Bug#358065: mesa: FTBFS on hurd-i386 because of dri/drm

[Marcelo E. Magallon]

On Thu, May 04, 2006 at 06:14:13PM +0200, Michael Banck wrote:

 > the attached patch fixes this problem in the most elegant way I could
 > come up with given the make-me-harder attitude of mesa's
 > debian/rules.  If somebody has a better suggestion, I'd be happy to
 > know :)

 Whatever.  If don't understand why debian/rules is the way it is, keep
 your comments to yourself.  If you can come up with a better solution
 to the problem of packaging Mesa, by all means post a patch.

 > As the current version has reached testing some time ago, and a
 > missing mesa seems to be breaking a lot, I plan to NMU this in a
 > couple of days if there are no objections.

 The proposed patch is not acceptable on grounds of not scaling well.

 This in particular:

 > +ifeq (hurd-i386,$(DEB_HOST_ARCH))
 > +select_target_any = $(shell dh_listpackages -s | grep -v '^$$' | while read p ; do cut -d ' ' -f 1,3 debian/libdir.map | grep -w $$p ; done | grep -v _ | grep -v dri | cut -d ' ' -f 1 | sort -u | grep ^$(1)- | cut -d - -f 2-)
 > +else
 >  select_target_any = $(shell dh_listpackages -s | grep -v '^$$' | while read p ; do cut -d ' ' -f 1,3 debian/libdir.map | grep -w $$p ; done | grep -v _ | cut -d ' ' -f 1 | sort -u | grep ^$(1)- | cut -d - -f 2-)
 > +endif

 If you inspect debian/rules you'll see there not a single explicit
 mention of architectures in there.  I'd like it to stay that way.  In
 the past debian/rules was a mess of architecture special casing and I
 don't want to go back to that.

 If you can provide a comprehensive list of architectures where
 libgl1-mesa-dri SHOULD be built, I'll accept that.  If you are so
 inclined, moan about the lack of exclusion lists for the Architecture
 field with the dpkg developers.

 Is DRM linux specific or does it exist on FreeBSD, too?

 Marcelo