[FFmpeg-cvslog] Merge commit '2e06758479650f6e2a8820c7105f2d193a701175' into release/1.1

Reimar Döffinger Reimar.Doeffinger at gmx.de
Sun May 12 12:23:08 CEST 2013


On Sun, May 12, 2013 at 12:13:36PM +0200, Michael Niedermayer wrote:
> ffmpeg | branch: release/1.1 | Michael Niedermayer <michaelni at gmx.at> | Sun May 12 12:02:26 2013 +0200| [50a3442120e685963649417a6455ce90cf45b653] | committer: Michael Niedermayer
> 
> Merge commit '2e06758479650f6e2a8820c7105f2d193a701175' into release/1.1
> 
> * commit '2e06758479650f6e2a8820c7105f2d193a701175':
>   Prepare for 9.5 Release
>   update Changelog
>   add missed CVE reference in 9.2 release
>   fate: fetch samples that match the release series
> 
> Conflicts:
> 	Changelog
> 	RELEASE
> 	tests/Makefile
> 
> The rsync change is not merged
> We need to maintain the ability to checkout and test old revissions
> from master. This implies that the default sample repository has the
> needed samples for both older and newer revissions. Thus there is no
> need for a seperate one for each release.
> Comments & Suggestions of course welcome

Which tests are affected anyway?
E.g. nothing should change for H.264 (and admittedly the references are
in git anyway), for audio codecs that do not specify behaviour we should
be a method that allows being off (and how much they are allowed to be
off is stored in git). I very much think that we should not generally
remove or change samples, but if we do it is still possible to adapt the
branches accordingly.
So yes, I don't really know why there would be any advantage in having
a fate checkout different for each release?


More information about the ffmpeg-cvslog mailing list