em>(8 bytes, optional)</em><br><br></td>
	</tr>
	<tr>
	  <td>Link Name Character Set <em>(optional)</em></td>
	  <td>Length of Link Name (variable size)</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>
	<tr>
	  <td colspan=4>Link Name (variable size)</td>
	</tr>
	<tr>
	  <td colspan=4><br>Link Information (variable size)<br><br></td>
	</tr>
    </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number for this message. This document describes version 1.</P>
	  </td>
	</tr>

	<tr>
	<td>Flags</td>
	  <td><P>This field contains information about the link and controls
                the presence of other fields below.
                <table class=list>
                    <tr>
                      <th width="30%">Bits</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0-1</code></td>
                      <td>Determines the size of the <em>Length of Link Name</em>
                        field.
                        <table class=list>
                            <tr>
                              <th width="30%">Value</th>
                              <th align=left>Description</th>
                            </tr>

                            <tr>
                              <td align=center><code>0</code></td>
                              <td>The size of the <em>Length of Link Name</em>
                                field is 1 byte.
                              </td>
                            </tr>
                            <tr>
                              <td align=center><code>1</code></td>
                              <td>The size of the <em>Length of Link Name</em>
                                field is 2 bytes.
                              </td>
                            </tr>
                            <tr>
                              <td align=center><code>2</code></td>
                              <td>The size of the <em>Length of Link Name</em>
                                field is 4 bytes.
                              </td>
                            </tr>
                            <tr>
                              <td align=center><code>3</code></td>
                              <td>The size of the <em>Length of Link Name</em>
                                field is 8 bytes.
                              </td>
                            </tr>
                        </table>
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2</code></td>
                      <td>Creation Order Field Present: if set, the <em>Creation
                        Order</em> field is present.  If not set, creation order
                        information is not stored for links in this group.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>3</code></td>
                      <td>Link Type Field Present: if set, the link is not
                        a hard link and the <em>Link Type</em> field is present.
                        If not set, the link is a hard link.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>4</code></td>
                      <td>Link Name Character Set Field Present: if set, the
                        link name is not represented with the ASCII character
                        set and the <em>Link Name Character Set</em> field is
                        present.  If not set, the link name is represented with
                        the ASCII character set.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>5-7</code></td>
                      <td>Reserved (zero).
                      </td>
                    </tr>
                </table>
          </P>
          </td>
	</tr>

	<tr>
	  <td>Link type</td>
	  <td><P>This is the link class type and can be one of the following
                values:
                <table class=list>
                    <tr>
                      <th width="30%">Value</th>
                      <th align=left>Description</th>
                    </tr>

                    <tr>
                      <td align=center><code>0</code></td>
                      <td>A hard link (should never be stored in the file)
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>1</code></td>
                      <td>A soft link.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>2-63</code></td>
                      <td>Reserved for future HDF5 internal use.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>64</code></td>
                      <td>An external link.
                      </td>
                    </tr>
                    <tr>
                      <td align=center><code>65-255</code></td>
                      <td>Reserved, but available for user-defined link types.
                      </td>
                    </tr>
                </table>
            </P>
	    <P>This field is present if bit 3 of <em>Flags</em> is set.</P>
	  </td>
	</tr>

	<tr>
	  <td>Creation Order</td>
	  <td><P>This 64-bit value is an index of the link's creation time within
            the group.  Values start at 0 when the group is created an increment
            by one for each link added to the group.  Removing a link from a
            group does not change existing links' creation order field.
            </P>
	    <P>This field is present if bit 2 of <em>Flags</em> is set.</P>
	  </td>
	</tr>

	<tr>
	  <td>Link Name Character Set</td>
	  <td><P>This is the character set for encoding the link's name:
            <table width=100% class=list>
                <tr>
                  <th width="30%">Value</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>ASCII character set encoding (this should never be stored
                    in the file)
                  </td>
                </tr>

                <tr>
                  <td align=center><code>1</code></td>
                  <td>UTF-8 character set encoding
                  </td>
                </tr>
            </table>
            </P>
	    <P>This field is present if bit 4 of <em>Flags</em> is set.</P>
	  </td>
	</tr>

	<tr>
	  <td>Length of link name</td>
	  <td><P>This is the length of the link's name.  The size of this field 
	    depends on bits 0 and 1 of <em>Flags</em>.</P>
	  </td>
	</tr>

	<tr>
	  <td>Link name</td>
	  <td><P>This is the name of the link, non-NULL terminated.</P>
	  </td>
	</tr>

	<tr>
	  <td>Link information</td>
	  <td><P>The format of this field depends on the <em>link type</em>.</P>
            <P>For hard links, the field is formatted as follows:
            <table class=list>
                <tr>
                  <td width=30%><i>Size of Offsets</i> bytes:</td>
                  <td>The address of the object header for the object that the
                    link points to.
                  </td>
                </tr>
            </table>
            </P>

            <P>
            For soft links, the field is formatted as follows:
            <table class=list>
                <tr>
                  <td width=30%>Bytes 1-2:</td>
                  <td>Length of soft link value.</td>
                </tr>
                <tr>
                  <td><em>Length of soft link value</em> bytes:</td>
                  <td>A non-NULL-terminated string storing the value of the
                    soft link.
                  </td>
                </tr>
	    </table>

            <P>
            For external links, the field is formatted as follows:
            <table class=list>
                <tr>
                  <td width=30%>Bytes 1-2:</td>
                  <td>Length of external link value.</td>
                </tr>
                <tr>
                  <td><em>Length of external link value</em> bytes:</td>
                  <td>The first byte contains the version number in the upper 4 bits
		    and flags in the lower 4 bits for the external link. 
		    Both version and flags are defined to be zero in this document.
		    The remaining bytes consist of two NULL-terminated strings, with no padding between
                    them.  The first string is the name of the HDF5 file
                    containing the object linked to and the second string is
                    the full path to the object linked to, within the HDF5
                    file's group hierarchy.
                  </td>
                </tr>
            </table>

            <P>
            For user-defined links, the field is formatted as follows:
            <table class=list>
                <tr>
                  <td width=30%>Bytes 1-2:</td>
                  <td>Length of user-defined data.</td>
                </tr>
                <tr>
                  <td><em>Length of used-defined link value</em> bytes:</td>
                  <td>The data supplied for the user-defined link type.</td>
                </tr>
            </table>

          </td>
        </tr>
      </table>
    </div>

    <hr>
    <h4><a name="ExternalFileListMessage">Name: Data Storage -
	External Data Files</a></h4>
    <P class=item><B>Header Message Type:</B> 0x0007 </P>
    <P class=item><B>Length:</B> varies</P>
    <P class=item><B>Status:</B> Optional, may not be repeated.</P>

    <P class=item><B>Purpose and Description:</B> The external data storage
      message indicates that the data for an object is stored outside the HDF5
      file.  The filename of the object is stored as a Universal
      Resource Location (URL) of the actual filename containing the
      data. An external file list record also contains the byte offset
      of the start of the data within the file and the amount of space
      reserved in the file for that data.</P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  External File List Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td colspan=3>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=2>Allocated Slots</td>
	  <td colspan=2>Used Slots</td>
	</tr>

	<tr>
	  <td colspan=4><br>Heap Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Slot Definitions...<br><br></td>
	</tr>
      </table>
    </div>	

    <br>
    <div align=center>
	<table class=desc>
	  <tr>
	    <th width="30%">Field Name</th>
	    <th>Description</th>
	  </tr>
	
	  <tr>
	    <td>Version</td>
	    <td>
	      <P>The version number information is used for changes in the format of
                External Data Storage Message and is described here:
		 <table class=list>
		     <tr>
			<th width="30%">Version</th>
			<th align=left>Description</th>
		     </tr>
		     <tr>
			<td align=center><code>0</code></td>
			<td>Never used.
		     </tr>
		     <tr>
			<td align=center><code>1</code></td>
			<td>The current version used by the library.
		     </tr>
		  </table>		
	      </P>	
	    </td>
	  </tr>

	  <tr>
	    <td>Allocated Slots</td>
	    <td>
	      <P>The total number of slots allocated in the message.  Its value must be at least as
		 large as the value contained in the Used Slots field. (The current library simply
		 uses the number of Used Slots for this message)</P>
	    </td>
	  </tr>

	  <tr>
	    <td>Used Slots</td>
	    <td>
	      <P>The number of initial slots which contains valid information.</P>
	    </td>
	  </tr>

	  <tr>
	    <td>Heap Address</td>
	    <td>
	      <P>This is the address of a local heap which contains the names for the external 
		 files (The local heap information can be found in Disk Format Level 1D in this
		 document).  The name at offset zero in the heap is always the empty string.</P>
	    </td>
	  </tr>

	  <tr>
	    <td>Slot Definitions</td>
	    <td>
	      <P>The slot definitions are stored in order according to the array addresses they 
		 represent.</P>
	    </td>
	  </tr>

	</table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  External File List Slot 
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td colspan=4><br>Name Offset in Local Heap<sup>L</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Offset in External Data File<sup>L</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Data Size in External File<sup>L</sup><br><br></td>
	</tr>
      </table>
    </div>	

    <br>
    <div align=center>
	<table class=desc>
	  <tr>
	    <th width="30%">Field Name</th>
	    <th>Description</th>
	  </tr>
	
	  <tr>
	    <td>Name Offset in Local Heap</td>
	    <td>
		<P>The byte offset within the local name heap for the name
	    	of the file.  File names are stored as a URL which has a
	    	protocol name, a host name, a port number, and a file
	    	name:
		<code><em>protocol</em>:<em>port</em>//<em>host</em>/<em>file</em></code>.
		If the protocol is omitted then "file:" is assumed.  If
		the port number is omitted then a default port for that
		protocol is used.  If both the protocol and the port
		number are omitted then the colon can also be omitted. If
		the double slash and host name are omitted then
		"localhost" is assumed.  The file name is the only
		mandatory part, and if the leading slash is missing then
		it is relative to the application's current working
		directory (the use of relative names is not
		recommended).
                </P>
	    </td>
	  </tr>

	  <tr>
	    <td>Offset in External Data File</td>
	    <td>
	  	<P>This is the byte offset to the start of the data in the
	        specified file. For files that contain data for a single
	        dataset this will usually be zero.</P>
	    </td>
	  </tr>

	  <tr>
	    <td>Data Size in External File</td>
	    <td>
		<P>This is the total number of bytes reserved in the
		specified file for raw data storage.  For a file that
		contains exactly one complete dataset which is not
		extendable, the size will usually be the exact size of the
		dataset.  However, by making the size larger one allows
		HDF5 to extend the dataset. The size can be set to a value
		larger than the entire file since HDF5 will read zeros
		past the end of the file without failing.</P>
	    </td>
	  </tr>
	</table>
    </div>
    </P>


    <hr>
    <h4><a name="LayoutMessage">Name: Data Storage - Layout</a></h4>

    <P class=item><B>Header Message Type:</B> 0x0008</P>
    <P class=item><B>Length:</B> varies</P>
    <P class=item><B>Status:</B> Required for datasets, may not be repeated.</P>

    <P class=item><B>Purpose and Description:</B> Data layout describes how the
      elements of a multi-dimensional array are stored in the HDF5 file.
      Three types of data layout are supported:

    <ol>
      <li>Contiguous: The array is stored in one contiguous area of the file.
	This layout requires that the size of the array be constant and
	does not permit chunking, compression, checksums, encryption,
	etc.  The message stores the total storage size of the array.  The
	offset of an element from the beginning of the storage area is
	computed as in a C array.

      <li>Chunked: The array domain is regularly decomposed into chunks and
	each chunk is allocted and stored separately.  This layout supports
	arbitrary element traversals, compression, encryption, and
	checksums, etc.  (these features are described in other
	messages).  The message stores the size of a chunk instead of
	the size of the entire array; the storage size of the entire array can
	be calculated by traversing the B-tree that stores the chunk
	addresses.

      <li>Compact: The array is stored in one contiguous block, as part of
        this object header message.
    </ol>
    
    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Data Layout Message (Versions 1 and 2)
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Dimensionality</td>
	  <td>Layout Class</td>
	  <td>Reserved <em>(zero)</em></td>
	</tr>

	<tr>
	  <td colspan=4>Reserved <em>(zero)</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Data Address<sup>O</sup> <em>(optional)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Dimension 0 Size</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension 1 Size</td>
	</tr>
	
	<tr>
	  <td colspan=4>...</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension #n Size</td>
	</tr>
	
	<tr>
	  <td colspan=4>Dataset Element Size <em>(optional)</em></td>
	</tr>
	
	<tr>
	  <td colspan=4>Compact Data Size <em>(optional)</em></td>
	</tr>
	
	<tr>
	  <td colspan=4><br>Compact Data... <em>(variable size, optional)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
	  <th width="30%">Field Name</th>
	  <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>
	    <P>The version number information is used for changes in the format of the data 
	    layout message and is described here:</P>
	      <table class=list>
		<tr>
		  <th width="30%">Version</th>
		  <th align=left>Description</th>
		</tr>

		<tr>
		  <td align=center><code>0</code></td>
		  <td>Never used.</td>
		</tr>

		<tr>
		  <td align=center><code>1</code></td>
		  <td>Used by version 1.4 and before of the library to encode layout information.
		      Data space is always allocated when the data set is created.</td>
		</tr>

		<tr>
		  <td align=center><code>2</code></td>
		  <td>Used by version 1.6.x of the library to encode layout information.
		      Data space is allocated only when it is necessary.</td>
		</tr>
	      </table>
	  </td>
	</tr>

	<tr>
	  <td>Dimensionality</td>
	  <td><P>An array has a fixed dimensionality.  This field
	    specifies the number of dimension size fields later in the
	    message.  The value stored for chunked storage is 1 greater than
            the number of dimensions in the dataset's dataspace (i.e. 2 is
            stored for a 1 dimensional dataset, etc).
            </P>
          </td>
	</tr>

	<tr>
	  <td>Layout Class</td>
	  <td><P>The layout class specifies the type of storage for the data
            and how the other fields of the layout message are to be
            interpreted.
	      <table class=list>
		<tr>
		  <th width="30%">Value</th>
		  <th align=left>Description</th>
		</tr>

		<tr>
		  <td align=center><code>0</code></td>
		  <td>Compact Storage
                  </td>
		</tr>

		<tr>
		  <td align=center><code>1</code></td>
		  <td>Contiguous Storage
                  </td>
		</tr>

		<tr>
		  <td align=center><code>2</code></td>
		  <td>Chunked Storage
                  </td>
		</tr>
	      </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Data Address</td>
	  <td><P>For contiguous storage, this is the address of the raw
	    data in the file.  For chunked storage this is the address
	    of the v1 B-tree that is used to look up the addresses of the
	    chunks.  This field is not present for compact storage.
            If the version for this message is >1, the address
            may have the "undefined address" value, to indicate that
            storage has not yet been allocated for this array.</P>
          </td>
	</tr>

	<tr>
	  <td>Dimension #n Size</td>
	  <td><P>For contiguous and compact storage the dimensions define
            the entire size of the array while for chunked storage they define
            the size of a single chunk.  In all cases, they are in units of 
            array elements (not bytes).  The first dimension stored in the list
            of dimensions is the slowest changing dimension and the last
            dimension stored is the fastest changing dimension.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dataset Element Size</td>
	  <td><P>The size of a dataset element, in bytes.  This field is only
            present for chunked storage.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Compact Data Size</td>
	  <td><P>This field is only present for compact data storage.
            It contains the size of the raw data for the dataset array, in
            bytes.</P>
          </td>

	<tr>
	  <td>Compact Data</td>
	  <td><P>This field is only present for compact data storage.
            It contains the raw data for the dataset array.</P>
          </td>
	</tr>
      </table>
    </div>

    <br>
    <P>Version 3 of this message re-structured the format into specific
        properties that are required for each layout class.

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  <B>Data Layout Message (Version 3)</B>
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Layout Class</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Properties <em>(variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
	  <th width="30%">Field Name</th>
	  <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>
	    <P>The version number information is used for changes in the format of layout message
	       and is described here:</P>
	      <table class=list>	
		<tr>
		  <th width="30%">Version</th>
		  <th align=left>Description</th>
		</tr>

		<tr>
		  <td align=center><code>3</code></td>
		  <td>Used by the version 1.6.3 and later of the library to store properties 
		      for each layout class.</td>
		</tr>
	      </table>
	  </td>
	</tr>

	<tr>
	  <td>Layout Class</td>
	  <td><P>The layout class specifies the type of storage for the data
            and how the other fields of the layout message are to be
            interpreted.
	      <table class=list>
		<tr>
		  <th width="30%">Value</th>
		  <th align=left>Description</th>
		</tr>

		<tr>
		  <td align=center><code>0</code></td>
		  <td>Compact Storage
                  </td>
		</tr>

		<tr>
		  <td align=center><code>1</code></td>
		  <td>Contiguous Storage
                  </td>
		</tr>

		<tr>
		  <td align=center><code>2</code></td>
		  <td>Chunked Storage
                  </td>
		</tr>
	      </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Properties</td>
	  <td><P>This variable-sized field encodes information specific to each
            layout class and is described below.  If there is no property
            information specified for a layout class, the size of this field
            is zero bytes.</P></td>
	</tr>
      </table>
    </div>

    <br>
    <P>Class-specific information for compact layout (Class 0):  (Note: The dimensionality information
       is in the Dataspace message)

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Compact Storage Property Description
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td colspan=2>Size</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>
	
	<tr>
	  <td colspan=4><br>Raw Data... <em>(variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Size</td>
	  <td><P>This field contains the size of the raw data for the dataset
                array, in bytes.
              </P>
          </td>
	</tr>

	<tr>
	  <td>Raw Data</td>
	  <td><P>This field contains the raw data for the dataset array.</P></td>
	</tr>
      </table>
    </div>
    </P>

    <br>
    <P>Class-specific information for contiguous layout (Class 1):  (Note: The dimensionality information
       is in the Dataspace message)

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Contiguous Storage Property Description
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td colspan=4><br>Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Size<sup>L</sup><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Address</td>
	  <td><P>This is the address of the raw data in the file.
            The address may have the "undefined address" value, to indicate
            that storage has not yet been allocated for this array.</P></td>
	</tr>

	<tr>
	  <td>Size</td>
	  <td><P>This field contains the size allocated to store the raw data,
                in bytes.
              </P>
          </td>
	</tr>
      </table>
    </div>
    </P>

    <br>
    <P>Class-specific information for chunked layout (Class 2):

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Chunked Storage Property Description
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Dimensionality</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Dimension 0 Size</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension 1 Size</td>
	</tr>
	
	<tr>
	  <td colspan=4>...</td>
	</tr>

	<tr>
	  <td colspan=4>Dimension #n Size</td>
	</tr>

	<tr>
	  <td colspan=4>Dataset Element Size</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Dimensionality</td>
	  <td><P>A chunk has a fixed dimensionality.  This field specifies 
	    the number of dimension size fields later in the message.</P></td>
	</tr>

	<tr>
	  <td>Address</td>
	  <td><P>This is the address of the v1 B-tree that is used to look up the
            addresses of the chunks that actually store portions of the array
            data.  The address may have the "undefined address" value, to
            indicate that storage has not yet been allocated for this array.</P></td>
	</tr>

	<tr>
	  <td>Dimension #n Size</td>
	  <td><P>These values define the dimension size of a single chunk, in
            units of array elements (not bytes).  The first dimension stored in
            the list of dimensions is the slowest changing dimension and the
            last dimension stored is the fastest changing dimension.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dataset Element Size</td>
	  <td><P>The size of a dataset element, in bytes.
            </P>
          </td>
	</tr>
      </table>
    </div>

    <hr>
    <h4><a name="BogusMessage">Name: Bogus</a></h4>
    <P class=item><B>Header Message Type:</B> 0x0009</P>
    <P class=item><B>Length:</B> 4 bytes</P>
    <P class=item><B>Status:</B> For testing only, should never be stored in valid file</P>
    <P class=item><B>Purpose and Description:</B> This message is used for
        testing the HDF5 library's response to an "unknown" message type and
        should never be encounted in a valid HDF5 file.
    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Bogus Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td colspan=4>Bogus Value</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
	  <th width="30%">Field Name</th>
	  <th>Description</th>
	</tr>

	<tr>
	  <td>Bogus Value</td>
	  <td>
	    <P>This value should always be: <code>0xdeadbeef</code>.
	  </td>
	</tr>
      </table>
    </div>
    </P>

    <hr>
    <h4><a name="GroupInfoMessage">Name: Group Info</a></h4>
    <P class=item><B>Header Message Type:</B> 0x000A</P>
    <P class=item><B>Length:</B> varies</P>
    <P class=item><B>Status:</B> Optional, may not be repeated.</P>
    <P class=item><B>Description:</B> This message stores information for the
        constants defining a "new style" group's behavior.  Constant information
        will be stored in this message and variable information will be stored
        in the <a href="#LinkInfoMessage">Link Info</a> message.</P>

    <P>Note: the "estimated entry" information below is used when determining
        the size of the object header for the group when it is created.
    </P>

    <P class=item><B>Format of Data:</B>
    <div align=center>
      <table class=format>
	<caption>
	  Group Info Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Flags</td>
	  <td colspan=2>Link Phase Change: Maximum Compact Value <em>(optional)</em></td>
	</tr>
	<tr>
	  <td colspan=2>Link Phase Change: Minimum Dense Value <em>(optional)</em></td>
	  <td colspan=2>Estimated Number of Entries <em>(optional)</em></td>
	</tr>
	<tr>
	  <td colspan=2>Estimated Link Name Length of Entries <em>(optional)</em></td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number for this message. This document describes version 0.</P>
	  </td>
	</tr>

	<tr>
	<td>Flags</td>
	  <td><P>This is the group information flag with the following definition:
	      </P>
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>If set, link phase change values are stored.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>1</code></td>
                  <td>If set, the estimated entry information is non-default
                    and is stored.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>2-7</code></td>
                  <td>Reserved</td>
                </tr>
	    </table>
          </td>
	</tr>

	<tr>
	  <td>Link Phase Change: Maximum Compact Value</td>
	  <td><P>The is the maximum number of links to store "compactly" (in
                the group's object header).</P>
	      <P>This field is present if bit 0 of <em>Flags</em> is set.</P>
	  </td>
	</tr>

	<tr>
	  <td>Link Phase Change: Minimum Dense Value</td>
	  <td><P>This is the minimum number of links to store "densely" (in
                the group's fractal heap).  The fractal heap's address is
                located in the <a href="#LinkInfoMessage">Link Info</a>
                message.</P>
	      <P>This field is present if bit 0 of <em>Flags</em> is set.</P>
	  </td>
	</tr>

	<tr>
	  <td>Estimated Number of Entries</td>
	  <td><P>This is the estimated number of entries in groups.</P>
	      <P>If this field is not present, the default value of <code>4</code>
                will be used for the estimated number of group entries.</P>
	      <P>This field is present if bit 1 of <em>Flags</em> is set.</P>
	  </td>
	</tr>

	<tr>
	  <td>Estimated Link Name Length of Entries</td>
	  <td><P>This is the estimated length of entry name.</P>
	      <P>If this field is not present, the default value of <code>8</code>
                will be used for the estimated link name length of group entries.</P>
	      <P>This field is present if bit 1 of <em>Flags</em> is set.</P>
	  </td>
	</tr>

      </table>
    </div>
    </P>

    <hr>
    <h4><a name="FilterMessage">Name: Data Storage - Filter Pipeline</a></h4>
    <P class=item><B>Header Message Type:</B> 0x000B</P>
    <P class=item><B>Length:</B> varies</P>
    <P class=item><B>Status:</B> Optional, may not be repeated.</P>

    <P class=item><B>Description:</B> This message describes the
      filter pipeline which should be applied to the data stream by
      providing filter identification numbers, flags, a name, and
      client data.</P>

    <P>This message may be present in the object headers of both
        dataset and group objects. For datasets, it specifies the
        filters to apply to raw data. For groups, it specifies the
        filters to apply to the group's fractal heap. Currently,
        only datasets using chunked data storage use the filter
        pipeline on their raw data.</P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Filter Pipeline Message - Version 1
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>  

	<tr>
	  <td>Version</td>
	  <td>Number of Filters</td>
	  <td colspan=2>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4><br>Filter Description List <em>(variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number for this message.  This table
	    describes version 1.</P></td>
	</tr>

	<tr>
	  <td>Number of Filters</td>
	  <td><P>The total number of filters described in this
	    message. The maximum possible number of filters in a
	    message is 32.</P></td>
	</tr>

	<tr>
	  <td>Filter Description List</td>
	  <td><P>A description of each filter.  A filter description
	    appears in the next table.</P></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Filter Description
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>  

	<tr>
	  <td colspan=2>Filter Identification Value</td>
	  <td colspan=2>Name Length</td>
	</tr>

	<tr>
	  <td colspan=2>Flags</td>
	  <td colspan=2>Number Client Data Values</td>
	</tr>

	<tr>
	  <td colspan=4><br>Name <em>(variable size, optional)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Client Data <em>(variable size, optional)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Padding <em>(variable size, optional)</em></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Filter Identification Value</td>
	  <td>
            <p>
            This value, often referred to as a filter identifier,
            is designed to be a unique identifier for the filter. 
            Values from zero through 32,767 are reserved for filters 
            supported by The HDF Group in the HDF5 library and for 
            filters requested and supported by third parties. 
            Filters supported by The HDF Group are documented immediately
            below.  Information on 3rd-party filters can be found at 
            The HDF Group&rsquo;s 
            <a href="http://www.hdfgroup.org/services/contributions.html">
            Contributions</a> page.
            
            <p>
            To request a filter identifier, please contact 
            The HDF Group&rsquo;s Help Desk at
            <img src="Graphics/help.png" valign="center" height=14>.
            You will be asked to provide the following information:
            <ol>
              <li>Contact information for the developer requesting the
                  new identifier 
              <li>A short description of the new filter 
              <li>Links to any relevant information, including licensing 
                  information 
            </ol>
            <p>
            Values from 32768 to 65535 are reserved for non-distributed uses 
            (for example, internal company usage) or for application usage 
            when testing a feature. The HDF Group does not track or document 
            the use of the filters with identifiers from this range.

            <p>
            The filters currently in library version 1.8.0 are 
	    listed below:
	    <table class=list>
	      <tr>
		<th width="30%">Identification</th>
		<th align=left>Name</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
		<td>N/A</td>
		<td>Reserved</td>
	      </tr>

	      <tr>
		<td align=center><code>1</code></td>
		<td>deflate</td>
		<td>GZIP deflate compression</td>
	      </tr>

	      <tr>
		<td align=center><code>2</code></td>
		<td>shuffle</td>
		<td>Data element shuffling</td>
	      </tr>

	      <tr>
		<td align=center><code>3</code></td>
		<td>fletcher32</td>
		<td>Fletcher32 checksum</td>
	      </tr>

	      <tr>
		<td align=center><code>4</code></td>
		<td>szip</td>
		<td>SZIP compression</td>
	      </tr>

	      <tr>
		<td align=center><code>5</code></td>
		<td>nbit</td>
		<td>N-bit packing</td>
	      </tr>

	      <tr>
		<td align=center><code>6</code></td>
		<td>scaleoffset</td>
		<td>Scale and offset encoded values</td>
	      </tr>
	    </table>
	  </P></td>
	</tr>

	<tr>
	  <td>Name Length</td>
	  <td><P>Each filter has an optional null-terminated ASCII name
	    and this field holds the length of the name including the
	    null termination padded with nulls to be a multiple of
	    eight. If the filter has no name then a value of zero is
	    stored in this field.</P></td>
	</tr>

	<tr>
	  <td>Flags</td>
	  <td><P>The flags indicate certain properties for a filter.  The 
	    bit values defined so far are:</P>
	    <table class=list>
	      <tr>
		<th width="30%">Bit</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
	        <td>If set then the filter is an optional filter.
		During output, if an optional filter fails it will be
		silently skipped in the pipeline.</td>
	      </tr>

	      <tr>
		<td align=center><code>1-15</code></td>
	        <td>Reserved (zero)</td>
	      </tr>
	    </table>
	  </td>
	</tr>

	<tr>
	  <td>Number of Client Data Values</td>
	  <td><P>Each filter can store integer values to control
	    how the filter operates.  The number of entries in the
	    <em>Client Data</em> array is stored in this field.</P></td>
	</tr>

	<tr>
	  <td>Name</td>
	  <td><P>If the <em>Name Length</em> field is non-zero then it will
	    contain the size of this field, padded to a multiple of eight.  This 
	    field contains a null-terminated, ASCII character
	    string to serve as a comment/name for the filter.</P></td>
	</tr>

	<tr>
	  <td>Client Data</td>
	  <td><P>This is an array of four-byte integers which will be
	    passed to the filter function.  The <em>Client Data Number</em> of
	    Values determines the number of elements in the array.</P></td>
	</tr>

	<tr>
	  <td>Padding</td>
	  <td><P>Four bytes of zeros are added to the message at this
	    point if the Client Data Number of Values field contains
	    an odd number.</P></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Filter Pipeline Message - Version 2
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>  

	<tr>
	  <td>Version</td>
	  <td>Number of Filters</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Filter Description List <em>(variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number for this message.  This table
	    describes version 2.</P></td>
	</tr>

	<tr>
	  <td>Number of Filters</td>
	  <td><P>The total number of filters described in this
	    message. The maximum possible number of filters in a
	    message is 32.</P></td>
	</tr>

	<tr>
	  <td>Filter Description List</td>
	  <td><P>A description of each filter.  A filter description
	    appears in the next table.</P></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Filter Description
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>  

	<tr>
	  <td colspan=2>Filter Identification Value</td>
	  <td colspan=2>Name Length <em>(optional)</em></td>
	</tr>

	<tr>
	  <td colspan=2>Flags</td>
	  <td colspan=2>Number Client Data Values</td>
	</tr>

	<tr>
	  <td colspan=4><br>Name <em>(variable size, optional)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Client Data <em>(variable size, optional)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Filter Identification Value</td>
	  <td>
            <p>
            This value, often referred to as a filter identifier,
            is designed to be a unique identifier for the filter. 
            Values from zero through 32,767 are reserved for filters 
            supported by The HDF Group in the HDF5 library and for 
            filters requested and supported by third parties. 
            Filters supported by The HDF Group are documented immediately
            below.  Information on 3rd-party filters can be found at 
            The HDF Group&rsquo;s 
            <a href="http://www.hdfgroup.org/services/contributions.html">
            Contributions</a> page.
            
            <p>
            To request a filter identifier, please contact 
            The HDF Group&rsquo;s Help Desk at
            <img src="Graphics/help.png" valign="center" height=14>.
            You will be asked to provide the following information:
            <ol>
              <li>Contact information for the developer requesting the
                  new identifier 
              <li>A short description of the new filter 
              <li>Links to any relevant information, including licensing 
                  information 
            </ol>
            <p>
            Values from 32768 to 65535 are reserved for non-distributed uses 
            (for example, internal company usage) or for application usage 
            when testing a feature. The HDF Group does not track or document 
            the use of the filters with identifiers from this range.

            <p>
            The filters currently in library version 1.8.0 are 
	    listed below:
	    <table class=list>
	      <tr>
		<th width="30%">Identification</th>
		<th align=left>Name</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
		<td>N/A</td>
		<td>Reserved</td>
	      </tr>

	      <tr>
		<td align=center><code>1</code></td>
		<td>deflate</td>
		<td>GZIP deflate compression</td>
	      </tr>

	      <tr>
		<td align=center><code>2</code></td>
		<td>shuffle</td>
		<td>Data element shuffling</td>
	      </tr>

	      <tr>
		<td align=center><code>3</code></td>
		<td>fletcher32</td>
		<td>Fletcher32 checksum</td>
	      </tr>

	      <tr>
		<td align=center><code>4</code></td>
		<td>szip</td>
		<td>SZIP compression</td>
	      </tr>

	      <tr>
		<td align=center><code>5</code></td>
		<td>nbit</td>
		<td>N-bit packing</td>
	      </tr>

	      <tr>
		<td align=center><code>6</code></td>
		<td>scaleoffset</td>
		<td>Scale and offset encoded values</td>
	      </tr>
	    </table>
	  </P></td>
	</tr>

	<tr>
	  <td>Name Length</td>
	  <td><P>Each filter has an optional null-terminated ASCII name
	    and this field holds the length of the name including the
	    null termination padded with nulls to be a multiple of
	    eight. If the filter has no name then a value of zero is
	    stored in this field.</P>
            <P>Filters with IDs less than 256 (i.e. those that are defined in
                this format documentation)
                do not store the <em>Name Length</em> or
                <em>Name</em> fields.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Flags</td>
	  <td><P>The flags indicate certain properties for a filter.  The 
	    bit values defined so far are:</P>
	    <table class=list>
	      <tr>
		<th width="30%">Bit</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
	        <td>If set then the filter is an optional filter.
		During output, if an optional filter fails it will be
		silently skipped in the pipeline.</td>
	      </tr>

	      <tr>
		<td align=center><code>1-15</code></td>
	        <td>Reserved (zero)</td>
	      </tr>
	    </table>
	  </td>
	</tr>

	<tr>
	  <td>Number of Client Data Values</td>
	  <td><P>Each filter can store integer values to control
	    how the filter operates.  The number of entries in the
	    <em>Client Data</em> array is stored in this field.</P></td>
	</tr>

	<tr>
	  <td>Name</td>
	  <td><P>If the <em>Name Length</em> field is non-zero then it will
                contain the size of this field, <em>not</em> padded to a multiple
                of eight.  This field contains a <em>non-</em>null-terminated,
                ASCII character string to serve as a comment/name for the filter.
            </P>
            <P>Filters that are defined in this format documentation (deflate,
                shuffle, etc.) do not store the <em>Name Length</em> or
                <em>Name</em> fields.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Client Data</td>
	  <td><P>This is an array of four-byte integers which will be
	    passed to the filter function.  The Client Data Number of
	    Values</em> determines the number of elements in the array.</P>
          </td>
	</tr>
      </table>
    </div>
    <p>
    </P>

    <hr>
    <h4><a name="AttributeMessage">Name: Attribute</a></h4>
    <P class=item><B>Header Message Type:</B> 0x000C
    <P class=item><B>Length:</B> varies
    <P class=item><B>Status:</B> Optional, may be repeated.
    
    <P class=item><B>Description:</B>  The <em>Attribute</em>
      message is used to store objects in the HDF5 file which are used
      as attributes, or &ldquo;metadata&rdquo; about the current object.  An
      attribute is a small dataset; it has a name, a datatype, a data
      space, and raw data.  Since attributes are stored in the object
      header, they should be relatively small (in other words, less than 64KB). 
      They can be associated with any type of object which has an object 
      header (groups, datasets, or committed (named) datatypes).</p>

    <p class="item2">In 1.8.x versions of the library, attributes can be larger than 
      64KB. See the 
      <a href="UG/13_Attributes.html#SpecIssues">&ldquo;Special Issues&rdquo;</a> 
      section of the Attributes chapter in the <cite>HDF5 User's Guide</cite> 
      for more information.</p>
    
    <P class=item2>Note: Attributes on an object must have unique names 
      (The HDF5 library currently enforces this by causing the creation 
      of an attribute with a duplicate name to fail). Attributes on 
      different objects may have the same name, however.</p>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Attribute Message (Version 1)
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Reserved (zero)</td>
	  <td colspan=2>Name Size</td>
	</tr>

	<tr>
	  <td colspan=2>Datatype Size</td>
	  <td colspan=2>Dataspace Size</td>
	</tr>

	<tr>
	  <td colspan=4><br>Name <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Datatype <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Dataspace <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Data <em>(variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number information is used for changes in the format of the 
	    attribute message and is described here:</P>
	    <table class=list>
	      <tr>
		<th width="30%">Version</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
		<td>Never used.</td>
	      </tr>

	      <tr>
		<td align=center><code>1</code></td>
		<td>Used by the library before version 1.6 to encode attribute message.
		    This version does not support shared datatypes.</td>
	      </tr>
	    </table>
	  </td>
	</tr>

	<tr>
	  <td>Name Size</td>
	  <td><P>The length of the attribute name in bytes including the
	    null terminator.  Note that the <em>Name</em> field below may
	    contain additional padding not represented by this
	    field.</P></td>
	</tr>

	<tr>
	  <td>Datatype Size</td>
	  <td><P>The length of the datatype description in the <em>Datatype</em>
	    field below.  Note that the <em>Datatype</em> field may contain
	    additional padding not represented by this field.</P></td>
	</tr>

	<tr>
	  <td>Dataspace Size</td>
	  <td><P>The length of the dataspace description in the <em>Dataspace</em>
	    field below.  Note that the <em>Dataspace</em> field may contain
	    additional padding not represented by this field.</P></td>
	</tr>

	<tr>
	  <td>Name</td>
	  <td><P>The null-terminated attribute name.  This field is
	    padded with additional null characters to make it a
	    multiple of eight bytes.</P></td>
	</tr>

	<tr>
	  <td>Datatype</td>
	  <td><P>The datatype description follows the same format as
	    described for the datatype object header message.  This
	    field is padded with additional zero bytes to make it a
	    multiple of eight bytes.</P></td>
	</tr>

	<tr>
	  <td>Dataspace</td>
	  <td><P>The dataspace description follows the same format as
	    described for the dataspace object header message.  This
	    field is padded with additional zero bytes to make it a
	    multiple of eight bytes.</P></td>
	</tr>

	<tr>
	  <td>Data</td>
	  <td><P>The raw data for the attribute.  The size is determined
	    from the datatype and dataspace descriptions.  This
	    field is <em>not</em> padded with additional bytes.</P></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Attribute Message (Version 2)
	</caption>

	<tr align=center>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Flags</td>
	  <td colspan=2>Name Size</td>
	</tr>

	<tr>
	  <td colspan=2>Datatype Size</td>
	  <td colspan=2>Dataspace Size</td>
	</tr>

	<tr>
	  <td colspan=4><br>Name <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Datatype <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Dataspace <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Data <em>(variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number information is used for changes in the
                format of the attribute message and is described here:</P>
                <table class=list width="90%">
                  <tr>
                    <th width="30%">Version</th>
                    <th align=left>Description</th>
                  </tr>

                  <tr>
                    <td align=center><code>2</code></td>
                    <td>Used by the library of version 1.6.x and after to encode
                        attribute messages.
                        This version supports shared datatypes.  The fields of
                        name, datatype, and dataspace are not padded with
                        additional bytes of zero.
                    </td>
                  </tr>
                </table>
	  </td>
	</tr>

	<tr>
	  <td>Flags</td>
	  <td><P>This bit-field contains extra information about 
                interpreting the attribute message:
              </P>
	    <table class=list width="90%">
	      <tr>
		<th width="30%">Bit</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
		<td>If set, datatype is shared.</td>
	      </tr>

	      <tr>
		<td align=center><code>1</code></td>
		<td>If set, dataspace is shared.</td>
	      </tr>
	    </table>
	  </td>
	</tr>

	<tr>
	  <td>Name Size</td>
	  <td><P>The length of the attribute name in bytes including the
	    null terminator.</P></td>
	</tr>

	<tr>
	  <td>Datatype Size</td>
	  <td><P>The length of the datatype description in the <em>Datatype</em>
	    field below.</P></td>
	</tr>

	<tr>
	  <td>Dataspace Size</td>
	  <td><P>The length of the dataspace description in the <em>Dataspace</em>
	    field below.</P></td>
	</tr>

	<tr>
	  <td>Name</td>
	  <td><P>The null-terminated attribute name.  This field is <em>not</em>
	    padded with additional bytes.</P></td>
	</tr>

	<tr>
	  <td>Datatype</td>
	  <td><P>The datatype description follows the same format as
                described for the datatype object header message.
            </P>
            <P>If the
                <em>Flag</em> field indicates this attribute's datatype is
                shared, this field will contain a "shared message" encoding
                instead of the datatype encoding.
            </P>
            <P>This field is <em>not</em> padded with additional bytes.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dataspace</td>
	  <td><P>The dataspace description follows the same format as
                described for the dataspace object header message.
            </P>
            <P>If the
                <em>Flag</em> field indicates this attribute's dataspace is
                shared, this field will contain a "shared message" encoding
                instead of the dataspace encoding.
            </P>
            <P>This field is <em>not</em> padded with additional bytes.</P>
          </td>
	</tr>

	<tr>
	  <td>Data</td>
	  <td><P>The raw data for the attribute.  The size is determined
                from the datatype and dataspace descriptions.
            </P>
            <P>This field is <em>not</em> padded with additional zero bytes.
            </P>
          </td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Attribute Message (Version 3)
	</caption>

	<tr align=center>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Flags</td>
	  <td colspan=2>Name Size</td>
	</tr>

	<tr>
	  <td colspan=2>Datatype Size</td>
	  <td colspan=2>Dataspace Size</td>
	</tr>

	<tr>
	  <td>Name Character Set Encoding</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Name <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Datatype <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Dataspace <em>(variable size)</em><br><br></td>
	</tr>

	<tr>
	  <td colspan=4><br>Data <em>(variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number information is used for changes in the
                format of the attribute message and is described here:</P>
                <table class=list width="90%">
                  <tr>
                    <th width="30%">Version</th>
                    <th align=left>Description</th>
                  </tr>

                  <tr>
                    <td align=center><code>3</code></td>
                    <td>Used by the library of version 1.8.x and after to
                        encode attribute messages.
                        This version supports attributes with non-ASCII names.
                    </td>
                  </tr>
                </table>
	  </td>
	</tr>

	<tr>
	  <td>Flags</td>
	  <td><P>This bit-field contains extra information about 
                interpreting the attribute message:
              </P>
	    <table class=list width="90%">
	      <tr>
		<th width="30%">Bit</th>
		<th align=left>Description</th>
	      </tr>

	      <tr>
		<td align=center><code>0</code></td>
		<td>If set, datatype is shared.</td>
	      </tr>

	      <tr>
		<td align=center><code>1</code></td>
		<td>If set, dataspace is shared.</td>
	      </tr>
	    </table>
	  </td>
	</tr>

	<tr>
	  <td>Name Size</td>
	  <td><P>The length of the attribute name in bytes including the
	    null terminator.</P></td>
	</tr>

	<tr>
	  <td>Datatype Size</td>
	  <td><P>The length of the datatype description in the <em>Datatype</em>
	    field below.</P></td>
	</tr>

	<tr>
	  <td>Dataspace Size</td>
	  <td><P>The length of the dataspace description in the <em>Dataspace</em>
	    field below.</P></td>
	</tr>

	<tr>
	  <td>Name Character Set Encoding</td>
	  <td><P>The character set encoding for the attribute's name:
            <table width=100% class=list>
                <tr>
                  <th width="30%">Value</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>ASCII character set encoding
                  </td>
                </tr>

                <tr>
                  <td align=center><code>1</code></td>
                  <td>UTF-8 character set encoding
                  </td>
                </tr>
            </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Name</td>
	  <td><P>The null-terminated attribute name.  This field is <em>not</em>
	    padded with additional bytes.</P></td>
	</tr>

	<tr>
	  <td>Datatype</td>
	  <td><P>The datatype description follows the same format as
                described for the datatype object header message.
            </P>
            <P>If the
                <em>Flag</em> field indicates this attribute's datatype is
                shared, this field will contain a "shared message" encoding
                instead of the datatype encoding.
            </P>
            <P>This field is <em>not</em> padded with additional bytes.
            </P>
          </td>
	</tr>

	<tr>
	  <td>Dataspace</td>
	  <td><P>The dataspace description follows the same format as
                described for the dataspace object header message.
            </P>
            <P>If the
                <em>Flag</em> field indicates this attribute's dataspace is
                shared, this field will contain a "shared message" encoding
                instead of the dataspace encoding.
            </P>
            <P>This field is <em>not</em> padded with additional bytes.</P>
          </td>
	</tr>

	<tr>
	  <td>Data</td>
	  <td><P>The raw data for the attribute.  The size is determined
                from the datatype and dataspace descriptions.
            </P>
            <P>This field is <em>not</em> padded with additional zero bytes.
            </P>
          </td>
	</tr>
      </table>
    </div>
    </P>
    
    <hr>
    <h4><a name="CommentMessage">Name: Object Comment</a></h4>

    <P class=item><B>Header Message Type:</B> 0x000D</P>
    <P class=item><B>Length:</B> varies</P>
    <P class=item><B>Status:</B> Optional, may not be repeated.</P>

    <P class=item><B>Description:</B> The object comment is
      designed to be a short description of an object.  An object comment
      is a sequence of non-zero (<code>\0</code>) ASCII characters with no
      other formatting included by the library.</P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Name Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td colspan=4><br>Comment <em>(variable size)</em><br><br></td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Name</td>
	  <td>A null terminated ASCII character string.</td>
	</tr>
      </table>
    </div>
    </P>
    
    <hr>
    <h4><a name="OldModificationTimeMessage">Name: Object Modification Time (Old)</a></h4>

    <P class=item><B>Header Message Type:</B> 0x000E</P>
    <P class=item><B>Length:</B> fixed</P>
    <P class=item><B>Status:</B> Optional, may not be repeated.</P>

    <P class=item><B>Description:</B>  The object modification date
      and time is a timestamp which indicates (using ISO-8601 date and
      time format) the last modification of an object.  The time is
      updated when any object header message changes according to the
      system clock where the change was posted.  All fields of this message
      should be interpreted as coordinated universal time (UTC).

      <br><br>This modification time message is deprecated in favor of the "new"
      <a href="#ModificationTimeMessage">Object Modification Time</a>
      message and is no longer written
      to the file in versions of the HDF5 library after the 1.6.0 version.
    </P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Modification Time Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td colspan=4>Year</td>
	</tr>

	<tr>
	  <td colspan=2>Month</td>
	  <td colspan=2>Day of Month</td>
	</tr>

	<tr>
	  <td colspan=2>Hour</td>
	  <td colspan=2>Minute</td>
	</tr>

	<tr>
	  <td colspan=2>Second</td>
	  <td colspan=2>Reserved</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Year</td>
	  <td><P>The four-digit year as an ASCII string. For example,
	    <code>1998</code>.
            </P></td>
	</tr>

	<tr>
	  <td>Month</td>
	  <td><P>The month number as a two digit ASCII string where
	    January is <code>01</code> and December is <code>12</code>.</P></td>
	</tr>

	<tr>
	  <td>Day of Month</td>
	  <td><P>The day number within the month as a two digit ASCII
	    string. The first day of the month is <code>01</code>.</P></td>
	</tr>

	<tr>
	  <td>Hour</td>
	  <td><P>The hour of the day as a two digit ASCII string where
	    midnight is <code>00</code> and 11:00pm is <code>23</code>.</P></td>
	</tr>

	<tr>
	  <td>Minute</td>
	  <td><P>The minute of the hour as a two digit ASCII string where
	    the first minute of the hour is <code>00</code> and 
	    the last is <code>59</code>.</P></td>
	</tr>

	<tr>
	  <td>Second</td>
	  <td><P>The second of the minute as a two digit ASCII string
	    where the first second of the minute is <code>00</code> 
	    and the last is <code>59</code>.</P></td>
	</tr>

	<tr>
	  <td>Reserved</td>
	  <td><P>This field is reserved and should always be zero.</P></td>
	</tr>
      </table>
    </div>
    </P>
    
    <hr>
    <h4><a name="SOHMTableMessage">Name: Shared Message Table</a></h4>
    <P class=item><B>Header Message Type:</B> 0x000F</P>
    <P class=item><B>Length:</B> fixed </P>
    <P class=item><B>Status:</B> Optional, may not be repeated.</P>

    <P class=item><B>Description:</B>  This message is used to locate the
        table of shared object header message (SOHM) indexes.  Each
        index consists of information to find the shared messages from either
        the heap or object header.  This message is <em>only</em> found in the
        superblock extension.
    </P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Shared Message Table Message
	</caption>

	<tr>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	  <th width="25%">byte</td>
	</tr>

	<tr>
	  <td>Version</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Shared Object Header Message Table Address<sup>O</sup><br><br></td>
	</tr>

	<tr>
	  <td>Number of Indices</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number for this message.  This document describes version 0.</td>
	</tr>

	<tr>
	  <td>Shared Object Header Message Table Address</td>
	  <td><P>This field is the address of the master table for shared
            object header message indexes.</P>
	  </td>
	</tr>

	<tr>
	  <td>Number of Indices</td>
	  <td><P>This field is the number of indices in the master table.</P>
	</tr>

      </table>
    </div>
    </P>

    <hr>
    <h4><a name="ContinuationMessage">Name: Object Header Continuation</a></h4>
    <P class=item><B>Header Message Type:</B> 0x0010</P>
    <P class=item><B>Length:</B> fixed</P>
    <P class=item><B>Status:</B> Optional, may be repeated.</P>
    <P class=item><B>Description:</B>  The object header continuation is the
        location in the file of a block containing more header messages for the
        current data object.  This can be used when header blocks become too
        large or are likely to change over time.</P>

    <P class=item><B>Format of Data:</B>
    <br>
    <div align=center>
      <table class=format>
        <caption>
          Object Header Continuation Message
        </caption>

        <tr>
          <th width=25%>byte</th>
          <th width=25%>byte</th>
          <th width=25%>byte</th>
          <th width=25%>byte</th>
        </tr>

        <tr>
          <td colspan=4><br>Offset<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Length<sup>L</sup><br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
        <tr>
          <th width=30%>Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>Offset</td>
          <td><P>This value is the address in the file where the
              header continuation block is located.</P></td>
        </tr>

        <tr>
          <td>Length</td>
          <td><P>This value is the length in bytes of the header continuation
              block in the file.</P></td>
        </tr>
      </table>
    </div>
    </P>

    <P>The format of the header continuation block that this message points
        to depends on the version of the object header that the message is
        contained within.
    </P>

    <P>
        Continuation blocks for version 1 object headers have no special
        formatting information, they are merely a list of object header
        message info sequences (type, size, flags, reserved bytes and data
        for each message sequence).  See the description 
	of <a href="#V1ObjectHeaderPrefix">Version 1 Data Object Header Prefix.</a>
    </P>

    <P>Continuation blocks for version 2 object headers <em>do</em> have
        special formatting information as described here
        (see also the description of
	<a href="#V2ObjectHeaderPrefix">Version 2 Data Object Header Prefix.</a>):
    <br>
    <div align=center>
      <table class=format>
	<caption>
	  Version 2 Object Header Continuation Block
	</caption>

	<tr>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	  <th>byte</th>
	</tr>

	<tr>
	  <td colspan=4>Signature</td>
	</tr>
	<tr>
	  <td>Header Message Type #1</td>
	  <td colspan=2>Size of Header Message Data #1</td>
	  <td>Header Message #1 Flags</td>
	</tr>

	<tr>
	  <td colspan=2>Header Message #1 Creation Order <em>(optional)</em></td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Header Message Data #1<br><br></td>
	</tr>

	<tr>
	  <td colspan=4>.<br>.<br>.<br></td>
	</tr>

	<tr>
	  <td>Header Message Type #n</td>
	  <td colspan=2>Size of Header Message Data #n</td>
	  <td>Header Message #n Flags</td>
	</tr>

	<tr>
	  <td colspan=2>Header Message #n Creation Order <em>(optional)</em></td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br>Header Message Data #n<br><br></td>
	</tr>

	<tr>
	  <td colspan=4>Gap <em>(optional, variable size)</em></td>
	</tr>

	<tr>
	  <td colspan=4>Checksum</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Signature</td>
	  <td>
            <P>The ASCII character string "<code>OCHK</code>"
            is used to indicate the
            beginning of an object header continuation block.  This gives file
            consistency checking utilities a better chance of reconstructing a
            damaged file.
	    </P>
	  </td>
	</tr>
	
	<tr>
	  <td>Header Message #n Type</td>
	  <td>
            <P>Same format as version 1 of the object header, described above.
          </td>
	</tr>

	<tr>
	  <td>Size of Header Message #n Data</td>
	  <td>
            <P>Same format as version 1 of the object header, described above.
          </td>
	</tr>

	<tr>
	  <td>Header Message #n Flags</td>
	  <td>
            <P>Same format as version 1 of the object header, described above.
          </td>
        </tr>

	<tr>
	  <td>Header Message #n Creation Order</td>
	  <td>
            <P>This field stores the order that a message of a given type
                was created in.
	    <P>This field is present if bit 2 of <em>flags</em> is set.</P>
          </td>
        </tr>

	<tr>
	  <td>Header Message #n Data</td>
	  <td>
            <P>Same format as version 1 of the object header, described above.
          </td>
	</tr>

	<tr>
	  <td>Gap</td>
	  <td>
            <P>A gap in an object header chunk is inferred by the end of the
                messages for the chunk before the beginning of the chunk's
                checksum.  Gaps are always smaller than the size of an
                object header message prefix (message type + message size +
                message flags).
            <P>Gaps are formed when a message (typically an attribute message)
                in an earlier chunk is deleted and a message from a later
                chunk that doesn't quite fit into the free space is moved
                into the earlier chunk.
          </td>
	</tr>

        <tr>
          <td>Checksum</td>
          <td>
            <P>This is the checksum for the object header chunk.
            </P>
          </td>
        </tr>
      </table>
    </div>
    </P>

    <hr>
    <h4><a name="SymbolTableMessage">Name: Symbol Table Message</a></h4>
    <P class=item><B>Header Message Type:</B> 0x0011</P>
    <P class=item><B>Length:</B> fixed</P>
    <P class=item><B>Status:</B> Required for "old style" groups, may not be repeated.</P>
    <P class=item><B>Description:</B> Each "old style" group has a v1 B-tree
        and a local heap for storing symbol table entries, which are located
        with this message.</P>

    <P class=item><B>Format of data:</B>

    <br>
    <div align=center>
      <table class=format>
        <caption>
          <B>Symbol Table Message</B>
        </caption>

        <tr>
          <th width="25%">byte</th>
          <th width="25%">byte</th>
          <th width="25%">byte</th>
          <th width="25%">byte</th>
        </tr>

        <tr>
          <td colspan=4><br>v1 B-tree Address<sup>O</sup><br><br></td>
        </tr>

        <tr>
          <td colspan=4><br>Local Heap Address<sup>O</sup><br><br></td>
        </tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
        <tr>
          <th width=30%>Field Name</th>
          <th>Description</th>
        </tr>

        <tr>
          <td>v1 B-tree Address</td>
          <td><P>This value is the address of the v1 B-tree containing the
            symbol table entries for the group.</P></td>
        </tr>

        <tr>
          <td>Local Heap Address</td>
          <td><P>This value is the address of the local heap containing
            the link names for the symbol table entries for the group.</P></td>
        </tr>
      </table>
    </div>
    </P>

    <hr>
    <h4><a name="ModificationTimeMessage">Name: Object Modification Time</a></h4>

    <P class=item><B>Header Message Type:</B> 0x0012 </P>
    <P class=item><B>Length:</B> Fixed </P>
    <P class=item><B>Status:</B> Optional, may not be repeated. </P>

    <P class=item><B>Description:</B> The object modification 
      time is a timestamp which indicates the time of the last modification of
      an object.  
      The time is updated when any object header message changes according to
      the system clock where the change was posted.
    </P>

    <P class=item><B>Format of Data:</B>
    <div align=center>
      <table class=format>
	<caption>
	  Modification Time Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td colspan=3>Reserved (zero)</td>
	</tr>

	<tr>
	  <td colspan=4>Seconds After UNIX Epoch</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number is used for changes in the format of Object Modification Time
	      and is described here:</P>
	     <table class=list>
		<tr>
		  <th width="30%">Version</th>
		  <th align=left>Description</th>
		</tr>

		<tr>
		  <td align=center><code>0</code></td>
		  <td>Never used.</td>
		</tr>

		<tr>
		  <td align=center><code>1</code></td>
		  <td>Used by Version 1.6.1 and after of the library to encode time.  In
		    this version, the time is the seconds after Epoch.</td>
		</tr>
	      </table>
	  </td>
	</tr>

	<tr>
	  <td>Seconds After UNIX Epoch</td>
	  <td><P>A 32-bit unsigned integer value that stores the number of
            seconds since 0 hours, 0 minutes, 0 seconds, January 1, 1970,
            Coordinated Universal Time.</P></td>
	</tr>
      </table>
    </div>
    </P>

    <hr>
    <h4><a name="BtreeKValuesMessage">Name: B-tree 'K' Values Message</a></h4>

    <P class=item><B>Header Message Type:</B> 0x0013 </P>
    <P class=item><B>Length:</B> Fixed </P>
    <P class=item><B>Status:</B> Optional, may not be repeated. </P>

    <P class=item><B>Description:</B>This message retrieves non-default 'K'
        values for internal and leaf nodes of a group or indexed storage
        v1 B-trees.  This message is <em>only</em> found in the
        superblock extension.
    </P>

    <P class=item><B>Format of Data:</B>
    <div align=center>
      <table class=format>
	<caption>
	  B-tree 'K' Values Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td colspan=2>Indexed Storage Internal Node K</td>
          <td bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=2>Group Internal Node K</td>
	  <td colspan=2>Group Leaf Node K</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number for this message. This document describes
            version 0.</P>
	  </td>
	</tr>

	<tr>
	  <td>Indexed Storage Internal Node K</td>
	  <td><P>This is the node 'K' value for each internal node of an
                indexed storage v1 B-tree.  See the description of this field
                in version 0 and 1 of the superblock as well the section on
                v1 B-trees.
	      </P>
	  </td>
	</tr>

	<tr>
	  <td>Group Internal Node K</td>
	  <td><P>This is the node 'K' value for each internal node of a group 
                v1 B-tree.  See the description of this field in version 0 and
                1 of the superblock as well as the section on v1 B-trees.
	      </P>
	  </td>
	</tr>

	<tr>
	  <td>Group Leaf Node K</td>
	  <td><P>This is the node 'K' value for each leaf node of a group v1
                B-tree.  See the description of this field in version 0 and 1
                of the superblock as well as the section on v1 B-trees.
	      </P>
	  </td>
	</tr>

      </table>
    </div>
    </P>

    <hr>
    <h4><a name="DrvInfoMessage">Name: Driver Info Message</a></h4>

    <P class=item><B>Header Message Type:</B> 0x0014 </P>
    <P class=item><B>Length:</B> varies </P>
    <P class=item><B>Status:</B> Optional, may not be repeated. </P>

    <P class=item><B>Description:</B>This message contains information needed by the
		file driver to reopen a file 
		(see description in <A href="#DriverInfo">File Driver Info</A>).
		This message is <em>only</em> found in the superblock extension.
    </P>

    <P class=item><B>Format of Data:</B>
    <div align=center>
      <table class=format>
	<caption>
	  Driver Info Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>
	<tr>
	  <td colspan=4><br>Driver Identification</td>
	</tr>

	<tr>
	  <td colspan=2>Driver Information Size</td>
          <td colspan=2 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4><br><br>Driver Information <em>(variable size)</em><br><br><br></td>
	</tr>
	
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number for this message. This document describes
                version 0.</P>
	  </td>
	</tr>

	<tr>
	<td>Driver Identification</td>
	  <td><P>This is an eight-byte ASCII string without null termination which
                identifies the driver.
	      </P>
	  </td>
	</tr>

	<tr>
	  <td>Driver Information Size</td>
	  <td><P>The size in bytes of the <em>Driver Information</em> field of this
                message.</P>
	  </td>
	</tr>

	<tr>
	  <td>Driver Information</td>
	  <td><P>Driver information is stored in a format defined by the file driver.</P>
	  </td>
	</tr>
      </table>
    </div>
    </P>

    <hr>
    <h4><a name="AinfoMessage">Name: Attribute Info Message</a></h4>

    <P class=item><B>Header Message Type:</B> 0x0015</P>
    <P class=item><B>Length:</B> varies </P>
    <P class=item><B>Status:</B> Optional, may not be repeated. </P>

    <P class=item><B>Description:</B>This message stores information about the
        attributes on an object, such as the maximum creation index for
        the attributes created and the location of the attribute storage when
        the attributes are stored "densely".
    </P>

    <P class=item><B>Format of Data:</B>
    <div align=center>
      <table class=format>
	<caption>
	  Attribute Info Message
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td>Flags</td>
	  <td colspan=2>Maximum Creation Index <em>(optional)</em></td>
	</tr>
	<tr>
	  <td colspan=4><br>Fractal Heap Address<sup>O</sup><br><br></td>
	</tr>
	<tr>
	  <td colspan=4><br>Attribute Name v2 B-tree Address<sup>O</sup><br><br></td>
	</tr>
	<tr>
	  <td colspan=4><br>Attribute Creation Order v2 B-tree Address<sup>O</sup> <em>(optional)</em><br><br></td>
	</tr>
	
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number for this message. This document describes
                version 0.</P>
	  </td>
	</tr>

	<tr>
	<td>Flags</td>
	  <td><P>This is the attribute index information flag with the
                following definition:
	      </P>
            <table class=list>
                <tr>
                  <th width="30%">Bit</th>
                  <th align=left>Description</th>
                </tr>

                <tr>
                  <td align=center><code>0</code></td>
                  <td>If set, creation order for attributes is tracked.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>1</code></td>
                  <td>If set, creation order for attributes is indexed.
                  </td>
                </tr>
                <tr>
                  <td align=center><code>2-7</code></td>
                  <td>Reserved</td>
                </tr>
	    </table>
            </P>
          </td>
	</tr>

	<tr>
	  <td>Maximum Creation Index</td>
	  <td><P>The is the maximum creation order index value for the
                attributes on the object.</P>
	      <P>This field is present if bit 0 of <em>Flags</em> is set.</P>
	  </td>
	</tr>

	<tr>
	  <td>Fractal Heap Address</td>
	  <td><P>This is the address of the fractal heap to store dense
                attributes.</P>
	  </td>
	</tr>

	<tr>
	  <td>Attribute Name v2 B-tree Address</td>
	  <td><P>This is the address of the version 2 B-tree to index the
                names of densely stored attributes.</P>
	  </td>
	</tr>

	<tr>
	  <td>Attribute Creation Order v2 B-tree Address</td>
	  <td><P>This is the address of the version 2 B-tree to index the
                creation order of densely stored attributes.</P>
	      <P>This field is present if bit 1 of <em>Flags</em> is set.</P>
	  </td>
	</tr>

      </table>
    </div>
    </P>

    <hr>
    <h4><a name="RefCountMessage">Name: Object Reference Count</a></h4>

    <P class=item><B>Header Message Type:</B> 0x0016 </P>
    <P class=item><B>Length:</B> fixed </P>
    <P class=item><B>Status:</B> Optional, may not be repeated. </P>

    <P class=item><B>Description:</B>This message stores the number of hard
        links (in groups or objects) pointing to an object (i.e. its
        <em>reference count</em>).
    </P>

    <P class=item><B>Format of Data:</B>
    <div align=center>
      <table class=format>
	<caption>
	  Object Reference Count
	</caption>

	<tr>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	  <th width="25%">byte</th>
	</tr>

	<tr>
	  <td>Version</td>
          <td colspan=3 bgcolor=#DDDDDD><em>This space inserted only to align table nicely</em></td>
	</tr>

	<tr>
	  <td colspan=4>Reference count</td>
	</tr>
      </table>
    </div>

    <br>
    <div align=center>
      <table class=desc>
	<tr>
          <th width="30%">Field Name</th>
          <th>Description</th>
	</tr>

	<tr>
	  <td>Version</td>
	  <td><P>The version number for this message. This document describes
            version 0.</P>
	  </td>
	</tr>

	<tr>
	  <td>Reference Count</td>
	  <td><P>The unsigned 32-bit integer is the reference count for the
                object.  This message is only present in "version 2" (or later)
                object headers, and if not present those object header versions,
                the reference count for the object is assumed to be 1.
            </P>
	  </td>
	</tr>

      </table>
    </div>
    </P>

<hr>
<h3><a name="DataStorage">Disk Format: Level 2B - Data Object Data Storage</a></h3>

<P>The data for an object is stored separately from its header
    information in the file and may not actually be located in the HDF5 file
    itself if the header indicates that the data is stored externally.  The
    information for each record in the object is stored according to the
    dimensionality of the object (indicated in the dataspace header message).
    Multi-dimensional array data is stored in C order, i.e. the
    "last" dimension changes fastest.

<P>Data whose elements are composed of atomic datatypes are stored in IEEE
    format, unless they are specifically defined as being stored in a different
    machine format with the architecture-type information from the datatype
    header message.  This means that each architecture will need to [potentially]
    byte-swap data values into the internal representation for that particular
    machine.

<P> Data with a variable-length datatype is stored in the global heap
    of the HDF5 file.  Global heap identifiers are stored in the 
    data object storage.

<P>Data whose elements are composed of reference datatypes are stored in
    several different ways depending on the particular reference type involved.
    Object pointers are just stored as the offset of the object header being
    pointed to with the size of the pointer being the same number of bytes as
    offsets in the file.

<P>Dataset region references are stored as a heap-ID which points to the
    following information within the file-heap: an offset of the object pointed
    to, number-type information (same format as header message), dimensionality
    information (same format as header message), sub-set start and end
    information (i.e. a coordinate location for each), and field start and end
    names (i.e.  a [pointer to the] string indicating the first field included
    and a [pointer to the] string name for the last field).  

<P>Data of a compound datatype is stored as a contiguous stream of the items
    in the structure, with each item formatted according to its datatype.

<hr>
<h3><a name="Appendix">Appendix</a></h3>
<P>Definitions of various terms used in this document.
</P>

<P>The <A name="UndefinedAddress">"undefined address"</A> for a file is a
file address with all bits set, i.e. <code>0xffff...ff</code>.

<P>The <A name="UnlimitedDim">"unlimited size"</A> for a size is a
value with all bits set, i.e. <code>0xffff...ff</code>.

<!-- #BeginLibraryItem "/ed_libs/NavBar_ADevG.lbi" -->
<hr>
<center>
<table border=0 width=98%>
<tr><td valign=top align=left>
    <a href="index.html">HDF5 documents and links</a>&nbsp;<br>
    <a href="H5.intro.html">Introduction to HDF5</a>&nbsp;<br>
    <!--
    <a href="Glossary.html">Glossary</a><br>
    -->
</td>
<td valign=top align=right>
    <a href="UG/index.html">HDF5 User's Guide</a>&nbsp;<br>
    <a href="RM/RM_H5Front.html">HDF5 Reference Manual</a>&nbsp;<br>
    <a href="ADGuide.html">HDF5 Application Developer's Guide</a>&nbsp;<br>
</td></tr>
</table>
</center>
<hr>
<!-- #EndLibraryItem --><!-- #BeginLibraryItem "/ed_libs/Footer.lbi" -->
<address>
<table width="100%" border="0">
  <tr valign="top">
      <td align="left">
          <address>
          The HDF Group Help Desk: <img src="Graphics/help.png" align=top height=16>
          <br>
          Describes HDF5 Release 1.8.13, May 2014.
          </address>
      </td><td width="5%">&nbsp;</td>
      <td align="right">
          <a href="Copyright.html">Copyright</a> by
          <a href="http://www.hdfgroup.org">The HDF Group</a>
          <br>
          and the Board of Trustees of the University of Illinois
      </td>   
  </tr>   
</table>
</address>
<!-- #EndLibraryItem --><!-- hhmts start -->
Last modified: 12 January 2012
<!-- hhmts end -->

</body>
</html>