3
5
Fork 0
mirror of git://git.savannah.gnu.org/guix.git synced 2023-12-14 03:33:07 +01:00

doc: Move "System Configuration" higher.

* doc/guix.texi (GNU Distribution): Move "System Configuration" right
  after "System Installation".
This commit is contained in:
Ludovic Courtès 2014-07-16 11:35:45 +02:00
parent 79ad1c6999
commit cf4a912919

View file

@ -2604,12 +2604,12 @@ For information on porting to other architectures or kernels,
@menu
* System Installation:: Installing the whole operating system.
* System Configuration:: Configuring a GNU system.
* Installing Debugging Files:: Feeding the debugger.
* Package Modules:: Packages from the programmer's viewpoint.
* Packaging Guidelines:: Growing the distribution.
* Bootstrapping:: GNU/Linux built from scratch.
* Porting:: Targeting another platform or kernel.
* System Configuration:: Configuring a GNU system.
@end menu
Building this distribution is a cooperative effort, and you are invited
@ -2781,473 +2781,6 @@ guix system disk-image --image-size=800MiB gnu/system/install.scm
@file{gnu/system/install.scm} in the source tree for more information
about the installation image.
@node Installing Debugging Files
@section Installing Debugging Files
@cindex debugging files
Program binaries, as produced by the GCC compilers for instance, are
typically written in the ELF format, with a section containing
@dfn{debugging information}. Debugging information is what allows the
debugger, GDB, to map binary code to source code; it is required to
debug a compiled program in good conditions.
The problem with debugging information is that is takes up a fair amount
of disk space. For example, debugging information for the GNU C Library
weighs in at more than 60 MiB. Thus, as a user, keeping all the
debugging info of all the installed programs is usually not an option.
Yet, space savings should not come at the cost of an impediment to
debugging---especially in the GNU system, which should make it easier
for users to exert their computing freedom (@pxref{GNU Distribution}).
Thankfully, the GNU Binary Utilities (Binutils) and GDB provide a
mechanism that allows users to get the best of both worlds: debugging
information can be stripped from the binaries and stored in separate
files. GDB is then able to load debugging information from those files,
when they are available (@pxref{Separate Debug Files,,, gdb, Debugging
with GDB}).
The GNU distribution takes advantage of this by storing debugging
information in the @code{lib/debug} sub-directory of a separate package
output unimaginatively called @code{debug} (@pxref{Packages with
Multiple Outputs}). Users can choose to install the @code{debug} output
of a package when they need it. For instance, the following command
installs the debugging information for the GNU C Library and for GNU
Guile:
@example
guix package -i glibc:debug guile:debug
@end example
GDB must then be told to look for debug files in the user's profile, by
setting the @code{debug-file-directory} variable (consider setting it
from the @file{~/.gdbinit} file, @pxref{Startup,,, gdb, Debugging with
GDB}):
@example
(gdb) set debug-file-directory ~/.guix-profile/lib/debug
@end example
From there on, GDB will pick up debugging information from the
@code{.debug} files under @file{~/.guix-profile/lib/debug}.
In addition, you will most likely want GDB to be able to show the source
code being debugged. To do that, you will have to unpack the source
code of the package of interest (obtained with @code{guix build
--source}, @pxref{Invoking guix build}), and to point GDB to that source
directory using the @code{directory} command (@pxref{Source Path,
@code{directory},, gdb, Debugging with GDB}).
@c XXX: keep me up-to-date
The @code{debug} output mechanism in Guix is implemented by the
@code{gnu-build-system} (@pxref{Build Systems}). Currently, it is
opt-in---debugging information is available only for those packages
whose definition explicitly declares a @code{debug} output. This may be
changed to opt-out in the future, if our build farm servers can handle
the load. To check whether a package has a @code{debug} output, use
@command{guix package --list-available} (@pxref{Invoking guix package}).
@node Package Modules
@section Package Modules
From a programming viewpoint, the package definitions of the
GNU distribution are provided by Guile modules in the @code{(gnu packages
@dots{})} name space@footnote{Note that packages under the @code{(gnu
packages @dots{})} module name space are not necessarily ``GNU
packages''. This module naming scheme follows the usual Guile module
naming convention: @code{gnu} means that these modules are distributed
as part of the GNU system, and @code{packages} identifies modules that
define packages.} (@pxref{Modules, Guile modules,, guile, GNU Guile
Reference Manual}). For instance, the @code{(gnu packages emacs)}
module exports a variable named @code{emacs}, which is bound to a
@code{<package>} object (@pxref{Defining Packages}).
The @code{(gnu packages @dots{})} module name space is special: it is
automatically scanned for packages by the command-line tools. For
instance, when running @code{guix package -i emacs}, all the @code{(gnu
packages @dots{})} modules are scanned until one that exports a package
object whose name is @code{emacs} is found. This package search
facility is implemented in the @code{(gnu packages)} module.
Users can store package definitions in modules with different
names---e.g., @code{(my-packages emacs)}. In that case, commands such
as @command{guix package} and @command{guix build} have to be used with
the @code{-e} option so that they know where to find the package.
The distribution is fully @dfn{bootstrapped} and @dfn{self-contained}:
each package is built based solely on other packages in the
distribution. The root of this dependency graph is a small set of
@dfn{bootstrap binaries}, provided by the @code{(gnu packages
bootstrap)} module. For more information on bootstrapping,
@ref{Bootstrapping}.
@node Packaging Guidelines
@section Packaging Guidelines
The GNU distribution is nascent and may well lack some of your favorite
packages. This section describes how you can help make the distribution
grow. @xref{Contributing}, for additional information on how you can
help.
Free software packages are usually distributed in the form of
@dfn{source code tarballs}---typically @file{tar.gz} files that contain
all the source files. Adding a package to the distribution means
essentially two things: adding a @dfn{recipe} that describes how to
build the package, including a list of other packages required to build
it, and adding @dfn{package meta-data} along with that recipe, such as a
description and licensing information.
In Guix all this information is embodied in @dfn{package definitions}.
Package definitions provide a high-level view of the package. They are
written using the syntax of the Scheme programming language; in fact,
for each package we define a variable bound to the package definition,
and export that variable from a module (@pxref{Package Modules}).
However, in-depth Scheme knowledge is @emph{not} a prerequisite for
creating packages. For more information on package definitions,
@ref{Defining Packages}.
Once a package definition is in place, stored in a file in the Guix
source tree, it can be tested using the @command{guix build} command
(@pxref{Invoking guix build}). For example, assuming the new package is
called @code{gnew}, you may run this command from the Guix build tree:
@example
./pre-inst-env guix build gnew --keep-failed
@end example
Using @code{--keep-failed} makes it easier to debug build failures since
it provides access to the failed build tree. Another useful
command-line option when debugging is @code{--log-file}, to access the
build log.
If the package is unknown to the @command{guix} command, it may be that
the source file contains a syntax error, or lacks a @code{define-public}
clause to export the package variable. To figure it out, you may load
the module from Guile to get more information about the actual error:
@example
./pre-inst-env guile -c '(use-modules (gnu packages gnew))'
@end example
Once your package builds correctly, please send us a patch
(@pxref{Contributing}). Well, if you need help, we will be happy to
help you too. Once the patch is committed in the Guix repository, the
new package automatically gets built on the supported platforms by
@url{http://hydra.gnu.org/gnu/master, our continuous integration
system}.
@cindex substituter
Users can obtain the new package definition simply by running
@command{guix pull} (@pxref{Invoking guix pull}). When
@code{hydra.gnu.org} is done building the package, installing the
package automatically downloads binaries from there
(@pxref{Substitutes}). The only place where human intervention is
needed is to review and apply the patch.
@menu
* Software Freedom:: What may go into the distribution.
* Package Naming:: What's in a name?
* Version Numbers:: When the name is not enough.
* Python Modules:: Taming the snake.
* Perl Modules:: Little pearls.
@end menu
@node Software Freedom
@subsection Software Freedom
@c Adapted from http://www.gnu.org/philosophy/philosophy.html.
The GNU operating system has been developed so that users can have
freedom in their computing. GNU is @dfn{free software}, meaning that
users have the @url{http://www.gnu.org/philosophy/free-sw.html,four
essential freedoms}: to run the program, to study and change the program
in source code form, to redistribute exact copies, and to distribute
modified versions. Packages found in the GNU distribution provide only
software that conveys these four freedoms.
In addition, the GNU distribution follow the
@url{http://www.gnu.org/distros/free-system-distribution-guidelines.html,free
software distribution guidelines}. Among other things, these guidelines
reject non-free firmware, recommendations of non-free software, and
discuss ways to deal with trademarks and patents.
Some packages contain a small and optional subset that violates the
above guidelines, for instance because this subset is itself non-free
code. When that happens, the offending items are removed with
appropriate patches or code snippets in the package definition's
@code{origin} form (@pxref{Defining Packages}). That way, @code{guix
build --source} returns the ``freed'' source rather than the unmodified
upstream source.
@node Package Naming
@subsection Package Naming
A package has actually two names associated with it:
First, there is the name of the @emph{Scheme variable}, the one following
@code{define-public}. By this name, the package can be made known in the
Scheme code, for instance as input to another package. Second, there is
the string in the @code{name} field of a package definition. This name
is used by package management commands such as
@command{guix package} and @command{guix build}.
Both are usually the same and correspond to the lowercase conversion of
the project name chosen upstream, with underscores replaced with
hyphens. For instance, GNUnet is available as @code{gnunet}, and
SDL_net as @code{sdl-net}.
We do not add @code{lib} prefixes for library packages, unless these are
already part of the official project name. But see @pxref{Python
Modules} and @ref{Perl Modules} for special rules concerning modules for
the Python and Perl languages.
@node Version Numbers
@subsection Version Numbers
We usually package only the latest version of a given free software
project. But sometimes, for instance for incompatible library versions,
two (or more) versions of the same package are needed. These require
different Scheme variable names. We use the name as defined
in @ref{Package Naming}
for the most recent version; previous versions use the same name, suffixed
by @code{-} and the smallest prefix of the version number that may
distinguish the two versions.
The name inside the package definition is the same for all versions of a
package and does not contain any version number.
For instance, the versions 2.24.20 and 3.9.12 of GTK+ may be packaged as follows:
@example
(define-public gtk+
(package
(name "gtk+")
(version "3.9.12")
...))
(define-public gtk+-2
(package
(name "gtk+")
(version "2.24.20")
...))
@end example
If we also wanted GTK+ 3.8.2, this would be packaged as
@example
(define-public gtk+-3.8
(package
(name "gtk+")
(version "3.8.2")
...))
@end example
@node Python Modules
@subsection Python Modules
We currently package Python 2 and Python 3, under the Scheme variable names
@code{python-2} and @code{python} as explained in @ref{Version Numbers}.
To avoid confusion and naming clashes with other programming languages, it
seems desirable that the name of a package for a Python module contains
the word @code{python}.
Some modules are compatible with only one version of Python, others with both.
If the package Foo compiles only with Python 3, we name it
@code{python-foo}; if it compiles only with Python 2, we name it
@code{python2-foo}. If it is compatible with both versions, we create two
packages with the corresponding names.
If a project already contains the word @code{python}, we drop this;
for instance, the module python-dateutil is packaged under the names
@code{python-dateutil} and @code{python2-dateutil}.
@node Perl Modules
@subsection Perl Modules
Perl programs standing for themselves are named as any other package,
using the lowercase upstream name.
For Perl packages containing a single class, we use the lowercase class name,
replace all occurrences of @code{::} by dashes and prepend the prefix
@code{perl-}.
So the class @code{XML::Parser} becomes @code{perl-xml-parser}.
Modules containing several classes keep their lowercase upstream name and
are also prepended by @code{perl-}. Such modules tend to have the word
@code{perl} somewhere in their name, which gets dropped in favor of the
prefix. For instance, @code{libwww-perl} becomes @code{perl-libwww}.
@node Bootstrapping
@section Bootstrapping
@c Adapted from the ELS 2013 paper.
@cindex bootstrapping
Bootstrapping in our context refers to how the distribution gets built
``from nothing''. Remember that the build environment of a derivation
contains nothing but its declared inputs (@pxref{Introduction}). So
there's an obvious chicken-and-egg problem: how does the first package
get built? How does the first compiler get compiled? Note that this is
a question of interest only to the curious hacker, not to the regular
user, so you can shamelessly skip this section if you consider yourself
a ``regular user''.
@cindex bootstrap binaries
The GNU system is primarily made of C code, with libc at its core. The
GNU build system itself assumes the availability of a Bourne shell and
command-line tools provided by GNU Coreutils, Awk, Findutils, `sed', and
`grep'. Furthermore, build programs---programs that run
@code{./configure}, @code{make}, etc.---are written in Guile Scheme
(@pxref{Derivations}). Consequently, to be able to build anything at
all, from scratch, Guix relies on pre-built binaries of Guile, GCC,
Binutils, libc, and the other packages mentioned above---the
@dfn{bootstrap binaries}.
These bootstrap binaries are ``taken for granted'', though we can also
re-create them if needed (more on that later).
@unnumberedsubsec Preparing to Use the Bootstrap Binaries
@c As of Emacs 24.3, Info-mode displays the image, but since it's a
@c large image, it's hard to scroll. Oh well.
@image{images/bootstrap-graph,6in,,Dependency graph of the early bootstrap derivations}
The figure above shows the very beginning of the dependency graph of the
distribution, corresponding to the package definitions of the @code{(gnu
packages bootstrap)} module. At this level of detail, things are
slightly complex. First, Guile itself consists of an ELF executable,
along with many source and compiled Scheme files that are dynamically
loaded when it runs. This gets stored in the @file{guile-2.0.7.tar.xz}
tarball shown in this graph. This tarball is part of Guix's ``source''
distribution, and gets inserted into the store with @code{add-to-store}
(@pxref{The Store}).
But how do we write a derivation that unpacks this tarball and adds it
to the store? To solve this problem, the @code{guile-bootstrap-2.0.drv}
derivation---the first one that gets built---uses @code{bash} as its
builder, which runs @code{build-bootstrap-guile.sh}, which in turn calls
@code{tar} to unpack the tarball. Thus, @file{bash}, @file{tar},
@file{xz}, and @file{mkdir} are statically-linked binaries, also part of
the Guix source distribution, whose sole purpose is to allow the Guile
tarball to be unpacked.
Once @code{guile-bootstrap-2.0.drv} is built, we have a functioning
Guile that can be used to run subsequent build programs. Its first task
is to download tarballs containing the other pre-built binaries---this
is what the @code{.tar.xz.drv} derivations do. Guix modules such as
@code{ftp-client.scm} are used for this purpose. The
@code{module-import.drv} derivations import those modules in a directory
in the store, using the original layout. The
@code{module-import-compiled.drv} derivations compile those modules, and
write them in an output directory with the right layout. This
corresponds to the @code{#:modules} argument of
@code{build-expression->derivation} (@pxref{Derivations}).
Finally, the various tarballs are unpacked by the
derivations @code{gcc-bootstrap-0.drv}, @code{glibc-bootstrap-0.drv},
etc., at which point we have a working C tool chain.
@unnumberedsubsec Building the Build Tools
@c TODO: Add a package-level dependency graph generated from (gnu
@c packages base).
Bootstrapping is complete when we have a full tool chain that does not
depend on the pre-built bootstrap tools discussed above. This
no-dependency requirement is verified by checking whether the files of
the final tool chain contain references to the @file{/gnu/store}
directories of the bootstrap inputs. The process that leads to this
``final'' tool chain is described by the package definitions found in
the @code{(gnu packages base)} module.
@c See <http://lists.gnu.org/archive/html/gnu-system-discuss/2012-10/msg00000.html>.
The first tool that gets built with the bootstrap binaries is
GNU Make, which is a prerequisite for all the following packages.
From there Findutils and Diffutils get built.
Then come the first-stage Binutils and GCC, built as pseudo cross
tools---i.e., with @code{--target} equal to @code{--host}. They are
used to build libc. Thanks to this cross-build trick, this libc is
guaranteed not to hold any reference to the initial tool chain.
From there the final Binutils and GCC are built. GCC uses @code{ld}
from the final Binutils, and links programs against the just-built libc.
This tool chain is used to build the other packages used by Guix and by
the GNU Build System: Guile, Bash, Coreutils, etc.
And voilà! At this point we have the complete set of build tools that
the GNU Build System expects. These are in the @code{%final-inputs}
variables of the @code{(gnu packages base)} module, and are implicitly
used by any package that uses @code{gnu-build-system} (@pxref{Defining
Packages}).
@unnumberedsubsec Building the Bootstrap Binaries
Because the final tool chain does not depend on the bootstrap binaries,
those rarely need to be updated. Nevertheless, it is useful to have an
automated way to produce them, should an update occur, and this is what
the @code{(gnu packages make-bootstrap)} module provides.
The following command builds the tarballs containing the bootstrap
binaries (Guile, Binutils, GCC, libc, and a tarball containing a mixture
of Coreutils and other basic command-line tools):
@example
guix build bootstrap-tarballs
@end example
The generated tarballs are those that should be referred to in the
@code{(gnu packages bootstrap)} module mentioned at the beginning of
this section.
Still here? Then perhaps by now you've started to wonder: when do we
reach a fixed point? That is an interesting question! The answer is
unknown, but if you would like to investigate further (and have
significant computational and storage resources to do so), then let us
know.
@node Porting
@section Porting to a New Platform
As discussed above, the GNU distribution is self-contained, and
self-containment is achieved by relying on pre-built ``bootstrap
binaries'' (@pxref{Bootstrapping}). These binaries are specific to an
operating system kernel, CPU architecture, and application binary
interface (ABI). Thus, to port the distribution to a platform that is
not yet supported, one must build those bootstrap binaries, and update
the @code{(gnu packages bootstrap)} module to use them on that platform.
Fortunately, Guix can @emph{cross compile} those bootstrap binaries.
When everything goes well, and assuming the GNU tool chain supports the
target platform, this can be as simple as running a command like this
one:
@example
guix build --target=armv5tel-linux-gnueabi bootstrap-tarballs
@end example
Once these are built, the @code{(gnu packages bootstrap)} module needs
to be updated to refer to these binaries on the target platform. In
addition, the @code{glibc-dynamic-linker} procedure in that module must
be augmented to return the right file name for libc's dynamic linker on
that platform; likewise, @code{system->linux-architecture} in @code{(gnu
packages linux)} must be taught about the new platform.
In practice, there may be some complications. First, it may be that the
extended GNU triplet that specifies an ABI (like the @code{eabi} suffix
above) is not recognized by all the GNU tools. Typically, glibc
recognizes some of these, whereas GCC uses an extra @code{--with-abi}
configure flag (see @code{gcc.scm} for examples of how to handle this).
Second, some of the required packages could fail to build for that
platform. Lastly, the generated binaries could be broken for some
reason.
@node System Configuration
@section System Configuration
@ -3846,6 +3379,472 @@ on-line documentation. Thus, the commands @command{deco start ncsd},
would expect (@pxref{Invoking deco,,, dmd, GNU dmd Manual}).
@node Installing Debugging Files
@section Installing Debugging Files
@cindex debugging files
Program binaries, as produced by the GCC compilers for instance, are
typically written in the ELF format, with a section containing
@dfn{debugging information}. Debugging information is what allows the
debugger, GDB, to map binary code to source code; it is required to
debug a compiled program in good conditions.
The problem with debugging information is that is takes up a fair amount
of disk space. For example, debugging information for the GNU C Library
weighs in at more than 60 MiB. Thus, as a user, keeping all the
debugging info of all the installed programs is usually not an option.
Yet, space savings should not come at the cost of an impediment to
debugging---especially in the GNU system, which should make it easier
for users to exert their computing freedom (@pxref{GNU Distribution}).
Thankfully, the GNU Binary Utilities (Binutils) and GDB provide a
mechanism that allows users to get the best of both worlds: debugging
information can be stripped from the binaries and stored in separate
files. GDB is then able to load debugging information from those files,
when they are available (@pxref{Separate Debug Files,,, gdb, Debugging
with GDB}).
The GNU distribution takes advantage of this by storing debugging
information in the @code{lib/debug} sub-directory of a separate package
output unimaginatively called @code{debug} (@pxref{Packages with
Multiple Outputs}). Users can choose to install the @code{debug} output
of a package when they need it. For instance, the following command
installs the debugging information for the GNU C Library and for GNU
Guile:
@example
guix package -i glibc:debug guile:debug
@end example
GDB must then be told to look for debug files in the user's profile, by
setting the @code{debug-file-directory} variable (consider setting it
from the @file{~/.gdbinit} file, @pxref{Startup,,, gdb, Debugging with
GDB}):
@example
(gdb) set debug-file-directory ~/.guix-profile/lib/debug
@end example
From there on, GDB will pick up debugging information from the
@code{.debug} files under @file{~/.guix-profile/lib/debug}.
In addition, you will most likely want GDB to be able to show the source
code being debugged. To do that, you will have to unpack the source
code of the package of interest (obtained with @code{guix build
--source}, @pxref{Invoking guix build}), and to point GDB to that source
directory using the @code{directory} command (@pxref{Source Path,
@code{directory},, gdb, Debugging with GDB}).
@c XXX: keep me up-to-date
The @code{debug} output mechanism in Guix is implemented by the
@code{gnu-build-system} (@pxref{Build Systems}). Currently, it is
opt-in---debugging information is available only for those packages
whose definition explicitly declares a @code{debug} output. This may be
changed to opt-out in the future, if our build farm servers can handle
the load. To check whether a package has a @code{debug} output, use
@command{guix package --list-available} (@pxref{Invoking guix package}).
@node Package Modules
@section Package Modules
From a programming viewpoint, the package definitions of the
GNU distribution are provided by Guile modules in the @code{(gnu packages
@dots{})} name space@footnote{Note that packages under the @code{(gnu
packages @dots{})} module name space are not necessarily ``GNU
packages''. This module naming scheme follows the usual Guile module
naming convention: @code{gnu} means that these modules are distributed
as part of the GNU system, and @code{packages} identifies modules that
define packages.} (@pxref{Modules, Guile modules,, guile, GNU Guile
Reference Manual}). For instance, the @code{(gnu packages emacs)}
module exports a variable named @code{emacs}, which is bound to a
@code{<package>} object (@pxref{Defining Packages}).
The @code{(gnu packages @dots{})} module name space is special: it is
automatically scanned for packages by the command-line tools. For
instance, when running @code{guix package -i emacs}, all the @code{(gnu
packages @dots{})} modules are scanned until one that exports a package
object whose name is @code{emacs} is found. This package search
facility is implemented in the @code{(gnu packages)} module.
Users can store package definitions in modules with different
names---e.g., @code{(my-packages emacs)}. In that case, commands such
as @command{guix package} and @command{guix build} have to be used with
the @code{-e} option so that they know where to find the package.
The distribution is fully @dfn{bootstrapped} and @dfn{self-contained}:
each package is built based solely on other packages in the
distribution. The root of this dependency graph is a small set of
@dfn{bootstrap binaries}, provided by the @code{(gnu packages
bootstrap)} module. For more information on bootstrapping,
@ref{Bootstrapping}.
@node Packaging Guidelines
@section Packaging Guidelines
The GNU distribution is nascent and may well lack some of your favorite
packages. This section describes how you can help make the distribution
grow. @xref{Contributing}, for additional information on how you can
help.
Free software packages are usually distributed in the form of
@dfn{source code tarballs}---typically @file{tar.gz} files that contain
all the source files. Adding a package to the distribution means
essentially two things: adding a @dfn{recipe} that describes how to
build the package, including a list of other packages required to build
it, and adding @dfn{package meta-data} along with that recipe, such as a
description and licensing information.
In Guix all this information is embodied in @dfn{package definitions}.
Package definitions provide a high-level view of the package. They are
written using the syntax of the Scheme programming language; in fact,
for each package we define a variable bound to the package definition,
and export that variable from a module (@pxref{Package Modules}).
However, in-depth Scheme knowledge is @emph{not} a prerequisite for
creating packages. For more information on package definitions,
@ref{Defining Packages}.
Once a package definition is in place, stored in a file in the Guix
source tree, it can be tested using the @command{guix build} command
(@pxref{Invoking guix build}). For example, assuming the new package is
called @code{gnew}, you may run this command from the Guix build tree:
@example
./pre-inst-env guix build gnew --keep-failed
@end example
Using @code{--keep-failed} makes it easier to debug build failures since
it provides access to the failed build tree. Another useful
command-line option when debugging is @code{--log-file}, to access the
build log.
If the package is unknown to the @command{guix} command, it may be that
the source file contains a syntax error, or lacks a @code{define-public}
clause to export the package variable. To figure it out, you may load
the module from Guile to get more information about the actual error:
@example
./pre-inst-env guile -c '(use-modules (gnu packages gnew))'
@end example
Once your package builds correctly, please send us a patch
(@pxref{Contributing}). Well, if you need help, we will be happy to
help you too. Once the patch is committed in the Guix repository, the
new package automatically gets built on the supported platforms by
@url{http://hydra.gnu.org/gnu/master, our continuous integration
system}.
@cindex substituter
Users can obtain the new package definition simply by running
@command{guix pull} (@pxref{Invoking guix pull}). When
@code{hydra.gnu.org} is done building the package, installing the
package automatically downloads binaries from there
(@pxref{Substitutes}). The only place where human intervention is
needed is to review and apply the patch.
@menu
* Software Freedom:: What may go into the distribution.
* Package Naming:: What's in a name?
* Version Numbers:: When the name is not enough.
* Python Modules:: Taming the snake.
* Perl Modules:: Little pearls.
@end menu
@node Software Freedom
@subsection Software Freedom
@c Adapted from http://www.gnu.org/philosophy/philosophy.html.
The GNU operating system has been developed so that users can have
freedom in their computing. GNU is @dfn{free software}, meaning that
users have the @url{http://www.gnu.org/philosophy/free-sw.html,four
essential freedoms}: to run the program, to study and change the program
in source code form, to redistribute exact copies, and to distribute
modified versions. Packages found in the GNU distribution provide only
software that conveys these four freedoms.
In addition, the GNU distribution follow the
@url{http://www.gnu.org/distros/free-system-distribution-guidelines.html,free
software distribution guidelines}. Among other things, these guidelines
reject non-free firmware, recommendations of non-free software, and
discuss ways to deal with trademarks and patents.
Some packages contain a small and optional subset that violates the
above guidelines, for instance because this subset is itself non-free
code. When that happens, the offending items are removed with
appropriate patches or code snippets in the package definition's
@code{origin} form (@pxref{Defining Packages}). That way, @code{guix
build --source} returns the ``freed'' source rather than the unmodified
upstream source.
@node Package Naming
@subsection Package Naming
A package has actually two names associated with it:
First, there is the name of the @emph{Scheme variable}, the one following
@code{define-public}. By this name, the package can be made known in the
Scheme code, for instance as input to another package. Second, there is
the string in the @code{name} field of a package definition. This name
is used by package management commands such as
@command{guix package} and @command{guix build}.
Both are usually the same and correspond to the lowercase conversion of
the project name chosen upstream, with underscores replaced with
hyphens. For instance, GNUnet is available as @code{gnunet}, and
SDL_net as @code{sdl-net}.
We do not add @code{lib} prefixes for library packages, unless these are
already part of the official project name. But see @pxref{Python
Modules} and @ref{Perl Modules} for special rules concerning modules for
the Python and Perl languages.
@node Version Numbers
@subsection Version Numbers
We usually package only the latest version of a given free software
project. But sometimes, for instance for incompatible library versions,
two (or more) versions of the same package are needed. These require
different Scheme variable names. We use the name as defined
in @ref{Package Naming}
for the most recent version; previous versions use the same name, suffixed
by @code{-} and the smallest prefix of the version number that may
distinguish the two versions.
The name inside the package definition is the same for all versions of a
package and does not contain any version number.
For instance, the versions 2.24.20 and 3.9.12 of GTK+ may be packaged as follows:
@example
(define-public gtk+
(package
(name "gtk+")
(version "3.9.12")
...))
(define-public gtk+-2
(package
(name "gtk+")
(version "2.24.20")
...))
@end example
If we also wanted GTK+ 3.8.2, this would be packaged as
@example
(define-public gtk+-3.8
(package
(name "gtk+")
(version "3.8.2")
...))
@end example
@node Python Modules
@subsection Python Modules
We currently package Python 2 and Python 3, under the Scheme variable names
@code{python-2} and @code{python} as explained in @ref{Version Numbers}.
To avoid confusion and naming clashes with other programming languages, it
seems desirable that the name of a package for a Python module contains
the word @code{python}.
Some modules are compatible with only one version of Python, others with both.
If the package Foo compiles only with Python 3, we name it
@code{python-foo}; if it compiles only with Python 2, we name it
@code{python2-foo}. If it is compatible with both versions, we create two
packages with the corresponding names.
If a project already contains the word @code{python}, we drop this;
for instance, the module python-dateutil is packaged under the names
@code{python-dateutil} and @code{python2-dateutil}.
@node Perl Modules
@subsection Perl Modules
Perl programs standing for themselves are named as any other package,
using the lowercase upstream name.
For Perl packages containing a single class, we use the lowercase class name,
replace all occurrences of @code{::} by dashes and prepend the prefix
@code{perl-}.
So the class @code{XML::Parser} becomes @code{perl-xml-parser}.
Modules containing several classes keep their lowercase upstream name and
are also prepended by @code{perl-}. Such modules tend to have the word
@code{perl} somewhere in their name, which gets dropped in favor of the
prefix. For instance, @code{libwww-perl} becomes @code{perl-libwww}.
@node Bootstrapping
@section Bootstrapping
@c Adapted from the ELS 2013 paper.
@cindex bootstrapping
Bootstrapping in our context refers to how the distribution gets built
``from nothing''. Remember that the build environment of a derivation
contains nothing but its declared inputs (@pxref{Introduction}). So
there's an obvious chicken-and-egg problem: how does the first package
get built? How does the first compiler get compiled? Note that this is
a question of interest only to the curious hacker, not to the regular
user, so you can shamelessly skip this section if you consider yourself
a ``regular user''.
@cindex bootstrap binaries
The GNU system is primarily made of C code, with libc at its core. The
GNU build system itself assumes the availability of a Bourne shell and
command-line tools provided by GNU Coreutils, Awk, Findutils, `sed', and
`grep'. Furthermore, build programs---programs that run
@code{./configure}, @code{make}, etc.---are written in Guile Scheme
(@pxref{Derivations}). Consequently, to be able to build anything at
all, from scratch, Guix relies on pre-built binaries of Guile, GCC,
Binutils, libc, and the other packages mentioned above---the
@dfn{bootstrap binaries}.
These bootstrap binaries are ``taken for granted'', though we can also
re-create them if needed (more on that later).
@unnumberedsubsec Preparing to Use the Bootstrap Binaries
@c As of Emacs 24.3, Info-mode displays the image, but since it's a
@c large image, it's hard to scroll. Oh well.
@image{images/bootstrap-graph,6in,,Dependency graph of the early bootstrap derivations}
The figure above shows the very beginning of the dependency graph of the
distribution, corresponding to the package definitions of the @code{(gnu
packages bootstrap)} module. At this level of detail, things are
slightly complex. First, Guile itself consists of an ELF executable,
along with many source and compiled Scheme files that are dynamically
loaded when it runs. This gets stored in the @file{guile-2.0.7.tar.xz}
tarball shown in this graph. This tarball is part of Guix's ``source''
distribution, and gets inserted into the store with @code{add-to-store}
(@pxref{The Store}).
But how do we write a derivation that unpacks this tarball and adds it
to the store? To solve this problem, the @code{guile-bootstrap-2.0.drv}
derivation---the first one that gets built---uses @code{bash} as its
builder, which runs @code{build-bootstrap-guile.sh}, which in turn calls
@code{tar} to unpack the tarball. Thus, @file{bash}, @file{tar},
@file{xz}, and @file{mkdir} are statically-linked binaries, also part of
the Guix source distribution, whose sole purpose is to allow the Guile
tarball to be unpacked.
Once @code{guile-bootstrap-2.0.drv} is built, we have a functioning
Guile that can be used to run subsequent build programs. Its first task
is to download tarballs containing the other pre-built binaries---this
is what the @code{.tar.xz.drv} derivations do. Guix modules such as
@code{ftp-client.scm} are used for this purpose. The
@code{module-import.drv} derivations import those modules in a directory
in the store, using the original layout. The
@code{module-import-compiled.drv} derivations compile those modules, and
write them in an output directory with the right layout. This
corresponds to the @code{#:modules} argument of
@code{build-expression->derivation} (@pxref{Derivations}).
Finally, the various tarballs are unpacked by the
derivations @code{gcc-bootstrap-0.drv}, @code{glibc-bootstrap-0.drv},
etc., at which point we have a working C tool chain.
@unnumberedsubsec Building the Build Tools
@c TODO: Add a package-level dependency graph generated from (gnu
@c packages base).
Bootstrapping is complete when we have a full tool chain that does not
depend on the pre-built bootstrap tools discussed above. This
no-dependency requirement is verified by checking whether the files of
the final tool chain contain references to the @file{/gnu/store}
directories of the bootstrap inputs. The process that leads to this
``final'' tool chain is described by the package definitions found in
the @code{(gnu packages base)} module.
@c See <http://lists.gnu.org/archive/html/gnu-system-discuss/2012-10/msg00000.html>.
The first tool that gets built with the bootstrap binaries is
GNU Make, which is a prerequisite for all the following packages.
From there Findutils and Diffutils get built.
Then come the first-stage Binutils and GCC, built as pseudo cross
tools---i.e., with @code{--target} equal to @code{--host}. They are
used to build libc. Thanks to this cross-build trick, this libc is
guaranteed not to hold any reference to the initial tool chain.
From there the final Binutils and GCC are built. GCC uses @code{ld}
from the final Binutils, and links programs against the just-built libc.
This tool chain is used to build the other packages used by Guix and by
the GNU Build System: Guile, Bash, Coreutils, etc.
And voilà! At this point we have the complete set of build tools that
the GNU Build System expects. These are in the @code{%final-inputs}
variables of the @code{(gnu packages base)} module, and are implicitly
used by any package that uses @code{gnu-build-system} (@pxref{Defining
Packages}).
@unnumberedsubsec Building the Bootstrap Binaries
Because the final tool chain does not depend on the bootstrap binaries,
those rarely need to be updated. Nevertheless, it is useful to have an
automated way to produce them, should an update occur, and this is what
the @code{(gnu packages make-bootstrap)} module provides.
The following command builds the tarballs containing the bootstrap
binaries (Guile, Binutils, GCC, libc, and a tarball containing a mixture
of Coreutils and other basic command-line tools):
@example
guix build bootstrap-tarballs
@end example
The generated tarballs are those that should be referred to in the
@code{(gnu packages bootstrap)} module mentioned at the beginning of
this section.
Still here? Then perhaps by now you've started to wonder: when do we
reach a fixed point? That is an interesting question! The answer is
unknown, but if you would like to investigate further (and have
significant computational and storage resources to do so), then let us
know.
@node Porting
@section Porting to a New Platform
As discussed above, the GNU distribution is self-contained, and
self-containment is achieved by relying on pre-built ``bootstrap
binaries'' (@pxref{Bootstrapping}). These binaries are specific to an
operating system kernel, CPU architecture, and application binary
interface (ABI). Thus, to port the distribution to a platform that is
not yet supported, one must build those bootstrap binaries, and update
the @code{(gnu packages bootstrap)} module to use them on that platform.
Fortunately, Guix can @emph{cross compile} those bootstrap binaries.
When everything goes well, and assuming the GNU tool chain supports the
target platform, this can be as simple as running a command like this
one:
@example
guix build --target=armv5tel-linux-gnueabi bootstrap-tarballs
@end example
Once these are built, the @code{(gnu packages bootstrap)} module needs
to be updated to refer to these binaries on the target platform. In
addition, the @code{glibc-dynamic-linker} procedure in that module must
be augmented to return the right file name for libc's dynamic linker on
that platform; likewise, @code{system->linux-architecture} in @code{(gnu
packages linux)} must be taught about the new platform.
In practice, there may be some complications. First, it may be that the
extended GNU triplet that specifies an ABI (like the @code{eabi} suffix
above) is not recognized by all the GNU tools. Typically, glibc
recognizes some of these, whereas GCC uses an extra @code{--with-abi}
configure flag (see @code{gcc.scm} for examples of how to handle this).
Second, some of the required packages could fail to build for that
platform. Lastly, the generated binaries could be broken for some
reason.
@c *********************************************************************
@node Contributing
@chapter Contributing