Contents
Check this for how to compile Evolution from SVN.
Why?
I noted that a lot GNOME bounties involve altering softwares like Evolution. In order to do that, you will need to get that beast compiled first of course. I assume a lot people have been scared away cause of that first task. However, if you know what and how, building it ain't really hard at all. So thats why I've setup this little guide.
So this document is for development purposes. Normal users (who aren't planning to start developing Evolution) shouldn't be reading this. Really. I'm serious about this! You will probably not understand a single thing about this documentation if you don't have a strong Linux-development background!
First!
Stop reading this document, go to this url, and read all of it. Yes all of it. You can skip "Building Evolution". It explains how to build Evolution using jhbuild. Jhbuild is nice but a.t.m. I find the current version of that documentation rather clumbsy. Then return here and continue.
Okay
okay, lets go
Platforms we have tried (feel free to add comments about the same procedures for your platform to this document)
Fedora Core 3
Ubuntu Warty and Hoary
Debian sarge
Gentoo
Prerequisites
sudo make install magic
For the 'sudo make install'-magic, put the following in your /etc/sudoers file. But don't do this on a system with multiple users or where security is of any importance.
username ALL = NOPASSWD:/usr/bin/make
distcc
Recommended if you have multiple hosts. Install ccache and distcc. If you do, add -j# as a parameter to each make-command. A good value for # is the amount of CPU's that will do the job multiplied by two.
export CC="ccache distcc" export DISTCC_HOSTS="host1withdistcc host2withdistcc localhost"
If you copy-paste from this wiki-page, you can use the $CPUS variable for this
export CPUS="-j6"
Important note: You'll need to rebuild the modules from scratch (redo the configure/autogen.sh steps) you want to switch from using distcc (with optionally ccache) to using a normal C-compiler like gcc. Also note that you need the same version of compiler on each host, or follow the steps about using different compilers (checkout the webpages of distcc for more about this). You can't mix (link them) for example gcc 3.3 object files with gcc 3.2 ones or gcc 3.3 ones with gcc 4.0 ones. It will fail.
on Debian sarge
apt-get install gnome-core-devel libnss-dev cvs distcc [cvs openssh]
On Fedora
Here are some -devel packages that where not installed on my system. Note that the configure-script of evolution-data-server will not warn you about libgcrypt. But the make step will fail if you don't have -lgcrypt. I already posted a bug about this on Ximians bugzilla. (note that this might be unrelated to evolution-data-server, I have not investigated this tiny issue in depth).
yum install libsoup-devel libgcrypt-devel [cvs openssh]
On Ubuntu
Simply run the following command(s):
[sudo apt-get install cvs openssh]
sudo apt-get build-dep evolution
On Gentoo
Simply run one of the following commands :
emerge -va gnome-light nss cvs
or
emerge -va gnome nss cvs
With gnome-light you get the only-needed stuff, like gnome-core on other dists. On Gentoo, gnome-core refers to Gnome v1.4. You'd probably save a ton of time doing so, emerging Gnome takes Time! (personally i did emerge gnome, just because i run gnome, but to others who dont, i guess gnome-light is sufficient)
you could also
emerge gnome-common
if you dont want to grab it from cvs. Though grabbing it from cvs is recomended. Then you get the most up-to-date version.
Building
Before the actual build takes place
Prepare a build-directory
mkdir -p ~/cvs/gnome cd ~/cvs/gnome
If you don't have a cvs account on gnome (If you hack on Evolution, they will give you one sooner or later)
export CVSROOT=:pserver:anonymous@anoncvs.gnome.org:/cvs/gnome cvs login Password: __press enter__
or if you have a CVS account on gnome
export CVS_RSH=ssh export CVSROOT=:ext:username@cvs.gnome.org:/cvs/gnome
Getting the sources
cvs -z3 co gnome-common gtkhtml gal evolution-data-server evolution
You'll also need the gnome-2-10 branch of the libsoup module
cvs -z3 co -r gnome-2-10 libsoup
Doing gnome-common
Note that most distributions (like Debian and Ubuntu for sure) have gnome-common available as a package. Often you don't have to overwrite this package for it's version is recent enough. If you still want to utilise the latest version of gnome-common while also having it installed as a package, consider installing it in a prefix other than /usr.
pushd gnome-common ./autogen.sh --prefix=/usr make $CPUS make install popd
Doing gal
(Needed parts of gal has been merged into the Evolution tree so this should not be needed)
pushd gal ./autogen.sh --prefix=/opt/evo/ make $CPUS sudo make install popd
Doing libsoup
pushd libsoup PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ ./autogen.sh --prefix=/opt/evo/ make $CPUS sudo make install popd
Doing gtkhtml
pushd gtkhtml PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ ./autogen.sh --prefix=/opt/evo/ make $CPUS sudo make install popd
Doing evolution-data-server
pushd evolution-data-server
Fedora Core 3 & Ubuntu
PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ ./autogen.sh --prefix=/opt/evo/
Debian sarge (this or install libnss and use the above method)
PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ ./autogen.sh --prefix=/opt/evo/ \ --with-nspr-includes=/usr/include/mozilla/nspr/
Then:
make $CPUS sudo make install popd
Having problems with a missing calendar/libical directory?
For some reason this directory didn't automatically get checked by with the previous commands for me. To solve this execute the following commands from the top of your e-d-s build tree
cd calendar cvs co libical
Doing evolution
pushd evolution PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ ./autogen.sh --prefix=/opt/evo/ \ --with-openldap=yes \ --enable-nntp=yes --enable-ipv6=yes --enable-test-component=yes \ --enable-nss=yes --enable-smime=yes --enable-plugins=all make $CPUS sudo make install popd
You're done
Optionally: Doing evolution-exchange
pushd evolution-exchange PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ ./autogen.sh --prefix=/opt/evo/ make $CPUS sudo make install popd
Update bonobo-activation-config.xml
Search for the file bonobo-activation-config.xml. Usually it's installed at /etc/bonobo-activation/bonobo-activation-config.xml. You'll need to add the location where your evolution-data-server binary is installed.
<?xml version="1.0"?>
<oafconfig>
<searchpath>
< ! - - Don't remove existing items - - >
<item>/opt/evo/lib/bonobo/servers/</item>
</searchpath>
</oafconfig>
Create a small wrapper-script
Copy it to, for example, /usr/local/bin/evolution-cvs or in /opt/evo/bin and put that path in your PATH environment variable.
You'll need GNU versions of tools like ps, pidof, bash or a sh that will do these scripts correctly, gdb, ddd, grep and cut. To use these scripts, it's important that you've succesfully build evolution from source. Also note that I did hardcoded some source-code filelocations in the scripts. For example $HOME/cvs/gnome/evolution/evolution-plugin.pc. If your checkout from cvs is differently located, you'll need to adjust these paths in all scripts you'll find here.
#! /bin/sh
prefix=`grep "^prefix=" $HOME/cvs/gnome/evolution/evolution-plugin.pc | cut -d "=" -f 2`
version=`grep "BASE_VERSION=" $HOME/cvs/gnome/evolution/configure.in | cut -d "=" -f 2`
function strstr () {
[ ${#2} -eq 0 ] && { echo "$1" ; return 0; }
case "$1" in
*$2*) ;;
*) return 1;;
esac
first=${1/$2*/}
echo "${1##$first}"
}
substr=`strstr "$prefix/lib/bonobo/servers" $BONOBO_ACTIVATION_PATH`
if test x"$substr" != x"$prefix/lib/bonobo/servers"
then
killall -9 bonobo-activation-server
# in case the file bonobo-activation-config.xml is incorrect
export BONOBO_ACTIVATION_PATH=$prefix/lib/bonobo/servers:$BONOBO_ACTIVATION_PATH
fi
substr=`strstr "$prefix/lib" $LD_LIBRARY_PATH`
if test x"$substr" != x"$prefix/lib"
then
export LD_LIBRARY_PATH=$prefix/lib:/usr/lib/evolution/$version/:/usr/lib/mozilla-firefox/:$LD_LIBRARY_PATH
fi
substr=`strstr "$prefix/bin/:$prefix/libexec/" $PATH`
if test x"$substr" != x"$prefixbin/:$prefix/libexec/"
then
export PATH=$prefix/bin/:$prefix/libexec/:$PATH
fi
$prefix/bin/evolution-$version $1 $2 $3 $4 $5ps. You can download it here.
Upgrading
When upgrading your build I'm assuming you want to get the very latest version thats available in CVS. Beware that it WILL be fscking unstable. But then again, real developers like it unstable for then they can fix those bugs. Right? No?
The rule of thumb is that if new files are added and/or if configure.in or configure.ac is modified, you'll probably need to make distclean the module and start building it from scratch. But it highly depends. An experienced GNOME programmer knows when there's a need to recompile a module from scatch.
cd ~/cvs/gnome/
ps. You can download a script that contains all thats below here.
It will update your sources using cvs and check for errors during the building. If an error happens, it will retry by reconfiguring the build-environment (configure/autogen.sh steps).
Upgrading gal
Upgrading gal is hardly ever needed
pushd gal cvs update -d make $CPUS && sudo make install || (make distclean; ./autogen.sh --prefix=/opt/evo/; make $CPUS && sudo make install) popd
Upgrading libsoup
Upgrading libsoup is harmly ever needed
pushd libsoup cvs update -d make $CPUS && sudo make install || (make distclean; PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ ./autogen.sh --prefix=/opt/evo/; make $CPUS && sudo make install) popd
Upgrading gtkhtml
Upgrading gtkhtml is sometimes interesting. Not always. And updating it might make your Evolution look very unstable. Often it's a simple bug in gtkhtml.
pushd gtkhtml cvs update -d make $CPUS && sudo make install || (make distclean; PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ ./autogen.sh --prefix=/opt/evo/; make $CPUS && sudo make install) popd
Upgrading evolution-data-server
You want to check for upgrades on evolution-data-server each time you're going to upgrade. As time moves on, lots of the functionality is actually implemented in evolution-data-server and simply being called by evolution using ORBit-2. Most of the times it's functionality that happens in the background or non userinterface related functionality. But there are plans to put some userinterface related things in e-d-s.
The desktop-integration efforts of the last few years caused that a few applications are also using e-d-s. Examples are recent versions of gnomemeeting and the panelapplet Clock. Note that if you make your e-d-s unstable, by installing a bleeding edge CVS version, you're also making other parts of your desktop unstable. But then again, and I repeat myself, thats the way developers like it!
pushd evolution-data-server cvs update -d make $CPUS && sudo make install || (make distclean; PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ ./autogen.sh --prefix=/opt/evo/; make $CPUS && sudo make install) popd
Upgrade evolution
Since you're going to Upgrade, I'm guessing you want to upgrade evolution itself.
pushd evolution cvs update -d make $CPUS && sudo make install || (make distclean; PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ \ ./autogen.sh --prefix=/opt/evo/ --with-openldap=yes \ --enable-nntp=yes --enable-ipv6=yes --enable-test-component=yes \ --enable-nss=yes --enable-smime=yes --enable-plugins=all; make $CPUS && sudo make install) popd
Optionally: Upgrade evolution-exchange
pushd evolution-exchange cvs update -d make $CPUS && sudo make install || (make distclean; PKG_CONFIG_PATH=/opt/evo/lib/pkgconfig/ \ ./autogen.sh --prefix=/opt/evo/; make $CPUS && sudo make install) popd
Running
Normal operation
evolution-cvs
Debugging
I use these two scripts to attach debuggers to running evolution and evolution-data-server processes.
Debugging a running Evolution process
Copy it to, for example, /usr/local/bin/debug-evolution or in /opt/evo/bin and put that path in your PATH environment variable.
#! /bin/sh
#debugger=ddd
debugger=gdb
prefix=`grep "^prefix=" $HOME/cvs/gnome/evolution/evolution-plugin.pc | cut -d "=" -f 2`
version=`grep "BASE_VERSION=" $HOME/cvs/gnome/evolution/configure.in | cut -d "=" -f 2`
if pidof evolution-$version>/dev/null
then
$debugger $prefix/bin/evolution-$version `ps --no-headers -o pid -C evolution-$version`
else
$debugger $prefix/bin/evolution-$version
fips. You can download it here.
Debugging a running evolution-data-server process
Copy it to, for example, /usr/local/bin/debug-evolution-data-server or in /opt/evo/bin and put that path in your PATH environment variable.
#! /bin/sh
#debugger=ddd
debugger=gdb
prefix=`grep "^prefix=" $HOME/cvs/gnome/evolution/evolution-plugin.pc | cut -d "=" -f 2`
version=`grep "EDS_PACKAGE=" $HOME/cvs/gnome/evolution/configure.in | cut -d "=" -f 2`
if pidof evolution-data-server-$version>/dev/null
then
$debugger $prefix/libexec/evolution-data-server-$version `ps --no-headers -o pid -C evolution-data-server-$version`
else
$debugger $prefix/libexec/evolution-data-server-$version
fips. You can download it here.
Create launchers on your panel
I probably don't have to explain how to do this. Nevertheless it's a great idea to force yourself to actually sometimes do a debugging session when that fscking Evolution from cvs is crashing and being very unstable.
~/.gnome2/panel2.d/default/launchers $ cat moe-005712dd25.desktop [Desktop Entry] Encoding=UTF-8 Version=1.0 Type=Application Exec=gnome-terminal -x debug-evolution TryExec= Icon=gnome-spider.png X-GNOME-DocPath= Terminal=false Name=Debug evolution GenericName=Debug evolution Comment=Debug evolution ~/.gnome2/panel2.d/default/launchers $
After starting the debugging session
gdb
If you've set the debugger-variable in the scripts to gdb, you'll have to type "cont" on the prompt in the new terminal. You can press CTRL+C at any time to interrupt the process you're debugging. Once interrupted you can set new breakpoints using "break functionname". You might want to teach yourself how to use gdb from the commandline.
ddd
A graphical userinterface for gdb is ddd. Ddd will show you the sourcecode and you can in the text-viewer set breakpoints and study whats currently in the stack in a graphically fasion.
Others
There's other debuggers that can be used. A very good example, when you're seeking memory problems and memory-leaks, is valgrind.
Making it your default E-mail client
gconftool-2 --set --type string /desktop/gnome/url-handlers/mailto/command "evolution-cvs %s"