Fix ChunkDemuxer so it properly outputs buffered ranges.

media/filters/chunk_demuxer.h

Index: media/filters/chunk_demuxer.h
diff --git a/media/filters/chunk_demuxer.h b/media/filters/chunk_demuxer.h
index 5945526aae61b2f58d86a2a62b8d1601ea0ae907..6cab74a6ac71a43e2291a2dc8782a1462836620d 100644
--- a/media/filters/chunk_demuxer.h
+++ b/media/filters/chunk_demuxer.h
@@ -125,6 +125,13 @@ class MEDIA_EXPORT ChunkDemuxer : public Demuxer {
   bool CopyIntoRanges(
       const SourceBufferStream::TimespanList& timespans,
       Ranges* ranges_out) const;
+
+  // Computes the intersection between the video & audio
+  // buffered ranges.
+  // Returns true if the intersection was computed and stored in |ranges_out|.
+  // Returns false if no data was buffered.
+  bool ComputeIntersection(Ranges* ranges_out) const;

[Ami GONE FROM CHROMIUM, 2012/06/18 16:18:31]

I don't get why make the return value bool instead of void.
The condition currently returned as "false" is equivalent to
ranges_out->empty().
In fact, the same question arises for GetBufferedRanges() above.  Chasing its
(non-test) callers down, I think the only use is in WMPP, which in turn is only
used by WebMediaPlayerImpl::sourceBuffered(), where the return value could just
as well be ignored (and the for loop turns into a no-op).

Ditto for CopyIntoRanges().

[acolwell GONE FROM CHROMIUM, 2012/06/19 06:12:34]

Done.

+
   void AddIntersectionRange(
       SourceBufferStream::Timespan timespan_a,
       SourceBufferStream::Timespan timespan_b,
@@ -142,8 +149,6 @@ class MEDIA_EXPORT ChunkDemuxer : public Demuxer {
   scoped_refptr<ChunkDemuxerStream> audio_;
   scoped_refptr<ChunkDemuxerStream> video_;
 
-  int64 buffered_bytes_;
-
   base::TimeDelta duration_;
 
   typedef std::map<std::string, StreamParser*> StreamParserMap;