Re: binNMU for gst0.10-python
On Tue, Feb 05, 2008 at 11:58:00AM +0100, Sebastian Dröge wrote:
> Am Montag, den 04.02.2008, 23:32 -0800 schrieb Steve Langasek:
> > On Wed, Jan 30, 2008 at 04:46:05PM +0100, Sebastian Dröge wrote:
> > > please schedule a binNMU for gst0.10-python on all archs against
> > > libgstreamer0.10-0 (>= 0.10.17) and libgstreamer-plugins-base0.10-0 (>=
> > > 0.10.17). The old version, 0.10.16, had an accidental ABI breakage.
> > Does this mean that version 0.10.17 restores ABI compatibility with 0.10.15
> > and earlier?
> In short: yes.
> The longer story is, that there was a ABI breakage in 0.10.14 that
> nobody noticed unfortunately. The last element(s) of two structs
> disappeared due to a stupid change that only happens when compiling with
> -DGST_DISABLE_DEPRECATED which was the default until 0.10.16. This broke
> nothing and since then everything seems to be rebuild.
> In 0.10.16 upstream started to not build releases with
> -DGST_DISABLE_DEPRECATED and the structs growed a bit again to the old,
> intended situation which broke applications. The fix in 0.10.17 is to
> have this struct fields removed in any case which of course is a ABI
> breakage compared to 0.10.13 and before... but as nobody ever noticed
> this seemed to be the correct decision.
I've scheduled binNMUs, but I have to say I don't agree this is the correct
decision. If the current libgstreamer0.10 in unstable is incompatible with
the not-rebuilt gst0.10-python in etch, then it's very possible that
packages will be installed-but-broken as part of a partial upgrade.
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer http://www.debian.org/