Help (voodoo, really) needed [Re: failed i386 build of iceweasel 11.0~b1-2]
On Mon, Feb 06, 2012 at 07:14:59PM +0100, Kurt Roeckx wrote:
> On Mon, Feb 06, 2012 at 09:25:19AM +0100, Mike Hommey wrote:
> > On Mon, Feb 06, 2012 at 05:16:16AM +0100, Kurt Roeckx wrote:
> > > On Sun, Feb 05, 2012 at 05:09:13PM +0100, Mike Hommey wrote:
> > > > Hi,
> > > >
> > > > This build error both makes no sense and can't be reproduced locally.
> > > > Could someone check what is going wrong there? (and/or respin a build)
> > >
> > > Given back.
> > Thanks. It happened again, though :(
> > The function that is on the failing line is declared as returning an int
> > in libasound2-dev, I fail to see how it could believe we're ignoring a
> > void value. I also fail to see how it could fail on the buildd while it
> > builds fine in a pbuilder here. It's not like the libasound headers have
> > changed spontaneously.
> So I see:
> sydney_audio_alsa.c: In function 'sa_stream_open':
> sydney_audio_alsa.c:156:3: warning: implicit declaration of function 'snd_pcm_set_params' [-Wimplicit-function-declaration]
> sydney_audio_alsa.c: In function 'sa_stream_write':
> sydney_audio_alsa.c:265:7: warning: implicit declaration of function 'snd_pcm_recover' [-Wimplicit-function-declaration]
> sydney_audio_alsa.c: In function 'sa_stream_set_volume_abs':
> sydney_audio_alsa.c:461:5: error: void value not ignored as it ought to be
> sydney_audio_alsa.c: In function 'sa_stream_get_volume_abs':
> sydney_audio_alsa.c:504:5: error: void value not ignored as it ought to be
> make: *** [sydney_audio_alsa.o] Error 1
> And I see an:
> #include <alsa/asoundlib.h>
> Which itself does:
> #include <alsa/pcm.h>
> #include <alsa/mixer.h>
> So it shouldn't give those warnings either. It's not
> making sense to me either.
This is getting annoying. For starters, the error makes no sense. I
also can't reproduce it locally. And it apparently happens reliably on
buildds (armel, sparc and i386).
The involved code has *not* changed since 9.0, and 10.0 still builds
fine, and as far as I can tell, 10 and 11 are built against the same
version of libasound-dev.
Would anyone have a clue as to what the hell is happening?