On Tue, Aug 07, 2012 at 11:07:26PM +0200, Diego Biurrun wrote: > On Mon, May 14, 2012 at 08:42:18PM +0100, Måns Rullgård wrote: > > Alex Converse <[email protected]> writes: > > > On Fri, May 11, 2012 at 10:35 AM, Mans Rullgard <[email protected]> wrote: > > >> > > >> These files all have pcm audio which is tested elsewhere. > > >> > > >> Signed-off-by: Mans Rullgard <[email protected]> > > >> --- > > >> tests/fate/demux.mak | 3 -- > > >> tests/fate/qt.mak | 2 +- > > >> tests/fate/screen.mak | 4 +- > > >> tests/fate/video.mak | 25 +++++---- > > >> tests/ref/fate/8bps | 24 --------- > > >> tests/ref/fate/armovie-escape124 | 5 -- > > >> tests/ref/fate/bethsoft-vid | 72 ------------------------- > > >> tests/ref/fate/bfi | 58 -------------------- > > >> tests/ref/fate/corepng | 20 ------- > > >> tests/ref/fate/creatureshock-avs | 37 ------------- > > >> tests/ref/fate/cyberia-c93 | 5 -- > > >> tests/ref/fate/dxtory | 1 - > > >> tests/ref/fate/id-cin-video | 53 ------------------ > > >> tests/ref/fate/siff | 4 -- > > >> tests/ref/fate/sp5x | 10 ---- > > >> tests/ref/fate/tiertex-seq | 78 --------------------------- > > >> tests/ref/fate/tmv | 111 > > >> -------------------------------------- > > >> tests/ref/fate/tscc-15bit | 16 ------ > > >> tests/ref/fate/wc3movie-xan | 36 ------------- > > >> 19 files changed, 17 insertions(+), 547 deletions(-) > > > > > > This seems to substantially lower our coverage for demuxers. In these > > > cases the code that demuxes the audio is still executed but now we no > > > longer get to see the timestamps or that we are grabbing the correct > > > bytes for each audio frame. What's the motivation for this? I under > > > stand trying to test features in isolation, but "decoding" linear PCM > > > audio isn't much more work than just copying, sometimes with a > > > byteswap. > > > > The ultimate goal is to make it possible to test even a very minimal > > configuration. How about adding demux-only tests for those of these > > containers that are not fully covered otherwise? > > I'm with Mans here - the long-term goal is to be able to test any config, > not just the full-featured ones. This will require making changes like > this one here and there. Sometimes you have to pass through a little > valley to reach the next summit.
Can we revisit this please? The patchset should go in, it will be long-term beneficial. Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
