Re: Interrupted builds: gs should remove broken output? [Re: [opensuse-contrib] lilypond]

[Dave Plater]

On 01/05/2010 04:10 PM, Jan Nieuwenhuizen wrote:
> Op dinsdag 05-01-2010 om 16:00 uur [tijdzone +0200], schreef Dave
> Plater:
>   
>> On 01/05/2010 03:16 PM, Jan Nieuwenhuizen wrote:
>>     
>>> Op dinsdag 05-01-2010 om 11:38 uur [tijdzone +0200], schreef Dave
>>> Plater:
>>>       
>   
>> "/usr/src/packages/BUILD/lilypond-2.12.3/out/lybook-db/snippet-names--7471402555679770406.ly"
>>
>> Can you make anything of this?
>>     
> Yes, look earlier in the log and you'll most probably see
> the gs segfault that I mailed and reported this morning.
>
> The trouble is that when gs fails it [may] still produce
> a .pdf which is not removed by gs or the current build system.
>
> Possibly our postscript->pdf or postscript->png functions
> should remove any leftovers of what they're building when
> they exit, although I think this is something that should
> be fixed in gs.
>
> Greetings,
> Jan.
>
>   
I made a log of the local rpmbuild of lilypond docs but I unfortunately
mixed up my redirectors and ended up with std error only in the rpmbuild
log, busy making a new one. The only real differences I've picked up so
far is that build service drops -g and -m64 from it's $RPM_OPT_FLAGS. Do
you get a successful build using my spec file with rpmbuild locally.
I've dropped all the patches and the doc tarball and you can comment out
"Source3: changelog" as it's not needed for "rpmbuild -bc". I have
openSUSE 11.2 x86_64 on my box but build service fails to compile the
docs on 11.0 through to factory. I've attached lilypond.spec.
Thanks
Dave P

lilypond.spec
Description: Text document