5.3. Structure of a Source Package

5.3.1. Format

A source package is usually comprised of three files, a .dsc, a .orig.tar.gz, and a .debian.tar.gz (or .diff.gz). They allow creation of binary packages (.deb files described above) from the source code files of the program, which are written in a programming language.

The .dsc (Debian Source Control) file is a short text file containing an RFC 2822 header (just like the control file studied in Section 5.2.1, “Description: the control File”) which describes the source package and indicates which other files are part thereof. It is signed by its maintainer, which guarantees authenticity. See Section 6.5, “Checking Package Authenticity” for further details on this subject.

Example 5.1. A .dsc file

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Format: 3.0 (quilt)
Source: zim
Binary: zim
Architecture: all
Version: 0.48-1
Maintainer: Emfox Zhou <emfox@debian.org>
Uploaders: Raphaël Hertzog <hertzog@debian.org>
Homepage: http://zim-wiki.org
Standards-Version: 3.9.0
Vcs-Browser: http://svn.debian.org/wsvn/collab-maint/deb-maint/zim/trunk?op=log
Vcs-Svn: svn://svn.debian.org/collab-maint/deb-maint/zim/trunk
Build-Depends: debhelper (>= 7.4.12), python-support (>= 0.8), xdg-utils, python (>= 2.5), libgtk2.0-0 (>= 2.6), python-gtk2, python-xdg, python-simplejson | python (>= 2.6)
Checksums-Sha1: 
 bd84fa5104de5ed85a49723d26b350856de93217 966899 zim_0.48.orig.tar.gz
 352111ff372a20579664416c9abd4970839835b3 9615 zim_0.48-1.debian.tar.gz
Checksums-Sha256: 
 77d8df7dc89b233fdc3aab1a8ad959c6888881ae160770f50bf880a56e02f895 966899 zim_0.48.orig.tar.gz
 0fceab5d3b099075cd38c225fa4002d893c1cdf4bbcc51d1391a34248e1e1a22 9615 zim_0.48-1.debian.tar.gz
Files: 
 88cfc18c0c7339528d5f5f463647bb5f 966899 zim_0.48.orig.tar.gz
 608b6e74aa14252dfc6236ab184bdb0c 9615 zim_0.48-1.debian.tar.gz

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Signed by Raphael Hertzog

iQEcBAEBCAAGBQJMSUAfAAoJEAOIHavrwpq5qjUIAKmM8p86GcHYTxMmKENoBUoW
UPi5R7DzrLMbFrUXKgXWLvEKQTXpmkJhh2aSWq2iY+5piBSHwMiITfaBTpdTRvzU
5nT/n9MlF8sJFESet/NgZaMPFDzWUbIy5aYbuG1TXmn/7XiDrBaQGiVqKkVLPrqc
yWhsotn3JNKIjbPDW/DjImYyKD5RZpXrbVjuIgDT1E6yxtNYwUyBlK0cx/GITNep
uV48hsT8cj0paqVXl5+P9Ww8XIE3clxNpE/45/tvKvkqGOeysc6OPAqsIw6HYFY9
0EnvMTfMpeQOA68ZqsNpUjomv5r/EGwdCbAWo5iJDsZzXQ1Feh6iSNrjv3yeRzg=
=qnbh
-----END PGP SIGNATURE-----

Note that the source package also has dependencies (Build-Depends) completely distinct from those of binary packages, since they indicate tools required to compile the software in question and construct its binary package.

CULTURE Why divide into several packages

Quite frequently, a source package (for a given software) can generate several binary packages. The split is justified by the possibility to use (parts of) the software in different contexts. Consider a shared library, it may be installed to make an application work (for example, libc6), or it can be installed to develop a new program (libc6-dev will then be the correct package). We find the same logic for client/server services where we want to install the server part on one machine and the client part on others (this is the case, for example, of openssh-server and openssh-client).

Just as frequently, the documentation is provided in a dedicated package: the user may install it independently from the software, and may at any time choose to remove it to save disk space. Additionally, this also saves disk space on the Debian mirrors, since the documentation package will be shared amongst all of the architectures (instead of having the documentation duplicated in the packages for each architecture).

PERSPECTIVE Different source package formats

Originally there was only one source package format. This is the 1.0 format, which associates an .orig.tar.gz archive to a .diff.gz “debianization” patch (there is also a variant, consisting of a single .tar.gz archive, which is automatically used if no .orig.tar.gz is available).

Since Debian Squeeze, Debian developers have the option to use new formats that correct many problems of the historical format. Format 3.0 (quilt) can combine multiple upstream archives in the same source package: in addition to the usual .orig.tar.gz, supplementary .orig-component.tar.gz archives can be included. This is useful with software that are distributed in several upstream components but for which a single source package is desired. These archives can also be compressed with bzip2 or xz rather than gzip (lzma is also supported by dpkg-source but not accepted into the official archive), which saves disk space and network resources. Finally, the monolithic patch, .diff.gz is replaced by a .debian.tar.gz archive containing the compiling instructions and a set of upstream patches contributed by the package maintainer. These last are recorded in a format compatible with quilt — a tool that facilitates the management of a series of patches.

The .orig.tar.gz file is an archive containing the source code as provided by the original developer. Debian package maintainers are asked to not modify this archive in order to be able to easily check the origin and integrity of the file (by simple comparison with a checksum) and to respect the wishes of some authors.

The .debian.tar.gz contains all of the modifications made by the Debian maintainer, especially the addition of a debian directory containing the instructions to execute to construct a Debian package.

TOOL Decompressing a source package

If you have a source package, you can use the dpkg-source command (from the dpkg-dev package) to decompress it:

$ dpkg-source -x package_0.7-1.dsc

You can also use apt-get to download a source package and unpack it right away. It requires that the appropriate deb-src lines be present in the /etc/apt/sources.list file, however (for further details, see Section 6.1, “Filling in the sources.list File”). These are used to list the “sources” of source packages (meaning the servers on which a group of source packages are hosted).

$ apt-get source package

5.3.2. Usage within Debian

The source package is the foundation of everything in Debian. All Debian packages come from a source package, and each modification in a Debian package is the consequence of a modification made to the source package. The Debian maintainers work with the source package, knowing, however, the consequences of their actions on the binary packages. The fruits of their labors are thus found in the source packages available from Debian: you can easily go back to them and everything stems from them.

When a new version of a package (source package and one or more binary packages) arrives on the Debian server, the source package is the most important. Indeed, it will then be used by a network of machines of different architectures for compilation on the various architectures supported by Debian. The fact that the developer also sends one or more binary packages for a given architecture (usually i386 or amd64) is relatively unimportant, since these could just as well have been automatically generated.