Control: severity -1 normal

On 03-08-17 21:32, Steve Langasek wrote:
> I've confirmed that this autopkgtest is testing the same presets.xml which
> is installed by winff in the winff-data package, so it appears that this
> will genuinely cause winff to fail at runtime (i.e., this is not just a bug
> in the test).  I'm therefore marking it 'grave'.  Feel free to downgrade if
> you think this is wrong - it looks like only the presets for three Nokia
> devices are affected.

The autopkgtest is meant to catch changes in ffmpeg/libav. As it affects
only a very limited set of presets, and the user can relatively easy
work around this, I downgrade this to normal. Winff will not fail, the
conversion using this preset will fail.

This is more severe for Ubuntu though, as I do recognize, but for Debian
this is a normal bug.

> I would tag this bug 'sid', except that since autopkgtests don't gate
> testing in Debian, this autopkgtest regression does not stop the new ffmpeg
> from migrating to testing and breaking winff there.


> I don't know the fix for this issue, as I'm not conversant with ffmpeg and
> so am not sure how to map the existing asyncts filter settings to the
> recommended af_aresample filter.  There does not appear to be a fix upstream
> for this yet in the winff github repo.

The most trivial way to fix it quickly is by just dropping the argument.
I'll try to fix it smarter. Upstream is basically in maintenance only
mode. Debian is one of the main reasons for changes (I make them). This
ffmpeg change is such a change.


Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to