[PATCH:modular] build.sh: remove xf86-input-mouse from Linux

Gaetan Nadon memsize at videotron.ca
Mon Jan 6 14:02:47 PST 2014

On 14-01-06 09:56 AM, Trevor Woerner wrote:
> A couple days ago I took the list of projects available on fd.o, sorted
> them by their last change, and noticed there are a number of them
> missing from the list of modules built by build.sh which appear to still
> be active projects. For example: mesa/glu, mesa/glut, mesa/demos,
> mesa/mesa-test, cairo, and a bunch of others (I haven't finished yet).
> Do you think it would be worth my time to submit a patch to add some of
> these active projects into the list of modules built by build.sh?
Interesting work... It all depends on how one sees the "mission" of

My opinion is the script main goal is to build all of the modules needed
to have a working copy of X on any platform. There are probably as many
"lists" as there are people. The bare minimum set of modules is the one
xorg publishes on the web for a release. That's the bare minimum. Add to
that legacy apps, drivers and libraries, and you get what you have today.

The list has been crafted over time by developers who add/remove modules
based on some consensus. It's not an exact science and there is a good
error margin.

Here is the challenge: if we add a module that is not maintained, it
breaks and adds noise on xorg-devel. Same goes for an unstable module
during early development, or a module with external dependencies not
available on distros. Same for test or demo modules.

So, yes, it is best to propose modules on a case by case basis (or an
RFC). For the mesa modules you mentioned, I don't think they should be
included as they are not directly involved in running X out-of-the-box
(I could be wrong on this). Mesa is not an X project, but a dependency,
so only the bits needed by X should be included.

Looking at all fdo modules may not be the right approach. This would
yield too many modules unrelated to xorg. Note the path in the git URL:
xorg/... which indicates an xorg module. Dependencies have been added
over time such as XCB (which is now part of X) which I documented here:


X has many dependencies and some of them may coincidentally be an fdo
module. We would not add it to build.sh if it is commonly available on
distros. Again, to reduce the complexity of the build and chances of
build breaks.

My impression is that build.sh has been fairly well maintained over the
years. Recently I revamped the jhbuild xorg.modules list so both are in
sync now. All the dependencies are listed, so you can start from scratch
specifying just one module and everything needed will be pulled in.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.x.org/archives/xorg-devel/attachments/20140106/4293260c/attachment.html>

More information about the xorg-devel mailing list