Re: libcroco and xz

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Sat, Jun 16, 2012 at 11:10:38PM +0000, David Holland wrote:
> On Sun, Jun 17, 2012 at 12:33:04AM +0200, Joerg Sonnenberger wrote:
>  > On Fri, Jun 15, 2012 at 07:31:40PM +0000, David Holland wrote:
>  > >  > The problem I have with the commit is that it also hides the reverse
>  > >  > case. A dependency is declared as build time only, used by a 
> dependency
>  > >  > as full dependency, but explicitly linked against. I'm more inclined 
> to
>  > >  > care about that case than the reverse.
>  > > 
>  > > I think that's less important. If it's an indirect dependency, it'll
>  > > always be present; the only way it'll fail at runtime is if an update
>  > > to libxml2 changes it to longer bring in xz. However, this will result
>  > > (if handled correctly) in xz being removed from xz's bl3.mk and
>  > > libcroco being revbumped and rebuilt, and then the liblzma NEEDED
>  > > entry will go away in the new build. IOW, it won't fail, so there's no
>  > > point worrying about it.
>  > 
>  > If we want to do that (I'm mildly opposed, but I agree that this will
>  > hit us a lot more often), I think the correct approach is to expand the
>  > list of full dependencies once. So move the pkg_info -r calls out of the
>  > AWK script and provide a second version of the .depends file that
>  > contains the full list of packages we are willing to accept shared
>  > linking against.
> 
> That sounds fine; however, can we have obache's version back first so
> we don't end up with pointlessly broken packages in 2012Q2?

Expanding the list obsoletes obache's change completely.

Joerg