Updated FFmpegDemuxerTest and FFmpegVideoDecoderTest to use the gmock-based MockFilterHost.

Updated FFmpegDemuxerTest and FFmpegVideoDecoderTest to use the gmock-based MockFilterHost.

Also deleted the old MockPipeline and MockFilterHost code as it has been completely deprecated.

BUG=16008
TEST=FFmpegDemuxerTest and FFmpegVideoDecoderTest

[scherkus (not reviewing), 2009-07-08 23:34:47]

Based on http://codereview.chromium.org/149350 (which I'll commit shortly)

Nothing too crazy here, fully deprecated the old MockFilterHost/MockPipeline and
switched things over to StrictMock<MockFilterHost> -- these tests are awesome!

[awong, 2009-07-08 23:38:30]

Die code! DIE! 
DDDDDIIIIIIIIIIIIIIIIIIIIIIIEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE!!!!!!!!!!!!!!!!!!!!!

*ahem*  I'm calm now.

http://codereview.chromium.org/149366/diff/1/4
File media/filters/ffmpeg_demuxer_unittest.cc (left):

http://codereview.chromium.org/149366/diff/1/4#oldcode258
Line 258: // Since we ignore data streams, the duration should be equal to the
longest
Do don't need an quivalent EXPECT somewhere?

[awong, 2009-07-08 23:38:52]

Oh, and uh, LGTM.

On 2009/07/08 23:38:30, awong wrote:
> Die code! DIE! 
>
DDDDDIIIIIIIIIIIIIIIIIIIIIIIEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE!!!!!!!!!!!!!!!!!!!!!
> 
> *ahem*  I'm calm now.
> 
> http://codereview.chromium.org/149366/diff/1/4
> File media/filters/ffmpeg_demuxer_unittest.cc (left):
> 
> http://codereview.chromium.org/149366/diff/1/4#oldcode258
> Line 258: // Since we ignore data streams, the duration should be equal to the
> longest
> Do don't need an quivalent EXPECT somewhere?

[scherkus (not reviewing), 2009-07-08 23:40:30]

http://codereview.chromium.org/149366/diff/1/4
File media/filters/ffmpeg_demuxer_unittest.cc (left):

http://codereview.chromium.org/149366/diff/1/4#oldcode258
Line 258: // Since we ignore data streams, the duration should be equal to the
longest
On 2009/07/08 23:38:30, awong wrote:
> Do don't need an quivalent EXPECT somewhere?

this was moved up into the InitializeDemuxer() method as a gmock expectation
(every time we have a successful demuxer initialization, we expect them to set
the duration, hence the expectation is set every time)

[Alpha Left Google, 2009-07-08 23:51:57]

I have two tiny questions. :)

http://codereview.chromium.org/149366/diff/1/4
File media/filters/ffmpeg_demuxer_unittest.cc (right):

http://codereview.chromium.org/149366/diff/1/4#newcode146
Line 146: }
Can we expect PIPELINE_OK here? Or because that we have initialized that we
don't care?

http://codereview.chromium.org/149366/diff/1/5
File media/filters/ffmpeg_video_decoder_unittest.cc (right):

http://codereview.chromium.org/149366/diff/1/5#newcode120
Line 120: scoped_refptr<StrictMock<MockFFmpegDemuxerStream> > demuxer_;
why this mock is scoped_refptr but unlike host_ be a member?

[scherkus (not reviewing), 2009-07-08 23:59:05]

http://codereview.chromium.org/149366/diff/1/4
File media/filters/ffmpeg_demuxer_unittest.cc (right):

http://codereview.chromium.org/149366/diff/1/4#newcode146
Line 146: }
On 2009/07/08 23:51:57, Alpha wrote:
> Can we expect PIPELINE_OK here? Or because that we have initialized that we
> don't care?

since we use a StrictMock but don't set an expectation on Error(), it's
impossible for PIPELINE_OK to get changed without the test failing

therefore we infer that everything is still ok :)

http://codereview.chromium.org/149366/diff/1/5
File media/filters/ffmpeg_video_decoder_unittest.cc (right):

http://codereview.chromium.org/149366/diff/1/5#newcode120
Line 120: scoped_refptr<StrictMock<MockFFmpegDemuxerStream> > demuxer_;
On 2009/07/08 23:51:57, Alpha wrote:
> why this mock is scoped_refptr but unlike host_ be a member?

The base MediaFilter interface is ref-counted, therefore all subclasses
(including the mocks) need to be destroyed via refcounts

The FilterHost interface isn't ref-counted, so we can get away with a statically
declared version

[awong, 2009-07-09 00:00:22]

On Wed, Jul 8, 2009 at 4:59 PM, <scherkus@chromium.org> wrote:

>
> http://codereview.chromium.org/149366/diff/1/4
> File media/filters/ffmpeg_demuxer_unittest.cc (right):
>
> http://codereview.chromium.org/149366/diff/1/4#newcode146
> Line 146: }
> On 2009/07/08 23:51:57, Alpha wrote:
>
>> Can we expect PIPELINE_OK here? Or because that we have initialized
>>
> that we
>
>> don't care?
>>
>
> since we use a StrictMock but don't set an expectation on Error(), it's
> impossible for PIPELINE_OK to get changed without the test failing
>
> therefore we infer that everything is still ok :)
>
> http://codereview.chromium.org/149366/diff/1/5
> File media/filters/ffmpeg_video_decoder_unittest.cc (right):
>
> http://codereview.chromium.org/149366/diff/1/5#newcode120
> Line 120: scoped_refptr<StrictMock<MockFFmpegDemuxerStream> > demuxer_;
> On 2009/07/08 23:51:57, Alpha wrote:
>
>> why this mock is scoped_refptr but unlike host_ be a member?
>>
>
> The base MediaFilter interface is ref-counted, therefore all subclasses
> (including the mocks) need to be destroyed via refcounts
>
> The FilterHost interface isn't ref-counted, so we can get away with a
> statically declared version


I ran into this issue gmocking other classes. It's annoying, but oh well.