[FFmpeg-devel] [PATCH 1/2] fate: Add fate-ts-opus-demux

[Kieran Kunhya]

On Mon, 4 Apr 2016 at 01:43 Michael Niedermayer <michael at niedermayer.cc>
wrote:

> On Sun, Apr 03, 2016 at 04:02:55AM +0200, Michael Niedermayer wrote:
> > On Sun, Apr 03, 2016 at 12:20:59AM +0200, Michael Niedermayer wrote:
> > > On Sat, Apr 02, 2016 at 11:14:29PM +0200, wm4 wrote:
> > > > On Sat, 2 Apr 2016 18:01:47 -0300
> > > > James Almer <jamrial at gmail.com> wrote:
> > > >
> > > > > On 4/2/2016 5:53 PM, wm4 wrote:
> > > > > > On Sat,  2 Apr 2016 22:05:35 +0200
> > > > > > Michael Niedermayer <michael at niedermayer.cc> wrote:
> > > > > >
> > > > > >> Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
> > > > > >> ---
> > > > > >>  tests/fate/demux.mak         |    3 +
> > > > > >>  tests/ref/fate/ts-opus-demux |  514
> ++++++++++++++++++++++++++++++++++++++++++
> > > > > >>  2 files changed, 517 insertions(+)
> > > > > >>  create mode 100644 tests/ref/fate/ts-opus-demux
> > > > > >>
> > > > > >
> > > > > > Maybe you should wait with all these new tests until the codecpar
> > > > > > changes have been merged.
> > > > >
> > > > > Why? None of these new tests should generate conflicts when the
> > > > > codecpar branch is rebased to current git head. And they may even
> > > > > reveal new regressions once they are run afterwards.
> > > >
> > > > I'm sure they will reveal regressions, because michaelni seems to be
> > > > specifically hunting for them. When the branch gets merged, there
> > > > should be no failing fate tests, so adding such tests would hold back
> > > > the merge.
> > > >
> > > > While it's normally a good thing to catch such regressions (and to
> fix
> > > > them before a big thing is merged), the codecpar merge has been going
> > > > on for weeks. It's holding back merging of other important features
> > > > (like vaapi encoding to name a particularly popular one), and
> there's a
> > > > limit to which extend we can load further work on daemon and nev.
> > > >
> > > > So I suggest that such arguably less important tests/regressions are
> > > > applied and fixed later, so that merging can go on and none of the
> devs
> > > > have to die from stress.
> > >
> > > We can add the tests now and then just disable them in the merge
> > > if the people working on the merge belive that that is the best choice.
> > > It should make it easier to keep track of issues
> >
> > note, i intend to push these soon
> > if someone objects, please state so
>
> tested with and without codecpar, pass in both cases now
> (so this should not add any stress to anyone)
>
> applied
>
> Where does this 7.1 sample come from?
Kieran