ChangeLog for SDoP
------------------

Version 0.90 (20-August-2020)
-----------------------------

1. If any of the characters 0x82, 0x83, 0xA0, or 0xAD were used with a
non-standardly encoded font, their width was incorrectly taken as zero when
calculating a string width. This could lead to alignment errors. (In a
standarly encoded font these Unicode characters don't always appear in the AFM,
but do have zero width.)

2. When searching for an AFM file, SDoP used to search for "fontmetrics/name"
within the installed /share/sdop or -S directories. Now, if that search fails,
it also searches for "name" without the "fontmetrics" prefix. This means that
-S can list the actual directory where AFM files are, without them having to be
in a subdirectory called "fontmetrics".

3. Minor code tidies to avoid compiler warnings.


Version 0.82 (20-October-2018)
------------------------------

 1. Fix typo in src/Makefile (LFLAGS should be LDFLAGS).

 2. Fix typo in error message.


Version 0.81 (29-August-2018)
-----------------------------

 1. Code tidies to get rid of compiler and sanitizer warnings.

 2. Malformed AFM files could cause crashes.


Version 0.80 (27-November-2013)
-------------------------------

 1. If index_headings_pdf_toc (a local processing parameter) is "yes",
    "pdfmark" data is inserted so that sections in an index (A, B, C, etc)
    appear in the PDF table of contents. The default is not to do this. Index
    sections never appear in the printed TOC.

 2. Removed three variables that were being set but never used.

 3. Updated the code to support characters in the Unicode Latin Extended-A code
    page (U+0100 to U+017F) which are not in the original PostScript fonts, but
    which can be found in newer versions and alternative fonts. Also updated
    the fontmetrics files for these characters, though not all alternative
    fonts support all the characters.

 4. Recognize --version.

 5. When table_warn_overflow="overprint" is specified, an actual overprint is
    not detected until the column that is overprinted is reached. The warning
    message was reporting the later column instead of the one that actually
    contained the overlong text.

 6. Added a facility to specify more fully the fonts for itemized and ordered
    list tags. Previously only roman/italic/bold/bolditalic could be specified;
    now the font's size and colour can be set and changed in different parts of
    the document. As part of this work, the names of some of the processing
    parameters have been changed to be more meaningful; the old names are
    retained, but cause a "deprecated" warning message.

 7. The PDF documentation was being created without embedded fonts, which meant
    that acroread did not display some characters in the tables at the end
    because they are not in the default fonts it uses. The fonts are now all
    embedded in the PDF.

 8. "Exotic" fonts were not being scaled by scale_typesize_base.


Version 0.71 (18-May-2012)
--------------------------

 1. There was no check for too many nested lists (maximum 10), leading to a
    segfault. Now there is a hard error.


Version 0.70 (26-Jan-2011)
--------------------------

 1. An off-by-one bug has been fixed (detected by valgrind in test 45).

 2. If the string specified for index_omit_sort contained any entities,
    references to uninitialized memory could occurn (detected by valgrind
    in test 48).

 3. A <literallayout> "paragraph" may contain blank lines. When such a
    paragraph had to be split at the foot of a page, and the first part
    consisted only of blank lines, SDoP crashed with a segfault.

 4. A long <literallayout> "paragraph" that continued over more than one page
    break caused SDoP to fail with the error "internal error: paragraph split
    end failure".

 5. Applied some patches from the Debian and Ubuntu distributors:

    (a) Add support for "make install DESTDIR=..."
    (b) Fixed a bug in make test that happened when sdop was not previously
        installed.
    (c) Make the testsuite die on error when not running interactively.
    (d) Fix typo LFLAGS (should read LDFLAGS) in src/Makefile.
    (e) Reorder linker command for the sdop binary, moving $(LIBS) to the end
        of the line so it gets added after the object files. This seems to be
        necessary for linking with --as-needed.

 6. If the text of an itemized list item was not in the roman font, the mark
    for the item was output in that font instead of roman.

 7. Added ilist_font and olist_font to specify the style (italic, roman, etc)
    of the marks for these lists. The size and family are those of the main
    text or footnote text, as appropriate.

 8. The colour of the footnote number in all but the first footnote on a page
    was the same as the last character in the previous footnote, instead of
    always being black.

 9. If a bullet character specified for an itemized list was in one of the
    auxiliary fonts (e.g. a character such as U+2666, a diamond that is in the
    ZapfDingbats font), the error "internal error: special auxiliary font
    missing" occurred.

10. Fixed an internal error in a data chain, which doesn't actually seem to
    have affected anything other than the debugging output.

11. Fixed another failed to initialize error which didn't show up until I was
    adding a new feature.

12. After certain errors (as well as after warnings), SDoP continues
    processing, and may even generate output. For some errors  (in particular
    missing cross references) it now suppresses the output and says so. For
    other errors, it now warns that the output may be incorrect.

13. The "hard space" character (U+00A0) does not appear in the AFM files for
    the standard fonts. Consequently, its width was being calculated as zero,
    thereby potentially affecting some spacings. It is actually output as a
    space character, so now its width is set equal to the width of a space.

14. Cross-references in the text (<xref>s) and page numbers in indexes now
    cause additional "pdfmark" data to be generated in the PostScript output.
    If this is converted to a PDF, the references and page numbers become
    clickable links. This feature can be disabled by setting the global
    processing parameter <?sdop xref_links="no"?>. By default, the text of the
    references is the colour of the surrounding text, and the text of page
    numbers in indexes is black. However, the xref_rgb parameter can be used to
    specify a different colour for them.

14. The name of the parameter orderedlist_format is inconsistent with other
    list parameters, which use olist_, ilist_, etc. Changed it to olist_format,
    but kept the old name as a synonym for compatibility.


Version 0.63
------------

 1. The latest gcc gives compiler warnings for discarded function return values
    when FORTIFY_SOURCE is set, even when the value is explicitly cast to
    (void). Alternative code has been used to avoid such warnings.

 2. A couple of "variable may be used unset" warnings have been fixed.

 3. The way in which the PNG library was being accessed was deprecated. It has
    been changed to use the "proper" current interface.

 4. The way JPEG files are decoded sometimes changes slightly between releases;
    this caused the tests to fail with libjpeg 7 because they were originally
    created with libjpeg 6. The JPEG test files have been adjusted by making
    them with cjpeg -sample 1x1 which is supposed to make them be decoded the
    same by the different libjpeg libraries.


Version 0.62
------------

 1. If more than one SDoP processing instruction preceded a <preface>, only the
    first of them was noticed when processing the preface.


Version 0.61
------------

 1. Added options to ./configure to force disabling of JPEG and PNG support.

 2. Some #ifdefs for JPEG and PNG support were missing.

 3. Arrange to skip JPEG and PNG tests in "make test" when the appropriate
    support is not compiled.


Version 0.60
------------

 1. Added support for subscript_small and superscript_small, both defaulting to
    "yes". At present, the small font is the footnote font. These parameters
    are ignored in titles.

 2. Added support for subscript_down and superscript_up, both defaulting to 33.

 3. Do not check for unsupported processing instructions unless output is
    generated, because when SDoP stops early after errors, not all of them will
    have been looked at.

 4. Added some words to the hyphenation list.

 5. If two index entries for the same item both had processing parameters in
    their content, the second gave rise to incorrect "unrecognized SDoP
    processing parameter" errors, because it was never actually processed.

 6. Index entries that differed only in font sizes, colours, or in having some
    characters subscripted or superscripted were not being treated as
    different.

 7. Allow spaces after commas in settings such as "yes, yes, no".

 8. Impose a minimum page length of 108 points (1.5").

 9. Added support for PNG files, but I have only tested a couple, so it may
    be incomplete.

10. Changed the fontmetrics files in the distribution to a set obtained with
    clear licensing conditions, and with appropriate licence files. The
    replacement files have more kerning data, which might be helpful. However,
    they lacked entries for the Euro character (which is in the fonts). I added
    Euro entries to the Helvetica, Courier, and Times fonts, copying from the
    previous AFM files, and noting in a comment that I had done so.

11. The name of the AFM file for the AvantGarde italic font was incorrect in
    the code ("Oblique" instead of "BookOblique").

12. The numeration in an ordered list item took on the font characteristics
    (size, italic, etc.) of the first character, which is clearly silly. Now
    the numeration uses the default font and size for the main text.

13. SDoP was reading all characters between & and ; when processing entities.
    Now it accepts only letters, digits, underscore, colon, hyphen, and dot.

14. Added &apos; and &quot; to the tables, as these are supposed to be
    recognized by all XML processors.

15. Added recognition of SDOP_SHARE (can be overridden by -S).

16. Added support for BPH (U+0082, "break permitted here") and NBH (U+0083, "no
    break here").

17. When SDoP's PostScript output was converted to a PDF, the index pointers
    for any prefaces were not working. I also added a PDF index item for the
    title page, when present.

18. Stop SDoP from crashing if an entity is unexpectedly its own partner,
    for example, <index/>.


Version 0.52
------------

 1. Because inserted files sometimes have bad bounding box values, and also
    for general aesthetic reasons, I've added a feature to <imageobject> that
    allows the image to be moved on the page. To do this, the "role" parameter
    has been abused, so as not to break the DocBook DTD.

 2. Added table_indent as a global processing instruction.

 3. The character U+2027 (a centred dot) was not being recognized; it
    corresponds to the PostScript character "periodcentered".

 4. The character U+0145, though correctly handled, was generating a spurious
    "unrecognized character" error.

 5. Added support for align="char" to <tgroup> and <colspec>. The defaults for
    "char" and "charoff" are "." and "50" respectively. If the given character
    is not found in a cell's text, SDoP behaves as if it were present at the
    end of the string. If "char" alignment would cause overflow or underflow,
    an error is generated and right or left alignment (respectively) is used
    instead.

 6. Added support for "align", "char", and "charoff" to <entry>.

 7. SDoP borrows some code from PCRE for checking whether a Unicode character
    is a letter. This is used when hyphenating. The way PCRE does this has been
    updated to be more efficient; the revised code has now been installed in
    SDoP. Efficiency isn't the only point: the main table is auto-generated
    from the Unicode tables for PCRE. The old way of generating is no longer
    maintained, so it's best to move to the new scheme.

 8. Added support for <?sdop extra_leading="n"?>.

 9. When a non-standardly encoded font (e.g. the PMW-Music font) was used,
    characters that are special in standard encoding (e.g. 160, the "hard
    space") were being treated specially, which should not be the case. All
    characters in such fonts are now just treated as data characters.

10. If a non-standardly encoded font did not contain a width for the '-'
    character, SDoP could crash.

11. If a non-standardly encoded font did not contain the character used for
    substituting for unknown characters, a hard internal error was given. Now,
    instead of trying to use the normal unknown character substitution, the
    first defined character in the special font is used, and a warning is
    output to say so.

12. SDoP could loop if a <literallayout> section was sufficiently long that it
    continued over two page breaks, and there was a forced page break at its
    end.

13. Added support for <?sdop index_sort_omit="...."?>.

14. Added support for <?sdop toc_printed_sections=".."?>, which limits the
    contents of the printed TOC, as opposed to the PDF bookmarks TOC. Setting
    toc_sections also sets toc_printed_sections.

15. Vertical stretching was not being applied to space specified before items;
    it was only being applied to the "after" spacing.

16. Allowed a little vertical stretching in indexes.

17. Ensure that at least one secondary index line appears in the same column as
    its primary - that is, avoid having a primary last in a column if it is
    followed by one or more secondaries.

18. When a list of secondary index entries flows over to a new column, arrange
    for a "(continued)" primary item to head the column.

19. The default column separation for indexes was not quite right.


Version 0.51
------------

 1. When a font contained a character with a negative width (the PMW-Music font
    has such characters), SDoP was generating a spurious "unknown character"
    error.

 2. An attempt to specify a single font (e.g. a section title) as bolditalic
    failed.

 3. Previously, all section titles, however deeply nested, used the same title
    font. Now the first uses the "section font" and any nested ones use the
    "subsection" font, which defaults to 11 point.

 4. When the table column overflow option was set to "overprint", spurious
    warnings could be given when an overflowing entry was the last one
    specified in a row. The code now also checks for overflowing into blank
    columns (previously overflowing into a blank column could trigger the
    warning if the justification was appropriate).


Version 0.50
------------

 1. Added support (or at least recognition) for these elements from Simplified
    DocBook:

      <abbrev>             No new processing is actually involved.
      <abstract>           No new processing is actually involved.
      <acronym>            No new processing is actually involved.
      <article>
      <articleinfo>
      <attribution>
      <audiodata>          Completely ignored
      <audioobject>        Completely ignored
      <authorblurb>        Completely ignored
      <caption>
      <citetitle>          Treated as <emphasis>
      <computeroutput>     Treated as <literal>
      <corpauthor>         Sets &author_corpauthor;
      <edition>            Sets &book_edition;
      <editor>             Sets various &editor_xxx; entities
      <email>              Italic by default
      <epigraph>           Works like <blockquote> but no title
      <example>            Parameters similar to <figure> and <table>
      <footnoteref>        But only when on the same page as the footnote
      <inlinemediaobject>  Treated as a synonym for <mediaobject>
      <issuenum>           Sets &book_issuenum;
      <jobtitle>
      <keyword>            Completely ignored
      <keywordset>         Completely ignored
      <legalnotice>        Use <?sdop insert="legalnotice"?> to insert
      <lineannotation>     Uses a small italic font.
      <link>               Completely ignored
      <note>               Works like <blockquote>, but no <attribution>
      <objectinfo>         Completely ignored
      <othercredit>        Sets various &othercredit_xxx; entities
      <pubdate>            Sets &book_pubdate;
      <publishername>      Sets &book_publishername;
      <releaseinfo>        Sets &book_releaseinfo;
      <replaceable>        Italic by default
      <revdescription>     Use <?sdop insert="revdescription"?> to insert
      <revremark>          Completely ignored
      <sectioninfo>        Completely ignored
      <sidebar>            Works like <blockquote>, but no <attribution>
      <subject>            Completely ignored
      <subjectset>         Completely ignored
      <subjectterm>        Completely ignored
      <systemitem>         Completely ignored
      <trademark>          Completely ignored
      <userinput>          Monospaced by default
      <videodata>          Completely ignored
      <videoobject>        Completely ignored
      <volumenum>          Sets &book_volumenum;

    Note that &book_revnumber; gets the value of the last <revnumber> to be
    encountered if there is more than one.

 2. Added support for <subtitle> on chapters (previously it was only acted on
    for book titles), which makes it also work for articles. Added
    font_chapter_subtitle to vary the font used.

 3. Nobody had noticed that the PostScript output contained comments like
    "%%Page 3 5" rather than the correct form "%%Page: 3 5". Fixed.

 4. Added support for <emphasis role="roman"> to make it possible to revert to
    roman type in the middle of non-roman (which might have been turned on by
    some other element, e.g. <citetitle>).

 5. Added support for <honorific>, <lineage>, and <othername> inside <author>,
    though all it does is set up special entities. They are not used in the
    default title templates.

 6. Added support for <emphasis role="smallfont"> to implement a smaller font
    that can be used for anything, but in particular, for "legalnotice" type
    stuff on title verso pages. The small fonts are also used for
    <lineannotation>s.

 7. Added support for <title> in <blockquote> and processing instructions
    font_blockquote_title and blockquote_title_justify.

 8. Added <?sdop blockquote_ruled="yes"?> to put horizontal rules above and
    below block quotes.


Version 0.06
------------

 1. If a chapter or section title contained a special entity, and head or foot
    lines were set up to contain the chapter or section title, SDoP did not
    recognize the entities in the head/foot lines, and gave an "unrecognized"
    error.

 2. If an automatically inserted hyphen followed an "fi" ligature, or any other
    character whose code point is greater than 255, the hyphen character was
    incorrectly output.

 3. Most Unicode characters greater than 255 in chapter or section titles are
    converted to '?' in the PDFMARK data that is included in the PostScript so
    that a PDF table of contents can be created. Some new replacements have
    been added: opening and closing quotes are converted to ASCII versions, and
    en and em dashes are converted to hyphens.


Version 0.05
------------

 1. Treat a paragraph consisting of just a single zero-width space character
    (&#x200b;) specially by removing the line of data. This causes the vertical
    white space to be a bit less than the case when an actually blank line is
    output. The latter can be caused, for example, by using a hard space
    (&nbsp;).


Version 0.04
------------

 1. When a title was followed by a mediaobject (image) and the pair of them
    could not fit on a page, sdop looped because it always tries to keep a
    title together with what follows. Now it outputs the title on its own; if
    the media object itself won't fit, that will be caught elsewhere.

 2. A chapter without a title that contained only a mediaobject and no other
    text was not forcing a new page.

 3. Added auth-enti-cation to the hyphenation dictionary.

 4. Add --help to hytest and buildhy.

 5. When a background color was set for a page, the PostScript path was not
    being deleted; if another drawing was on the page, it got messed up.

 6. The distribution was missing the *-t and *-p data files, and the Makefile
    wasn't installing them.

 7. The typo "while isspace(*p)" wasn't noticed because the code was tested
    on a system where isspace() is a macro that ends up in parentheses. The
    missing parens have been added.


Version 0.03
------------

 1. Added page_full_length, page_line_width, page_foot_line_width, and
    page_head_line_width.

 2. Added many processing instructions for changing details of specific fonts:
    scale_typesize_base and lots of font_xxx...

 3. Missing substrings for toc_line_chapter_strings were not being set empty.

 4. Added paper_size and margin_{bottom,left_recto,left_verso} processing
    instructions.

 5. Re-jigged the font handling to avoid binding different virtual font numbers
    with the same characteristics.

 6. Added background_rgb to set background colour.

 7. Fixed a bug with the indentation of <term>s in <variablelist>s. By default
    they would end up 2 points to the left the main text (not always
    noticeable). Invented a separate set of layout parameters for the titles of
    <variablelists>s.

 8. Added many processing instructions for changing the indentation of many
    different types of paragraph.


Version 0.02
------------

 1. Added -p and -pf to select specific output pages.

 2. Added processing instruction chapter_head_skip="no" to skip head for a
    chapter starting page.

 3. Added a whole lot more character entity definitions.

 4. Added processing instructions to change the three font familes.

 5. Added index_headings processing instruction.

 6. Added support for <figure>.

 7. Added support for <title> in <table>s.

 8. Added support for <sectn>, but always as a synonym for <section>.

 9. Added support for <programlisting> as a synonym for <screen>.

10. Added support for <![CDATA[...]]> sections.

11. Added support for <subscript> and <superscript>.

12. If an unsupported character's code was less than 384, no warning message
    was given, and the substitute character's width was incorrect.

13. Added -qc, -wc, and processing parameter warn_unsupported_characters.

14. SDoP wasn't properly skipping processing instructions for other processors
    if they extended over more than one line.


Version 0.01
------------

 1. When a line that ended with </literal> was joined onto the next line, the
    inserted space was in the monospaced font and not in the ordinary font,
    thus often looking far too wide.

 2. Tidy up error messages for files that fail to open.

 3. Do not output a table of contents if there is nothing to go in it.

 4. A hyphen at the start of a line or table cell was being taken as a possible
    hyphenation point by mistake.

 5. Added support for one or more <preface> elements.

 6. The toc_sections processing instruction is now local rather than global, so
    it can be changed while processing. In parcticular, a different setting can
    be used within a preface.

 7. Added separate header and footer templates for the preface and the TOC, and
    processing instructions to vary the left/centre/right variables. Page
    numbers in the TOC are now in italic roman numerals by default.

 8. Added support for appendices and colophons.

 9. Added support for footnotes.

10. Added support for <mediaobject> with <textobject> and <imageobject> in
    EPS or JPEG formats.

11. Some refactoring to make the handling of newlines within the markup more
    straightforward.

12. Added support for exotic fonts via

      <emphasis role="exotic-font-name/size/leading">

    If size and leading are omitted, the values from the current font are used.
    Size can be given as 0 to mean "the current size" if just additional
    leading is needed. Leading space happens above the line. Fonts with
    standard encoding are re-encoded as Unicode; for others, only characters in
    the range 0-255 are supported, in their native encoding.

** End **