[FFmpeg-devel] GSoC Weely report (libswscale)

[Pedro Arthur]

I'll check it, I think most of these asserts are from some unused variables
which were replaced by the SwsSlice struct.

2015-07-11 23:51 GMT-03:00 Michael Niedermayer <michael at niedermayer.cc>:

> On Sat, Jul 11, 2015 at 04:57:22PM -0300, Pedro Arthur wrote:
> > Here is the full patch rebased including all previous changes.
>
> if you build ffmpeg with --assert-level=2
>
> fails fate
> for example in fate-vsynth1-cinepak
>
> its probably best to always build with --assert-level=2
> so you wont miss such failures
>
> --- ./tests/ref/vsynth/vsynth1-cinepak  2015-07-11 12:25:01.268257903 +0200
> +++ tests/data/fate/vsynth1-cinepak     2015-07-12 04:11:06.041453781 +0200
> @@ -1,4 +0,0 @@
> -546c7c1069f9e418aa787f469b693b94 *tests/data/fate/vsynth1-cinepak.mov
> -99465 tests/data/fate/vsynth1-cinepak.mov
> -bee091c200262be3427a233a2812388c
> *tests/data/fate/vsynth1-cinepak.out.rawvideo
> -stddev:    8.46 PSNR: 29.58 MAXDIFF:  105 bytes:  7603200/   456192
>
> Assertion lumSrcPtr + vLumFilterSize - 1 < (const int16_t **)lumPixBuf +
> vLumBufSize * 2 failed at libswscale/swscale.c:694
>
> there are also many other fate tests which fail with --assert-level=2
>
> also it might make more sense to stash all these commits together
> for submission,
> if you like you could keep them seperate in your branch (could make
> sense for debug/bisect)
>
> but the commits showing the step wise development are rather hard to
> review and would be impossible to push to master git
> commits for master git should not introduce known issues that are
> fixed in later commits and should be split in selfcontained changesets
>
> a single stashed together patch should be fine and easy to generate
>
> thanks
>
> [...]
>
> --
> Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
>
> Avoid a single point of failure, be that a person or equipment.
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
>