Re: RFS: libgeier (updated package)

[Olaf Dietsche <olaf--mail.debian-mentors@olafdietsche.de>]

Uwe Kleine-König <ukleinek@strlen.de> writes:

> On Mon, Jul 11, 2011 at 02:17:55PM +0200, Olaf Dietsche wrote:
>> 
>> > P: libgeier-dev: no-upstream-changelog
>> > P: libgeier0: no-upstream-changelog
>> 
>>   Doesn't make sense, since upstream changelog is empty.
> So overwrite it.

I don't understand that. What do you mean with overwrite it?

>> > E: libgeier-dev: non-empty-dependency_libs-in-la-file usr/lib/libgeier.la
>> 
>>   AFAICT, this comes from depending on xmlsec1, which itself depends on
>>   these libraries.
> I didn't look into your package, but still having an empty
> dependency_libs entry is important for multiarch support. According to
> http://release.debian.org/~aba/la/current.txt (which is linked from
> http://lists.debian.org/debian-devel/2011/05/msg01003.html) xmlsec1 is
> OK, so you should really clean this entry in your libgeier.la.

I don't understand this either.

If xmlsec1 is ok and libtool copies these dependencies into libgeier.la,
why is it an error for libgeier to include this information, but not for
xmlsec1? If this is a bug for libgeier, libtool shouldn't copy these
dependencies in the first place.

And if an empty dependency_libs entry is important for multiarch
support, why is it important for libgeier, but not for xmlsec1?

>> > I: libgeier0: no-symbols-control-file usr/lib/libgeier.so.0.0.0
>> 
>>   Won't fix: Severity: wishlist
>> 
>> I also won't fix upstream warnings, since I think this is a separate
>> issue.
> IMHO this should be done for all serious library packages. If upstream
> doesn't care for API stability it's hardly possible, but having a
> symbols file is really worthwile.

Regards, Olaf