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

Bug#676433: nmu: kmymoney_4.6.2-3



Hi,

On Thu, 2012-06-07 at 00:07 +0200, Micha Lenk wrote:
> Am 06.06.2012 23:41, schrieb Adam D. Barratt:
> > On Wed, 2012-06-06 at 23:10 +0200, Micha Lenk wrote:
> >> there was a new upstream release that introduced a soname bump in libaqbanking.
> > 
> > That makes this a(n uncoordinated) transition, not a binNMU request.
> > Thankfully one that doesn't appear to clash with any other ongoing
> > transitions that I can see.
> 
> Sorry, you're right. I did not anticipate that correctly. Please accept
> my apologies.

No worries.  It looks like a rather small set of packages and
technically you were only requesting a single binNMU. :-)

> > Why do the -dev package names contain (and therefore change with) the
> > SONAME?  Given that they all Provide "libaqbanking-dev", could the -dev
> > package not simply be a concrete package named libaqbanking-dev?  That
> > also makes transitions much easier and avoids the need for source
> > uploads on each change (unless there are API changes which require the
> > source changes).
> 
> The package had a rather unstable API change history. For that reason I
> didn't dare to provide the real -dev package without the SONAME. But
> recently the API seems to have settled a bit, so I will consider to
> remove the SONAME from the package name at the next occasion (e.g. the
> next SONAME bump).

Okay, thanks for the explanation.

> >> The package kmymoney has a build-dep on 'libaqbanking33-dev |
> >> libaqbanking-dev'. For this reason I hope that a binNMU is enough as a short
> >> term solution to the problem.
> > 
> > It's not even a short term solution, I'm afraid.  The buildds will only
> > consider the first branch of an alternative build-dependency, which
> > means libaqbanking33-dev will always be installed.
> 
> Huh, I was afraid that might happen - now it really happened. If the
> binNMU really isn't sufficient I will raise the severity of #676430
> accordingly and do a NMU as needed.

It makes builds more reproducible, as you can guarantee which of the
alternatives will be installed.  Having said that, libaqbanking33-dev
will stay in unstable anyway until it's manually removed, which is
unlikely to happen until all the {build-,}dependencies have migrated (or
someone asks ftpmaster to remove it even though stuff breaks) so afaics
the result would be the same in any case.

Regards,

Adam




Reply to: