[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [PATCH 7/7] rules: Don't fail on new symbols.



On 2013-08-25 15:50 +0200, Emilio Pozuelo Monfort wrote:

> On 24/08/13 19:39, Sven Joachim wrote:
>> On 2013-08-22 10:35 +0200, Maarten Lankhorst wrote:
>> 
>>> Op 21-08-13 23:19, Jon Severinsson schreef:
>>>> ---
>>>>  debian/changelog |    1 +
>>>>  debian/rules     |    5 -----
>>>>  2 filer ändrade, 1 tillägg(+), 5 borttagningar(-)
>>>>
>>>> diff --git a/debian/changelog b/debian/changelog
>>>> index 7f554e49..e523baa2 100644
>>>> --- a/debian/changelog
>>>> +++ b/debian/changelog
>>>> @@ -7,6 +7,7 @@ mesa (9.2~rc1-2) UNRELEASED; urgency=low
>>>>    * libgl1-mesa(-swx11)-dev.install.in: Install the correct pkgconfig/gl.pc files.
>>>>    * rules: Add --disable-shared-glapi to the swx11-i386-i686 target.
>>>>    * rules: Enable the llvm shader compiler for r600g.
>>>> +  * rules: Don't fail on new symbols.
>>>>  
>>>>   -- Jon Severinsson <jon@severinsson.net>  Tue, 20 Aug 2013 18:00:00 +0200
>>>>  
>>>> diff --git a/debian/rules b/debian/rules
>>>> index 113c7838..24a3a161 100755
>>>> --- a/debian/rules
>>>> +++ b/debian/rules
>>>> @@ -274,12 +274,7 @@ binary-arch: install
>>>>  	dh_strip $(foreach p,$(nodbgpkg),-p$(p))
>>>>  	dh_compress -s
>>>>  	dh_fixperms -s
>>>> -ifeq (,$(filter sparc sparc64,$(DEB_HOST_ARCH)))
>>>> -	dh_makeshlibs -s -- -c4
>>>> -else
>>>> -	# Temporarily work around FTBFS on sparc, ignore new symbols:
>>>>  	dh_makeshlibs -s -- -c1
>>>> -endif
>>>>  	dh_installdeb -s
>>>>  	dh_shlibdeps -s
>>>>  	dh_gencontrol -s
>>> Why?
>> 
>> In mesa 7.10, libgles1-mesa would gain new symbols on sparc[1], that's why
>> KiBi introduced this hack.  Since mesa 7.11 this no longer seems to be
>> the case[2], so the workaround should be removed.
>
> The workaround here was to pass -c1 to dh_makeshlibs, so then we should do the
> opposite of your patch,

For the record, it wasn't my patch, but Jon's.

> i.e. always pass -c4 to dh_makeshlibs.

For some reason I thought he had done that, although the commit message
clearly says otherwise. :-/

I've pushed a fix that does what you suggested.

Cheers,
       Sven


Reply to: