EncoderZlib/DecoderZlib for chromoting

EncoderZlib/DecoderZlib for chromoting

Encoder and decoder using zlib for chromoting. This implementation has zero
copy out of the decoder and encoder. The consequence is that we have to break
out the zlib stream into rect boundaries which requires the synchronication
flush feature in zlib. This feature will hurt compression ratio but the effect
still need to be measured.

This patch also provides tests for testing the Encoder and Decoder pair with
zlib.

TEST=remoting_unittests

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=53738

[Alpha Left Google, 2010-07-23 00:39:53]

ping.

[dmac, 2010-07-23 05:35:08]

http://codereview.chromium.org/2868062/diff/1/2
File remoting/base/codec_test.cc (right):

http://codereview.chromium.org/2868062/diff/1/2#newcode204
remoting/base/codec_test.cc:204: LOG(INFO) << "Testing Rect " << i;
did you mean to leave the log? I assume so, just checking.

http://codereview.chromium.org/2868062/diff/1/3
File remoting/base/compressor.h (right):

http://codereview.chromium.org/2868062/diff/1/3#newcode23
remoting/base/compressor.h:23: };
maybe typedef a type for this

typedef int CompressorFlush and then use it to be clear what we are passing
around?

http://codereview.chromium.org/2868062/diff/1/3#newcode49
remoting/base/compressor.h:49: int flush, int* consumed, int* written) = 0;
would use CompressorFlush (or whatever from above) in place of int flush here.

[Alpha Left Google, 2010-07-26 20:59:54]

http://codereview.chromium.org/2868062/diff/1/2
File remoting/base/codec_test.cc (right):

http://codereview.chromium.org/2868062/diff/1/2#newcode204
remoting/base/codec_test.cc:204: LOG(INFO) << "Testing Rect " << i;
I do mean to leave log, wants to see which case is failed.

http://codereview.chromium.org/2868062/diff/1/3
File remoting/base/compressor.h (right):

http://codereview.chromium.org/2868062/diff/1/3#newcode23
remoting/base/compressor.h:23: };
On 2010/07/23 05:35:08, dmac wrote:
> maybe typedef a type for this
> 
> typedef int CompressorFlush and then use it to be clear what we are passing
> around?

Done.

http://codereview.chromium.org/2868062/diff/1/3#newcode49
remoting/base/compressor.h:49: int flush, int* consumed, int* written) = 0;
On 2010/07/23 05:35:08, dmac wrote:
> would use CompressorFlush (or whatever from above) in place of int flush here.

Done.

[dmac, 2010-07-26 21:48:05]

LGTM