Do a better job of enforcing the semantics of the setRequireUnpremultipliedColors flag

Do a better job of enforcing the semantics of the setRequireUnpremultipliedColors flag

Committed: https://skia.googlesource.com/skia/+/3b570c6b023c5556613f435400798d5df37a78e3

[scroggo, 2014-06-10 13:50:23]

lgtm

https://codereview.chromium.org/322813005/diff/20001/src/images/SkImageDecode...
File src/images/SkImageDecoder_ktx.cpp (right):

https://codereview.chromium.org/322813005/diff/20001/src/images/SkImageDecode...
src/images/SkImageDecoder_ktx.cpp:72: bool bSrcIsPremul =
ktxFile.getValueForKey(premulKey) == SkString("True");
const

https://codereview.chromium.org/322813005/diff/20001/src/images/SkImageDecode...
src/images/SkImageDecoder_ktx.cpp:83: // premul, then we cannot honor their
wish.
FWIW, we *could* do a second pass and unpremultiply the data
(src/ports/SkImageDecoder_CG.cpp does this). I think it's okay to return false
though.

https://codereview.chromium.org/322813005/diff/20001/src/images/SkImageDecode...
src/images/SkImageDecoder_ktx.cpp:165: SkScaledBitmapSampler::Options opts
(*this);
It might be simpler to move this outside of the if statement:

SkScaledBitmapSampler::Options opts (*this);
if (bSrcIsPremul) {
  // asserts
  opts.fPremultiplyAlpha = false;
}
if (!sampler.begin(bm, SkScaledBitmapSampler::kRGBA, opts)) {
  return false;
}

[krajcevski, 2014-06-10 14:47:29]

https://codereview.chromium.org/322813005/diff/20001/src/images/SkImageDecode...
File src/images/SkImageDecoder_ktx.cpp (right):

https://codereview.chromium.org/322813005/diff/20001/src/images/SkImageDecode...
src/images/SkImageDecoder_ktx.cpp:72: bool bSrcIsPremul =
ktxFile.getValueForKey(premulKey) == SkString("True");
On 2014/06/10 13:50:23, scroggo wrote:
> const

Done.

https://codereview.chromium.org/322813005/diff/20001/src/images/SkImageDecode...
src/images/SkImageDecoder_ktx.cpp:83: // premul, then we cannot honor their
wish.
On 2014/06/10 13:50:23, scroggo wrote:
> FWIW, we *could* do a second pass and unpremultiply the data
> (src/ports/SkImageDecoder_CG.cpp does this). I think it's okay to return false
> though.

I agree, although I believe the use case for that is pretty limited so it's not
worth the time right now.

https://codereview.chromium.org/322813005/diff/20001/src/images/SkImageDecode...
src/images/SkImageDecoder_ktx.cpp:165: SkScaledBitmapSampler::Options opts
(*this);
On 2014/06/10 13:50:23, scroggo wrote:
> It might be simpler to move this outside of the if statement:
> 
> SkScaledBitmapSampler::Options opts (*this);
> if (bSrcIsPremul) {
>   // asserts
>   opts.fPremultiplyAlpha = false;
> }
> if (!sampler.begin(bm, SkScaledBitmapSampler::kRGBA, opts)) {
>   return false;
> }

Done.

[krajcevski, 2014-06-10 14:47:38]

The CQ bit was checked by krajcevski@google.com

[commit-bot: I haz the power, 2014-06-10 14:48:34]

CQ is trying da patch. Follow status at
 https://skia-tree-status.appspot.com/cq/krajcevski@google.com/322813005/40001

[commit-bot: I haz the power, 2014-06-10 16:17:55]

Change committed as 3b570c6b023c5556613f435400798d5df37a78e3