Tino Schwarze wrote:
> On Fri, Jul 25, 2003 at 08:37:24AM +0200, David Neary wrote:
> > CVS does not support specifying interdependencies - it
> > knows nothing about file structure. It versions files, period.
> > Luckily, that's all it has to do :) We use good old text files to
> > specify interdependencies.
>
> There is actually a "good old text file" in CVSROOT: modules. You can
> specify aliases for modules there and also group several modules
> together. It needs careful checking though - it's easy to mess things up
> this way.
>
> If the modules file is set up correctly, you can even run
> cvs -c
> to get a list of available modules. This is very useful IMO.
> Ah, I see, it actually works for GNOME cvs:
> cvs -d :pserver:anonymous@xxxxxxxxxxxxxxxx:/cvs/gnome co -c
>
> There's no gimp-gap nor gimp-perl though.
Up until pretty recently, GNOME CVS had a habit of creating a
module as an alias for a directory... I suppose this was so that
the aforementioned cvs co -c would work... there is actually no
way to know what directories exist in a cvs repository other than
looking, I think. Either with a gui tool, or with a file browser.
I may be wrong, though.
Anyway, listing the modules in GNOME CVS doesn't get us any further
along the way towards knowing what modules the gimp HEAD depends
on, which is what I understood of the question. Although I guess
(now that I'm reading the info file) tyhat it would be possible
to create a gimp_container module which had the following in the
modules file...
gimp_container &gimp >k+ &glib &pango &atk &libart_lgpl etc...
But then how could one specify a particular branch to be taken,
in the case whete the GIMP doesn't build against HEAD? I'm
guessing that we're going to build against GTK+ 2.2 from now on,
if we're in feature freeze, that means not taking the HEAD branch
from several modules.
Cheers,
Dave.
--
David Neary,
Lyon, France
E-Mail: bolsh@xxxxxxxx
