Fix FFmpegAudioDecoder monotonically increasing timestamp check to ignore kNoTimestamp()

Fix FFmpegAudioDecoder monotonically increasing timestamp check to ignore kNoTimestamp()


BUG=142738
TEST=Cr142738/FFmpegRegressionTest.BasicPlayback/0

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=151800

[acolwell GONE FROM CHROMIUM, 2012-08-15 00:10:06]

The clip in the bug report emits a buffer w/ kNoTimestamp() for the last buffer
of the stream. Digging further into the FFmpeg code may be necessary to
determine why, but this minor fix allows the clip to play properly, should be
harmless, and doesn't require an FFmpeg roll. :)

[scherkus (not reviewing), 2012-08-15 00:24:18]

can we turn that manual test into an automated test somehow? (i.e., ffmpeg
regression test that expects a decode error so if it does indeed get fixed we'll
know about it?)

[acolwell GONE FROM CHROMIUM, 2012-08-15 16:30:05]

On 2012/08/15 00:24:18, scherkus wrote:
> can we turn that manual test into an automated test somehow? (i.e., ffmpeg
> regression test that expects a decode error so if it does indeed get fixed
we'll
> know about it?)
I'd have to add an FFmpegDemuxer specific test that just checks timestamps since
this is really just about the timestamps FFmpeg is sending us. If this gets
fixed in FFmpeg we just won't get the buffer w/ kNoTimestamp(). The pipeline
shouldn't signal an error in either case.

[acolwell GONE FROM CHROMIUM, 2012-08-15 21:57:47]

PTAL... now w/ a FFmpegRegression test.

[scherkus (not reviewing), 2012-08-15 21:58:23]

lgtm!!!

thanks!

[commit-bot: I haz the power, 2012-08-15 21:59:28]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/acolwell@chromium.org/10857010/3

[commit-bot: I haz the power, 2012-08-16 00:24:24]

Change committed as 151800