Detecting soname bumps

[Kamil Paral]

> On Tue, 2012-12-04 at 23:24 -0500, Rahul Sundaram wrote:
> > Hi
> > 
> > On Tue, Dec 4, 2012 at 2:35 PM, Ville Skyttä <ville.skytta at iki.fi>
> > wrote:
> >         IMO if a maintainer of a shared lib package goes as far as
> >         submitting a
> >         koji build without noticing a soname bump in it, the
> >         maintainer should
> >         seriously consider stepping aside and finding someone else
> >         to
> >         take
> >         proper care of the package.
> > 
> > If we can convert runtime errors to buildtime errors by enforcing a
> > additional check, that's a good thing regardless of where the
> > problem
> > lies.   Question is, can rpmbuild do anything automatically instead
> > of
> > per spec changes?
> 
> This is the kind of thing that's traditionally been in rpmlint-type
> tools. It'd probably be trivial to add it to autoqa for e.g., as
> informative output on a bodhi update...

Are we talking about something like this:
http://autoqa.fedoraproject.org/results/471217-autotest/virt02.qa/rpmguard/results/libmemcached-1.0.14-.html
and this:
http://autoqa.fedoraproject.org/results/471217-autotest/virt02.qa/rpmguard/results/libmemcached-1.0.14-.html
?

We're doing very poor job announcing it, but we run the tool on every new package build in Koji.