On Thu, May 13, 2010 at 03:47:59AM +0100, Olly Betts wrote:
> On Wed, May 12, 2010 at 07:26:04PM -0700, tony mancill wrote:
> > Jari made the change because I was getting failures trying to build the
> > package inside a chroot environment, but it didn't affect the build. I
> > ended up unsetting my DISPLAY before the build. I agree that it would be
> > better not to disable the entire test suite; I'll test with your patch.
>
> Hmm, it's not good if the package fails to build if DISPLAY is set.
>
> It's hard to know what is happening without seeing the error you got, but my
> best guess is that DISPLAY was set to a value which doesn't work in the build
> environment. If that was the problem, then the change in the NMU isn't going
> to help at all, and neither will my patch.
OK, I've just tested, and my guess above was essentially correct.
But the change in the NMU does actually help, but only because it isn't doing
what was intended:
- $(MAKE) check
+ [ ! "$DISPLAY" ] || $(MAKE) check
Since this is a makefile, $DISPLAY is $D followed by the literal text ISPLAY so
the shell sees:
[ ! "ISPLAY" ] || $(MAKE) check
Which will simply never run the tests.
The best way to address this is probably just to ensure DISPLAY isn't set, so
we don't try to use an inherited value of DISPLAY which doesn't work in the
chroot-ed environment:
- $(MAKE) check
+ $(MAKE) check DISPLAY=
I'm tempted to do a general spring clean of the package - there's another open
bug and a few lintian warnings.
Wookey: OK to add myself as co-maintainer and do that?
Cheers,
Olly
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
