Don't clobber initial transform with SetMatrix.

Don't clobber initial transform with SetMatrix.

BUG=skia:2378
Committed: http://code.google.com/p/skia/source/detail?r=14778

[reed1, 2014-05-19 15:04:34]

https://codereview.chromium.org/290883004/diff/1/src/record/SkRecordDraw.cpp
File src/record/SkRecordDraw.cpp (right):

https://codereview.chromium.org/290883004/diff/1/src/record/SkRecordDraw.cpp#...
src/record/SkRecordDraw.cpp:45: ctm.postConcat(fInitialCTM);
seems fine.

NIT: Is it any more efficient to say:

SkMatrix ctm;
ctm.setConcat(fInitialCTM, r.matrix);

To avoid the initial copy?

https://codereview.chromium.org/290883004/diff/1/src/record/SkRecordDraw.h
File src/record/SkRecordDraw.h (right):

https://codereview.chromium.org/290883004/diff/1/src/record/SkRecordDraw.h#ne...
src/record/SkRecordDraw.h:48: const SkMatrix fInitialCTM;
10x4=40 bytes added to every record. I guess space is not an object (as we could
conceivably store an index to the finite set of different CTMs we encounter in a
given picture)?

[reed1, 2014-05-19 15:07:29]

I withdraw all misguided questions.

lgtm

[mtklein, 2014-05-19 15:11:23]

lgtm

https://codereview.chromium.org/290883004/diff/1/src/record/SkRecordDraw.cpp
File src/record/SkRecordDraw.cpp (right):

https://codereview.chromium.org/290883004/diff/1/src/record/SkRecordDraw.cpp#...
src/record/SkRecordDraw.cpp:45: ctm.postConcat(fInitialCTM);
On 2014/05/19 15:04:34, reed1 wrote:
> seems fine.
> 
> NIT: Is it any more efficient to say:
> 
> SkMatrix ctm;
> ctm.setConcat(fInitialCTM, r.matrix);
> 
> To avoid the initial copy?

Yes, I think that is a bit more efficient.  Done.

https://codereview.chromium.org/290883004/diff/1/src/record/SkRecordDraw.h
File src/record/SkRecordDraw.h (right):

https://codereview.chromium.org/290883004/diff/1/src/record/SkRecordDraw.h#ne...
src/record/SkRecordDraw.h:48: const SkMatrix fInitialCTM;
On 2014/05/19 15:04:34, reed1 wrote:
> 10x4=40 bytes added to every record. I guess space is not an object (as we
could
> conceivably store an index to the finite set of different CTMs we encounter in
a
> given picture)?

This is 40 bytes extra used on the stack while playing the picture back. 
Nothing extra called or stored per-op.

[commit-bot: I haz the power, 2014-05-19 15:11:30]

CQ is trying da patch. Follow status at
 https://skia-tree-status.appspot.com/cq/mtklein@chromium.org/290883004/20001

[commit-bot: I haz the power, 2014-05-19 15:15:30]

Change committed as 14778