From Fedora Project Wiki

No edit summary
(59 intermediate revisions by 22 users not shown)
Line 1: Line 1:
This is a tutorial on writing RPM files, suitable for someone familiar with
{{autolang|base=yes}}
the FOSS software building process, and with using pre-made RPM packages.
This is a short hands-on tutorial on writing RPM files, showing how to quickly step up to create simple source and binary software packages. It assumes  some familiarity with using pre-made RPM packages, and with the FOSS software building process.  


This tutorial uses the GNU "Hello World". While 'Hello World" is a trivial program, the GNU project
For a comprehensive information on how to create RPM files, including more detailed tips, refer to [[How to create an RPM package]]. If you plan to create an RPM package for the Fedora repository, follow the process for [[Join the package collection maintainers|How to join the Fedora Package Collection Maintainers]], including following the various Fedora guidance.
contains most of the usual peripheral components associated with a typical FOSS software project,
such as the configuration/build/install environment, documentation, internationalization, etc., so it
is a reasonable vehicle to practice building RPMs on.


I wrote this tutorial after reading Christoph Wickert's IRC class on building RPMs.
This tutorial demonstrates packaging of the GNU "Hello World" project. While the C program printing 'Hello World" to standard output is trivial, the GNU version contains most of the usual peripheral components associated with a typical FOSS software project, including the configuration/build/install environment, documentation, internationalization, etc. The GNU version, however, traditionally consists of a tar file containing the source code and configure/make scripts, but it does not include the packaging information. Therefore, it's a reasonable vehicle to practice building RPMs on.
Rahul Sundaram suggested using GNU "Hello World" as a test case, and it turns out
that Christian Lyder Jacobsen also had
[http://www.absolutepanic.org/blog/2009/07/building-a-gnu-hello-world-rpm this idea]
in 2009. Christian is not intending to update his website, so I decided to publish my
version in this wiki.


== Development environment ==
== Development environment ==


To build RPMs we need a set of development tools. This is a one-time-only setup,
To build RPMs we need a set of development tools. This is a one-time-only setup, installed by running those commands from a system administration (<code>root</code>) account:
installed most simply by
running those commands from a system administration (<code>root</code>) account:


<pre>
<pre>
yum groupinstall development-tools
# yum install @development-tools
yum install rpm-build rpmdevtools
# yum install fedora-packager
</pre>
</pre>


Those are the only commands requiring <code>root</code> privileges. All the remaining
If you want to test the build procedure in a clean chroot, you need to configure your non-privileged account to be a member of the 'mock' group:
work should be done from your regular, non-privileged account. Modern RPM-based
systems, including Fedora, are set up to build and test RPM packages purely from
within a non-privileged account, which can be set up by running the command


<pre>rpmdev-setuptree</pre>
<pre>
# usermod -a -G mock <your username>
</pre>
 
Those are the only commands requiring <code>root</code> privileges. All the remaining work should be done from your regular, non-privileged account, or even from a separate account created just for development work. Modern RPM-based systems, including Fedora, are set up to build and test RPM packages purely from within a non-privileged account. The command 
 
<pre>$ rpmdev-setuptree</pre>
 
sets up a RPM build area in your <code>~/rpmbuild</code>. This directory will contain several subdirectories, for the project source code, RPM configuration files and for the resulting source and binary packages.
 
== Building a "Hello World" RPM ==
 
We need the source code of the project we are packaging, often referred
to as the 'upstream' source. We will download it from the project's website  into the <code>~/rpmbuild/SOURCE</code>
directory. We are getting the compressed tarball archive, which happens to be a preferred distribution form for
most FOSS projects.


This command sets up a RPM build
<pre>
area in your home directory, usually in <code>~/rpmbuild</code> or <code>~/rpm</code>.
$ cd ~/rpmbuild/SOURCES
$ wget http://ftp.gnu.org/gnu/hello/hello-2.8.tar.gz
</pre>


== Building "Hello World" ==
The RPM package is configured by <code>.spec</code> files. We will create a template
file <code> hello.spec</code> in the appropriate directory:


We need the source code of the project we are packaging in the ~/rpm/SOURCE directory. We need
<pre>
a compressed tarball, which happens to be a preferred distribution form for most FOSS projects
$ cd ~/rpmbuild/SPECS
$ rpmdev-newspec hello
</pre>
 
Recent versions of <code>Emacs</code> and <code>vi</code> have .spec file editing modes which will also bring up a similar template upon creating a new file. So you can just use the following command for example to use the template automatically.


<pre>
<pre>
cd ~/rpm/SOURCE
vi hello.spec
wget  http://ftp.gnu.org/gnu/hello/hello-2.5.tar.gz
</pre>
</pre>


The
=== Inside a <code>.spec</code> file ===
Next, you want to create a template .spec file in the ~/rpm/SPECS directory:


cd ~/rpm/SPECS
The fields in our <code>.spec</code> file need slight editing. Please follow the [[How_to_create_an_RPM_package#Spec_file_pieces_explained|Fedora rules]] for these fields. In our case, the file might start as follows:
rpmdev-newspec hello
 
<pre>
Name:    hello
Version:  2.8
Release:  1
Summary:  The "Hello World" program from GNU
License:  GPLv3+
URL:      http://ftp.gnu.org/gnu/hello   
Source0:  http://ftp.gnu.org/gnu/hello/hello-2.8.tar.gz


This will create a template spec file, hello.spec. The fields in
%description
this file need slight editing as described below:
The "Hello World" program, done with all bells and whistles of a proper FOSS
project, including configuration, build, internationalization, help files, etc.


Name: hello
%changelog
Version: 2.5 (the version is from upstream while Release is from Fedora)
* Thu Jul 07 2011 The Coon of Ty <Ty@coon.org> - 2.8-1
Release: 1
- Initial version of the package
Summary: should be short and precise. Only keywords, simple phrases
</pre>
e.g. "Widget management application". First letter uppercase to avoid
rpmlint complaints
Group: from /usr/share/doc/rpm-4.6.0/GROUPS    but groups not used any more
License: GPLv3 (check the source for the license it is released under)
URL: http://ftp.gnu.org/gnu/hello    (The homepage of the program)
Source0: http://ftp.gnu.org/gnu/hello/hello-2.5.tar.gz (the URL for
the source distribution files)


Now comment out BuildRequires and Requires with a # for now,
The <code>Version</code> should mirror upstream while <code> Release</code> numbers our work within Fedora.


%description:
The first letter of the <code> Summary</code> should be uppercase to avoid rpmlint complaints.
Package summary; text starts on next line, ends with empty line


BuildRoot is the location where we are building the package,
It is your responsibility to check the <code>License</code> status of the software, by inspecting the source files and/or their LICENSE files, and/or by talking to the authors.
temporary folder, default is OK


%prep is for preparing , eg. extracting the source and applying patches
The <code> Group </code> tag was historically used to classify the package in accordance to the list in <code>/usr/share/doc/rpm-<version>/GROUPS</code>. It is being phased out so you will not see it added by default. However, it doesn't hurt to add it anyway.
if there are any


The actual build commands are specified in %build
The <code> %changelog</code> should document the work on preparing the RPM, especially if there are security and bug patches included on top of the base upstream source. Changelog data can be displayed by <code>rpm --changelog -q <packagename></code>, which is very useful for instance to find out if specific bug and security patches were included in the installed software, thanks to diligent Fedora packagers who include this info with the relevant [http://cve.mitre.org/ CVE] numbers.
usually you run ./configure and  make.
A recently popular 'make' replacement waf is sometimes used used for
builds; it automatically configures the build:
./waf build replaces %configure; make %{?_smp_mflags}


The changelog entry should include the version string to avoid rpmlint complaints.


After that the files are installed to a temporary location during
Multi-line sections like <code> %changelog</code> or <code> %description</code> start on a line under the directive, and end with an empty line.
%install, usually 'make install'


%install
Lines which aren't needed (e.g. <code>BuildRequires</code> and <code>Requires</code>) can be commented out with a hash ('#') for now.
rm -rf $RPM_BUILD_ROOT  remains as is for cleaning up old stuff
from a previous build that failed
"make install DESTDIR=$RPM_BUILD_ROOT" or
"DESTDIR=$RPM_BUILD_ROOT ./waf install"


%clean remains as is
Many lines in the template don't need to be changed at all in many cases, at least for the initial attempt.


The template %changelog is as follows (include the version to avoid
=== Building the package ===
rpmlint complains)
* Sun Apr 05 2009 Foo Bar <foo@bar.com> - 0.3.1-1
- Initial attempt


We are ready for the first  run to build source,  binary and debugging packages:
We are ready for the first  run to build source,  binary and debugging packages:


rpmbuild -ba hello.spec
<pre>
$ rpmbuild -ba hello.spec
</pre>
 
It will complain and list the unpackaged files, i.e. the files that would be  installed in the system that weren't  declared as belonging to the package. We need to declare them in the <code>%files</code> section. Do not hardcode names like <code>/usr/bin/</code>, but use macros, like <code>%{_bindir}/hello</code> instead. The manual pages should be declared in the <code>%doc</code> subsection: <code>%doc %{_mandir}/man1/hello.1.gz</code>.
 
This is an iterative process: after editing the <code>.spec</code> file, rerun <code>rpmbuild</code>.
 
Since our program uses translations and internationalization, we are seeing a lot of undeclared i18 files. The [[Packaging:Guidelines#Handling_Locale_Files|recommended method]] to declare them is:
 
* find the filenames in the <code>%install</code> step: <code> %find_lang %{name}</code>
* add the required build dependencies: <code>BuildRequires: gettext</code>
* use the found filenames <code>%files -f %{name}.lang</code>
 
If the program uses GNU info files, you need to make sure the installation and uninstallation
of the package does not interfere with other software on the system, by using this boilerplate:
 
* delete the 'dir' file in %install:  <code>rm -f %{buildroot}/%{_infodir}/dir</code>
* <code>Requires(post): info</code> and <code>Requires(preun): info</code>
* add those steps:
<pre>
%post
/sbin/install-info %{_infodir}/%{name}.info %{_infodir}/dir || :
 
%preun
if [ $1 = 0 ] ; then
/sbin/install-info --delete %{_infodir}/%{name}.info %{_infodir}/dir || :
fi
</pre>
 
=== A complete <code>hello.spec</code> file ===
 
Here's the initial version of <code>hello.spec</code>:
 
<pre>
Name:          hello
Version:        2.8
Release:        1%{?dist}
Summary:        The "Hello World" program from GNU
 
License:        GPLv3+
URL:            http://ftp.gnu.org/gnu/%{name}
Source0:        http://ftp.gnu.org/gnu/%{name}/%{name}-%{version}.tar.gz


It will probably complain about unpackaged files. We need to declare them in the
BuildRequires: gettext
%files section. When you add them, do not hardcode /usr/bin/, but use
     
%{_bindir}/hello instead
Requires(post): info
After editing, rerun rpmbuild.
Requires(preun): info


If the program uses translations, use %find_lang ${name}
%description
in %install and BuildRequires: gettext , and %files -f ${name}.lang as
The "Hello World" program, done with all bells and whistles of a proper FOSS
explained in
project, including configuration, build, internationalization, help files, etc.
https://fedoraproject.org/wiki/Packaging/Guidelines#Handling_Locale_Files


If the program uses GNU info files, you need to do some magic:
%prep
- delete the 'dir' file in %installrm -f $RPM_BUILD_ROOT/usr/share/info/dir
%autosetup
- Requires(post): info and Requires(preun): info
 
- add postinstall and preuninstall steps to configure info:
%build
%configure
make %{?_smp_mflags}
 
%install
%make_install
%find_lang %{name}
rm -f %{buildroot}/%{_infodir}/dir


%post
%post
Line 127: Line 170:
fi
fi


With this spec file, you should be able to successfully recreate the
%files -f %{name}.lang
RPMs. Next you should check them for conformance with RPM design
%{_mandir}/man1/hello.1.gz
rules, by running rpmlint
%{_infodir}/%{name}.info.gz
on the spec file and all RPMs: rpmlint hello.spec ../SRPMS/hello*
%{_bindir}/hello
../RPMS/*/hello*
 
If there are no warnings or errors, we've succeeded.
%doc
AUTHORS ChangeLog COPYING NEWS README THANKS TODO
 
%changelog
* Tue Sep 06 2011 The Coon of Ty <Ty@coon.org> 2.8-1
- Initial version of the package
</pre>
 
With this spec file, you should be able to successfully complete the build process, and create the source and binary RPM packages.
 
Next you should check them for conformance with RPM design rules, by running <code>rpmlint</code> on the spec file and all RPMs:  
 
<pre>
$ rpmlint hello.spec ../SRPMS/hello* ../RPMS/*/hello*
</pre>
 
If there are no warnings or errors, we've succeeded. Otherwise, use <code>rpmlint -i</code> or <code>rpmlint -I &lt;error_code&gt;</code> to see a more verbose description of the <code>rpmlint</code> diagnostics.
 
=== The <code>mock</code> builds ===
 
To check that the package build will succeed in the Fedora restricted build environment, check it with mock.  The default mock configuration builds the package against Rawhide - the Fedora development branch.
 
<pre>
$ mock --verbose ../SRPMS/hello-2.8-1.fc20.src.rpm
</pre>
 
== References ==


To check that the package build will succeed in the Fedora restricted
* [[How to create an RPM package]]
build environment, check it with mock. Your account needs to be in the
* [[Building RPM packages (20090405)]]
'mock' group for
* [[Using Mock to test package builds]]
mock to work, so you may need to do one-time setup, like so:
* [[Using the Koji build system]]


sudo usermod -a -G mock przemek
== History ==


and then run mock:
Przemek Klosowski wrote this tutorial when he worked through [[Building_RPM_packages_%2820090405%29|Christoph Wickert's IRC session on building RPMs]] using Rahul Sundaram suggestion of GNU "Hello World" as a test case. After he wrote up his experience, he found out about the excellent and extensive [[How to create an RPM package]] page on this wiki, as well as the [http://www.absolutepanic.org/blog/2009/07/building-a-gnu-hello-world-rpm Christian Lyder Jacobsen's website]. However, Christian isn't planning to update his site, and it seemed that a 5-minute 'fast food' alternative to the more extensive article might suit some people. More in-depth information on using and building RPM packages is available from [[Yum|other sources]].


mock -r fedora-12-i386 --rebuild ../SRPMS/hello-2.5-1.fc12.src.rpm
[[Category:Package Maintainers]][[Category:How to]]

Revision as of 14:36, 13 February 2015

This is a short hands-on tutorial on writing RPM files, showing how to quickly step up to create simple source and binary software packages. It assumes some familiarity with using pre-made RPM packages, and with the FOSS software building process.

For a comprehensive information on how to create RPM files, including more detailed tips, refer to How to create an RPM package. If you plan to create an RPM package for the Fedora repository, follow the process for How to join the Fedora Package Collection Maintainers, including following the various Fedora guidance.

This tutorial demonstrates packaging of the GNU "Hello World" project. While the C program printing 'Hello World" to standard output is trivial, the GNU version contains most of the usual peripheral components associated with a typical FOSS software project, including the configuration/build/install environment, documentation, internationalization, etc. The GNU version, however, traditionally consists of a tar file containing the source code and configure/make scripts, but it does not include the packaging information. Therefore, it's a reasonable vehicle to practice building RPMs on.

Development environment

To build RPMs we need a set of development tools. This is a one-time-only setup, installed by running those commands from a system administration (root) account:

# yum install @development-tools
# yum install fedora-packager

If you want to test the build procedure in a clean chroot, you need to configure your non-privileged account to be a member of the 'mock' group:

# usermod -a -G mock <your username>

Those are the only commands requiring root privileges. All the remaining work should be done from your regular, non-privileged account, or even from a separate account created just for development work. Modern RPM-based systems, including Fedora, are set up to build and test RPM packages purely from within a non-privileged account. The command

$ rpmdev-setuptree

sets up a RPM build area in your ~/rpmbuild. This directory will contain several subdirectories, for the project source code, RPM configuration files and for the resulting source and binary packages.

Building a "Hello World" RPM

We need the source code of the project we are packaging, often referred to as the 'upstream' source. We will download it from the project's website into the ~/rpmbuild/SOURCE directory. We are getting the compressed tarball archive, which happens to be a preferred distribution form for most FOSS projects.

$ cd ~/rpmbuild/SOURCES
$ wget http://ftp.gnu.org/gnu/hello/hello-2.8.tar.gz

The RPM package is configured by .spec files. We will create a template file hello.spec in the appropriate directory:

$ cd ~/rpmbuild/SPECS
$ rpmdev-newspec hello

Recent versions of Emacs and vi have .spec file editing modes which will also bring up a similar template upon creating a new file. So you can just use the following command for example to use the template automatically.

vi hello.spec

Inside a .spec file

The fields in our .spec file need slight editing. Please follow the Fedora rules for these fields. In our case, the file might start as follows:

Name:     hello
Version:  2.8
Release:  1
Summary:  The "Hello World" program from GNU
License:  GPLv3+
URL:      http://ftp.gnu.org/gnu/hello    
Source0:  http://ftp.gnu.org/gnu/hello/hello-2.8.tar.gz

%description
The "Hello World" program, done with all bells and whistles of a proper FOSS 
project, including configuration, build, internationalization, help files, etc.

%changelog
* Thu Jul 07 2011 The Coon of Ty <Ty@coon.org> - 2.8-1
- Initial version of the package

The Version should mirror upstream while Release numbers our work within Fedora.

The first letter of the Summary should be uppercase to avoid rpmlint complaints.

It is your responsibility to check the License status of the software, by inspecting the source files and/or their LICENSE files, and/or by talking to the authors.

The Group tag was historically used to classify the package in accordance to the list in /usr/share/doc/rpm-<version>/GROUPS. It is being phased out so you will not see it added by default. However, it doesn't hurt to add it anyway.

The %changelog should document the work on preparing the RPM, especially if there are security and bug patches included on top of the base upstream source. Changelog data can be displayed by rpm --changelog -q <packagename>, which is very useful for instance to find out if specific bug and security patches were included in the installed software, thanks to diligent Fedora packagers who include this info with the relevant CVE numbers.

The changelog entry should include the version string to avoid rpmlint complaints.

Multi-line sections like %changelog or %description start on a line under the directive, and end with an empty line.

Lines which aren't needed (e.g. BuildRequires and Requires) can be commented out with a hash ('#') for now.

Many lines in the template don't need to be changed at all in many cases, at least for the initial attempt.

Building the package

We are ready for the first run to build source, binary and debugging packages:

$ rpmbuild -ba hello.spec

It will complain and list the unpackaged files, i.e. the files that would be installed in the system that weren't declared as belonging to the package. We need to declare them in the %files section. Do not hardcode names like /usr/bin/, but use macros, like %{_bindir}/hello instead. The manual pages should be declared in the %doc subsection: %doc %{_mandir}/man1/hello.1.gz.

This is an iterative process: after editing the .spec file, rerun rpmbuild.

Since our program uses translations and internationalization, we are seeing a lot of undeclared i18 files. The recommended method to declare them is:

  • find the filenames in the %install step: %find_lang %{name}
  • add the required build dependencies: BuildRequires: gettext
  • use the found filenames %files -f %{name}.lang

If the program uses GNU info files, you need to make sure the installation and uninstallation of the package does not interfere with other software on the system, by using this boilerplate:

  • delete the 'dir' file in %install: rm -f %{buildroot}/%{_infodir}/dir
  • Requires(post): info and Requires(preun): info
  • add those steps:
%post
/sbin/install-info %{_infodir}/%{name}.info %{_infodir}/dir || :

%preun
if [ $1 = 0 ] ; then
/sbin/install-info --delete %{_infodir}/%{name}.info %{_infodir}/dir || :
fi

A complete hello.spec file

Here's the initial version of hello.spec:

Name:           hello
Version:        2.8
Release:        1%{?dist}
Summary:        The "Hello World" program from GNU

License:        GPLv3+
URL:            http://ftp.gnu.org/gnu/%{name}
Source0:        http://ftp.gnu.org/gnu/%{name}/%{name}-%{version}.tar.gz

BuildRequires: gettext
      
Requires(post): info
Requires(preun): info

%description 
The "Hello World" program, done with all bells and whistles of a proper FOSS 
project, including configuration, build, internationalization, help files, etc.

%prep
%autosetup

%build
%configure
make %{?_smp_mflags}

%install
%make_install
%find_lang %{name}
rm -f %{buildroot}/%{_infodir}/dir

%post
/sbin/install-info %{_infodir}/%{name}.info %{_infodir}/dir || :

%preun
if [ $1 = 0 ] ; then
/sbin/install-info --delete %{_infodir}/%{name}.info %{_infodir}/dir || :
fi

%files -f %{name}.lang
%{_mandir}/man1/hello.1.gz
%{_infodir}/%{name}.info.gz
%{_bindir}/hello

%doc
AUTHORS ChangeLog COPYING NEWS README THANKS TODO

%changelog
* Tue Sep 06 2011 The Coon of Ty <Ty@coon.org> 2.8-1
- Initial version of the package

With this spec file, you should be able to successfully complete the build process, and create the source and binary RPM packages.

Next you should check them for conformance with RPM design rules, by running rpmlint on the spec file and all RPMs:

$ rpmlint hello.spec ../SRPMS/hello* ../RPMS/*/hello*

If there are no warnings or errors, we've succeeded. Otherwise, use rpmlint -i or rpmlint -I <error_code> to see a more verbose description of the rpmlint diagnostics.

The mock builds

To check that the package build will succeed in the Fedora restricted build environment, check it with mock. The default mock configuration builds the package against Rawhide - the Fedora development branch.

$ mock --verbose ../SRPMS/hello-2.8-1.fc20.src.rpm

References

History

Przemek Klosowski wrote this tutorial when he worked through Christoph Wickert's IRC session on building RPMs using Rahul Sundaram suggestion of GNU "Hello World" as a test case. After he wrote up his experience, he found out about the excellent and extensive How to create an RPM package page on this wiki, as well as the Christian Lyder Jacobsen's website. However, Christian isn't planning to update his site, and it seemed that a 5-minute 'fast food' alternative to the more extensive article might suit some people. More in-depth information on using and building RPM packages is available from other sources.