skia2/dm
scroggo f2c96a2a6f Only decode opaque to premul in DM image
Previously, we would test decoding an opaque SkCodec to all three:
kUnpremul,
kPremul,
kOpaque

The image should look the same in all three cases. We already test for
that in CodecTest [1], where we require that the result matches exactly.
CodecTest runs on a smaller set of images, but it covers a variety of
opaque images. No need to test on all opaque images.

Running locally on my Mac laptop, the following command:

    dm --src image --images resources/

dropped from:
5932 srcs
11.2s

to:

3544 srcs
6.69s

for a 40% speedup.

BUG=skia:5307

[1] https://skia.googlesource.com/skia/+/master/tests/CodecTest.cpp#119
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1999593003

Review-Url: https://codereview.chromium.org/1999593003
2016-05-20 11:08:27 -07:00
..
DM.cpp Only decode opaque to premul in DM image 2016-05-20 11:08:27 -07:00
DMGpuSupport.h Rename lots of things from 'sRGB' to 'GammaCorrect', where appropriate 2016-04-13 13:10:14 -07:00
DMJsonWriter.cpp Revert of SK_DECLARE_STATIC_MUTEX -> static SkMutex (patchset #1 id:1 of https://codereview.chromium.org/1948193002/ ) 2016-05-04 17:12:46 -07:00
DMJsonWriter.h Add gamma_correct option field to dm.json 2016-02-29 07:38:01 -08:00
DMSrcSink.cpp Delete SkBitmapRegionCanvas 2016-05-18 06:23:57 -07:00
DMSrcSink.h Delete SkBitmapRegionCanvas 2016-05-18 06:23:57 -07:00
DMSrcSinkAndroid.cpp Simplify linkages to Android framework internals 2015-10-26 07:21:32 -07:00
DMSrcSinkAndroid.h DM: this must have been missing from the previous CL 2016-02-08 14:30:25 -08:00