Demuxer stream now uses a callback instead of Assignable template. Cleaning ...

Demuxer stream now uses a callback instead of Assignable template.  Cleaning up Pipeine interfaces to conform with Chrome Base design.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=12652

[scherkus (not reviewing), 2009-03-27 01:46:07]

LGTM with some comments

http://codereview.chromium.org/53126/diff/1/6
File media/base/filters.h (right):

http://codereview.chromium.org/53126/diff/1/6#newcode162
Line 162: virtual void Read(Callback1<Buffer*>::Type* read_callback) = 0;
here's just a crazy thought.. what if we used scoped_refptr<Buffer> instead? 
I'm not sure if its worth the hassle.

http://codereview.chromium.org/53126/diff/1/3
File media/filters/ffmpeg_demuxer.cc (right):

http://codereview.chromium.org/53126/diff/1/3#newcode148
Line 148: Callback1<Buffer*>::Type* read_callback;
may consider using scoped_ptr for this

[ralphl, 2009-03-27 05:36:36]

Changed to scoped_ptr in demuxer inner loop.  Since the tree is closed, I
decided to upload one more time to make 100% sure the try servers were happy
with the code.  

I'll check it in as soon as the tree opens.

http://codereview.chromium.org/53126/diff/1/6
File media/base/filters.h (right):

http://codereview.chromium.org/53126/diff/1/6#newcode162
Line 162: virtual void Read(Callback1<Buffer*>::Type* read_callback) = 0;
On 2009/03/27 01:46:08, scherkus wrote:
> here's just a crazy thought.. what if we used scoped_refptr<Buffer> instead? 
> I'm not sure if its worth the hassle.

In cases where the caller passes a pointer, the caller is always guaranteeing
that the pointer is valid for the duration of the call.  In the case of ref
counted objects, the called procedure must reference the object within the call
to claim ownership.

http://codereview.chromium.org/53126/diff/1/3
File media/filters/ffmpeg_demuxer.cc (right):

http://codereview.chromium.org/53126/diff/1/3#newcode148
Line 148: Callback1<Buffer*>::Type* read_callback;
On 2009/03/27 01:46:08, scherkus wrote:
> may consider using scoped_ptr for this

Good idea.  Changed it.