Add flag bits for partial GrContext reset

Add flag bits for partial GrContext reset

BUG=248728
R=bsalomon@google.com

Committed: https://code.google.com/p/skia/source/detail?r=9802

Committed: https://code.google.com/p/skia/source/detail?r=9814

[bsalomon, 2013-06-18 10:16:31]

Thanks for updating the CL to have a bitfield. I didn't expect you to flesh out
all the other bits besides texture bindings, but glad you did!

I have some comments inline

https://codereview.chromium.org/17027003/diff/2001/include/gpu/GrContext.h
File include/gpu/GrContext.h (right):

https://codereview.chromium.org/17027003/diff/2001/include/gpu/GrContext.h#ne...
include/gpu/GrContext.h:59: enum GrState {
For enums that are nested within a class we don't typically prefix them by Gr. I
think we either should move this to GrTypes or remove the Gr prefix.

Also in Skia the name of the enum comes last in enum values. So kTexture_GrState
rather than kGrState_Texture.

State seems a little bit vague, we're using the word "backend" in other places
to refer to the underlying 3D API. Could we call it "BackendState" or some
variation on that?

Also, some of this is fairly GL-specific (e.g. pixelstore) and wouldn't make
sense on a theoretical D3D backend. Why don't we make this GLBackendState, make
the function take a uint32_t, and have a comment that the flag bits are
dependent on which backend is used by the context. If we're going to possibly
have multiple sets of flags then I'd prefer to move this to GrTypes.h and keep
the Gr prefixing.

https://codereview.chromium.org/17027003/diff/2001/include/gpu/GrContext.h#ne...
include/gpu/GrContext.h:61: kGrState_Texture      = 0x01 << 1,
TextureBindings?

https://codereview.chromium.org/17027003/diff/2001/include/gpu/GrContext.h#ne...
include/gpu/GrContext.h:62: kGrState_View         = 0x01 << 2,
View could use a comment, not obvious what state it refers to.

https://codereview.chromium.org/17027003/diff/2001/include/gpu/GrContext.h#ne...
include/gpu/GrContext.h:84: void resetTextureBinding();
Let's just stick with the one resetContext() call and not have a specialized
function. Blink can pass kTexture.

[kangyuan.shu, 2013-06-24 03:52:30]

hello, it's ready for review. Anyone give a comment? thanks...!

[bsalomon, 2013-06-26 19:03:07]

basically looks good. Two small requests.

https://codereview.chromium.org/17027003/diff/6001/include/gpu/GrContext.h
File include/gpu/GrContext.h (right):

https://codereview.chromium.org/17027003/diff/6001/include/gpu/GrContext.h#ne...
include/gpu/GrContext.h:63: enum GLBackendState {
Let's move it now so that we don't have to go back later and update callers to
convert from GrContext::kTextureBinding_GLBackendState to
kTextureBinding_GrGLBackendState.

https://codereview.chromium.org/17027003/diff/6001/src/gpu/gl/GrGpuGL.cpp
File src/gpu/gl/GrGpuGL.cpp (right):

https://codereview.chromium.org/17027003/diff/6001/src/gpu/gl/GrGpuGL.cpp#new...
src/gpu/gl/GrGpuGL.cpp:307: // The windows NVIDIA driver has GL_ARB_imaging in
the extension string when using a core
Nit: can you wrap this comment to 100 col lines?

[kangyuan.shu, 2013-06-27 01:53:15]

done.

[bsalomon, 2013-06-27 14:56:03]

On 2013/06/27 01:53:15, kangyuan.shu wrote:
> done.

I fixed a couple of style things in patchset 5:
   re-added prefix Gr prefixes to the enum name/values since this is no longer
in GrContext
   Added braces to the one-line if statements.


I'll put this in the commit queue.

Probably bad form to "lgtm" my own patchset, but my updates were very minor :)
Also necessary to allow the commit queue to work.

[kangyuan.shu, 2013-06-28 01:16:37]

On 2013/06/27 14:56:03, bsalomon wrote:
> On 2013/06/27 01:53:15, kangyuan.shu wrote:
> > done.
> 
> I fixed a couple of style things in patchset 5:
>    re-added prefix Gr prefixes to the enum name/values since this is no longer
> in GrContext
>    Added braces to the one-line if statements.
> 
> 
> I'll put this in the commit queue.
> 
> Probably bad form to "lgtm" my own patchset, but my updates were very minor :)
> Also necessary to allow the commit queue to work.

thanks very much!

[bsalomon, 2013-06-28 14:33:15]

Committed patchset #5 manually as r9802 (presubmit successful).

[bsalomon, 2013-06-28 18:57:40]

Committed patchset #6 manually as r9814 (presubmit successful).