Building and Installing HTSlib
                    ==============================

Requirements
============

Building HTSlib requires a few programs and libraries to be present.
See the "System Specific Details" below for guidance on how to install
these.

At least the following are required:

    GNU make
    C compiler (e.g. gcc or clang)

In addition, building the configure script requires:

    autoheader
    autoconf
    autoreconf

Running the configure script uses awk, along with a number of
standard UNIX tools (cat, cp, grep, mv, rm, sed, among others).  Almost
all installations will have these already.

Running the test harness (make test) uses:

    bash
    perl

HTSlib uses the following external libraries.  Building requires both the
library itself, and include files needed to compile code that uses functions
from the library.  Note that some Linux distributions put include files in
a development ('-dev' or '-devel') package separate from the main library.

    zlib       (required)
    libbz2     (required, unless configured with --disable-bz2)
    liblzma    (required, unless configured with --disable-lzma)
    libcurl    (optional, but strongly recommended)
    libcrypto  (optional for Amazon S3 support; not needed on MacOS)
    libdeflate (optional, but strongly recommended for faster gzip)

Disabling libbzip2 and liblzma will make some CRAM files unreadable, so
is not recommended.

Using libcurl provides HTSlib with network protocol support, for
example it enables the use of ftp://, http://, and https:// URLs.
It is also required if direct access to Amazon S3 or Google Cloud
Storage is enabled.

Amazon S3 support requires an HMAC function to calculate a message
authentication code.  On MacOS, the CCHmac function from the standard
library is used.  Systems that do not have CCHmac will get this from
libcrypto.  libcrypto is part of OpenSSL or one of its derivatives (LibreSSL
or BoringSSL).

On Microsoft Windows we recommend use of Mingw64/Msys2.  Whilst the
code may work on Windows with other environments, these have not been
verified.  Use of the configure script is a requirement too.

Update htscodecs submodule
==========================

Note that this section only applies to git checkouts.  If you're building
from a release tar file, you can skip this section.

Some parts of HTSlib are provided by the external "htscodecs" project.  This
is included as a submodule.  When building from the git repository,
either clone the project using "git clone --recurse-submodules", or run:

    git submodule update --init --recursive

to ensure the correct version of the submodule is present.

It is also possible to link against an external libhtscodecs library
by using the '--with-external-htscodecs' configure option.  When
this is used, the submodule files will be ignored.

Building Configure
==================

This step is only needed if configure.ac has been changed, or if configure
does not exist (for example, when building from a git clone).  The
configure script and config.h.in can be built by running:

    autoreconf -i

Basic Installation
==================

To build and install HTSlib, 'cd' to the htslib-1.x directory containing
the package's source and type the following commands:

    ./configure
    make
    make install

The './configure' command checks your build environment and allows various
optional functionality to be enabled (see Configuration below).  If you
don't want to select any optional functionality, you may wish to omit
configure and just type 'make; make install' as for previous versions
of HTSlib.  However if the build fails you should run './configure' as
it can diagnose the common reasons for build failures.

The 'make' command builds the HTSlib library and various useful
utilities: bgzip, htsfile, and tabix.  If compilation fails you should
run './configure' as it can diagnose problems with your build environment
that cause build failures.

The 'make install' command installs the libraries, library header files,
utilities, several manual pages, and a pkgconfig file to /usr/local.
The installation location can be changed by configuring with --prefix=DIR
or via 'make prefix=DIR install' (see Installation Locations below).
Shared library permissions can be set via e.g. 'make install LIB_PERM=755'.


Configuration
=============

By default, './configure' examines your build environment, checking for
requirements such as the zlib development files, and arranges for a plain
HTSlib build.  The following configure options can be used to enable
various features and specify further optional external requirements:

--enable-plugins
    Use plugins to implement exotic file access protocols and other
    specialised facilities.  This enables such facilities to be developed
    and packaged outwith HTSlib, and somewhat isolates HTSlib-using programs
    from their library dependencies.  By default (or with --disable-plugins),
    any enabled pluggable facilities (such as libcurl file access) are built
    directly within HTSlib.

    Programs that are statically linked to a libhts.a with plugins enabled
    need to be linked using -rdynamic or a similar linker option.

    The <https://github.com/samtools/htslib-plugins> repository contains
    several additional plugins, including the iRODS (<http://irods.org/>)
    file access plugin previously distributed with HTSlib.

--with-plugin-dir=DIR
    Specifies the directory into which plugins built while building HTSlib
    should be installed; by default, LIBEXECDIR/htslib.

--with-plugin-path=DIR:DIR:DIR...
    Specifies the list of directories that HTSlib will search for plugins.
    By default, only the directory specified via --with-plugin-dir will be
    searched; you can use --with-plugin-path='DIR:$(plugindir):DIR' and so
    on to cause additional directories to be searched.

--with-external-htscodecs
    Build and link against an external copy of the htscodecs library
    instead of using the source files in the htscodecs directory.

--enable-libcurl
    Use libcurl (<http://curl.se/>) to implement network access to
    remote files via FTP, HTTP, HTTPS, etc.  By default or with
    --enable-libcurl=check, configure will probe for libcurl and include
    this functionality if libcurl is available.  Use --disable-libcurl
    to prevent this.

--enable-gcs
    Implement network access to Google Cloud Storage.  By default or with
    --enable-gcs=check, this is enabled when libcurl is enabled.

--enable-s3
    Implement network access to Amazon AWS S3.  By default or with
    --enable-s3=check, this is enabled when libcurl is enabled.

--disable-bz2
    Bzip2 is an optional compression codec format for CRAM, included
    in HTSlib by default.  It can be disabled with --disable-bz2, but
    be aware that not all CRAM files may be possible to decode.

--disable-lzma
    LZMA is an optional compression codec for CRAM, included in HTSlib
    by default.  It can be disabled with --disable-lzma, but be aware
    that not all CRAM files may be possible to decode.

--with-libdeflate
    Libdeflate is a heavily optimized library for DEFLATE-based compression
    and decompression.  It also includes a fast crc32 implementation.
    By default, ./configure will probe for libdeflate and use it if
    available.  To prevent this, use --without-libdeflate.

Each --enable-FEATURE/--disable-FEATURE/--with-PACKAGE/--without-PACKAGE
option listed also has an opposite, e.g., --without-external-htscodecs
or --disable-plugins.  However, apart from those options for which the
default is to probe for related facilities, using these opposite options
is mostly unnecessary as they just select the default configure behaviour.

The configure script also accepts the usual options and environment variables
for tuning installation locations and compilers: type './configure --help'
for details.  For example,

    ./configure CC=icc --prefix=/opt/icc-compiled

would specify that HTSlib is to be built with icc and installed into bin,
lib, etc subdirectories under /opt/icc-compiled.

If dependencies have been installed in non-standard locations (i.e. not on
the normal include and library search paths) then the CPPFLAGS and LDFLAGS
environment variables can be used to set the options needed to find them.
For example, NetBSD users may use:

    ./configure CPPFLAGS=-I/usr/pkg/include \
                LDFLAGS='-L/usr/pkg/lib -Wl,-R/usr/pkg/lib'

to allow compiling and linking against dependencies installed via the ports
collection.

Installation Locations
======================

By default, 'make install' installs HTSlib libraries under /usr/local/lib,
HTSlib header files under /usr/local/include, utility programs under
/usr/local/bin, etc.  (To be precise, the header files are installed within
a fixed 'htslib' subdirectory under the specified .../include location.)

You can specify a different location to install HTSlib by configuring
with --prefix=DIR or specify locations for particular parts of HTSlib by
configuring with --libdir=DIR and so on.  Type './configure --help' for
the full list of such install directory options.

Alternatively you can specify different locations at install time by
typing 'make prefix=DIR install' or 'make libdir=DIR install' and so on.
Consult the list of prefix/exec_prefix/etc variables near the top of the
Makefile for the full list of such variables that can be overridden.

You can also specify a staging area by typing 'make DESTDIR=DIR install',
possibly in conjunction with other --prefix or prefix=DIR settings.
For example,

    make DESTDIR=/tmp/staging prefix=/opt

would install into bin, lib, etc subdirectories under /tmp/staging/opt.


System Specific Details
=======================

Installing the prerequisites is system dependent and there is more
than one correct way of satisfying these, including downloading them
from source, compiling and installing them yourself.

For people with super-user access, we provide an example set of commands
below for installing the dependencies on a variety of operating system
distributions.  Note these are not specific recommendations on distribution,
compiler or SSL implementation.  It is assumed you already have the core set
of packages for the given distribution - the lists may be incomplete if
this is not the case.

Debian / Ubuntu
---------------

sudo apt-get update  # Ensure the package list is up to date
sudo apt-get install autoconf automake make gcc perl zlib1g-dev libbz2-dev liblzma-dev libcurl4-gnutls-dev libssl-dev libdeflate-dev

Note: libcurl4-openssl-dev can be used as an alternative to libcurl4-gnutls-dev.

RedHat / CentOS
---------------

sudo yum install autoconf automake make gcc perl-Data-Dumper zlib-devel bzip2 bzip2-devel xz-devel curl-devel openssl-devel libdeflate-devel

Note: On some versions perl FindBin will need to be installed to make the tests work.

sudo yum install perl-FindBin

Alpine Linux
------------

doas apk update  # Ensure the package list is up to date
doas apk add autoconf automake make gcc musl-dev perl bash zlib-dev bzip2-dev xz-dev curl-dev openssl-dev

Ideally also install a copy of libdeflate-dev for faster (de)compression.
This can be found in the Alpine community repository.

Note: some older Alpine versions use libressl-dev rather than openssl-dev.

OpenSUSE
--------

sudo zypper install autoconf automake make gcc perl zlib-devel libbz2-devel xz-devel libcurl-devel libopenssl-devel

Also install libdeflate-devel, available on OpenSUSE Leap 15.4 onwards
or directly via git releases above.

Windows MSYS2/MINGW64
---------------------

The configure script must be used as without it the compilation will
likely fail.

Follow MSYS2 installation instructions at
https://www.msys2.org/wiki/MSYS2-installation/

Then relaunch to MSYS2 shell using the "MSYS2 MinGW x64" executable.
Once in that environment (check $MSYSTEM equals "MINGW64") install the
compilers using pacman -S and the following package list:

base-devel mingw-w64-x86_64-toolchain
mingw-w64-x86_64-libdeflate mingw-w64-x86_64-zlib mingw-w64-x86_64-bzip2
mingw-w64-x86_64-xz mingw-w64-x86_64-curl mingw-w64-x86_64-autotools
mingw-w64-x86_64-tools-git

(The last is only needed for building libraries compatible with MSVC.)

HP-UX
-----

HP-UX requires that shared libraries have execute permission.  The
default for HTSlib is to install with permission 644 (read-write for
owner and read-only for group / other). This can be overridden by
setting the LIB_PERM variable at install time with:

    make install LIB_PERM=755