Remove SkImageGenerator pieces only for SkCodec.

Remove SkImageGenerator pieces only for SkCodec.

Follow up to the split between SkImageGenerator and SkCodec. Now that
SkCodec does not inherit from SkImageGenerator, SkImageGenerator no
longer needs Options or Result, which were added for SkCodec. Remove
them, but keep them behind a flag, since Chromium has its own
subclasses of SkImageGenerator which assume the old signature for
onGetPixels.

Committed: https://skia.googlesource.com/skia/+/5315fd4761a3c510dfff834a84e71e4c471951f9

[scroggo, 2015-07-08 18:53:13]

scroggo@google.com changed reviewers:
+ reed@google.com, scroggo@google.com

[scroggo, 2015-07-08 18:53:13]

Where do we land the flag in Chromium?

This might be a tricky transition due to the fact that one SkImageGenerator
subclass is in Chromium [1], and another is in Blink [2].

[1]
https://code.google.com/p/chromium/codesearch#chromium/src/media/blink/skcanv...
[2]
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...

[reed1, 2015-07-08 19:01:40]

seems like we should land the #define in chrome's SkUserConfig.h first, before
this CL.

[scroggo_chromium, 2015-07-09 15:19:57]

On 2015/07/08 19:01:40, reed1 wrote:
> seems like we should land the #define in chrome's SkUserConfig.h first, before
> this CL.

That define has landed in https://codereview.chromium.org/1227113002/ (DIFFERENT
ISSUE).

PTAL

[reed1, 2015-07-09 15:26:45]

https://codereview.chromium.org/1226023003/diff/1/include/core/SkImageGenerat...
File include/core/SkImageGenerator.h (right):

https://codereview.chromium.org/1226023003/diff/1/include/core/SkImageGenerat...
include/core/SkImageGenerator.h:178: bool getPixels(const SkImageInfo& info,
void* pixels, size_t rowBytes,
Can you just expose these variants of getPixels w/o the #else? That way other
callers can just switch to them unconditionally, and we only need to guard
(perhaps) the virtuals.

[scroggo, 2015-07-09 15:51:59]

https://codereview.chromium.org/1226023003/diff/1/include/core/SkImageGenerat...
File include/core/SkImageGenerator.h (right):

https://codereview.chromium.org/1226023003/diff/1/include/core/SkImageGenerat...
include/core/SkImageGenerator.h:178: bool getPixels(const SkImageInfo& info,
void* pixels, size_t rowBytes,
On 2015/07/09 15:26:45, reed1 wrote:
> Can you just expose these variants of getPixels w/o the #else? That way other
> callers can just switch to them unconditionally, and we only need to guard
> (perhaps) the virtuals.

Yes. I'd forgotten that all of the callers of getPixels are inside Skia, so no
need to stage the public interface.

[reed1, 2015-07-09 15:53:35]

Even if there are some in chrome you could still do


#ifdef LEGACY
   void oldGetPixels...
#endif

    void newGetPixels...

Assuming newGetPixels can call into the old virtual by passing in dummy args.

[scroggo, 2015-07-09 15:55:24]

On 2015/07/09 15:53:35, reed1 wrote:
> Even if there are some in chrome you could still do
> 
> 
> #ifdef LEGACY
>    void oldGetPixels...
> #endif
> 
>     void newGetPixels...
> 
> Assuming newGetPixels can call into the old virtual by passing in dummy args.

Indeed, but no need. I have removed the ifdefs for the public interface. PTAL

[reed1, 2015-07-09 15:57:24]

lgtm

[scroggo, 2015-07-09 15:58:16]

The CQ bit was checked by scroggo@google.com

[commit-bot: I haz the power, 2015-07-09 15:58:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1226023003/40001

[commit-bot: I haz the power, 2015-07-09 16:08:03]

Committed patchset #3 (id:40001) as
https://skia.googlesource.com/skia/+/5315fd4761a3c510dfff834a84e71e4c471951f9