Don't track FB depth attachment in debug GL context

Don't track FB depth attachment in debug GL context

Quits tracking depth attachment in the debug GL context. Skia does not
use depth buffers so this extra complexity is not required. We can
always add it back in the future if needed.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2085553004

[csmartdalton, 2016-06-21 14:01:06]

Description was changed from

==========
Don't track depth FB attachment in debug GL context

Quits tracking depth attachment in the debug GL context. Skia does not
use depth buffers so this extra complexity is not required. We can
always add it back in the future if needed.

BUG=skia:
==========

to

==========
Don't track depth FB attachment in debug GL context

Quits tracking depth attachment in the debug GL context. Skia does not
use depth buffers so this extra complexity is not required. We can
always add it back in the future if needed.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2085553004
==========

[csmartdalton, 2016-06-21 14:01:52]

Description was changed from

==========
Don't track depth FB attachment in debug GL context

Quits tracking depth attachment in the debug GL context. Skia does not
use depth buffers so this extra complexity is not required. We can
always add it back in the future if needed.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2085553004
==========

to

==========
Don't track FB depth attachment in debug GL context

Quits tracking depth attachment in the debug GL context. Skia does not
use depth buffers so this extra complexity is not required. We can
always add it back in the future if needed.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2085553004
==========

[csmartdalton, 2016-06-21 14:02:03]

csmartdalton@google.com changed reviewers:
+ bsalomon@google.com

[bsalomon, 2016-06-21 14:04:14]

Does this make something easier for a coming refactor?

We don't currently use the depth buffer, though that may change as we explore
shadow APIs.

Also, we do bind to the depth attachment when using depth-stencil formats.

[csmartdalton, 2016-06-21 14:49:59]

On 2016/06/21 14:04:14, bsalomon wrote:
> Does this make something easier for a coming refactor?
> 
> We don't currently use the depth buffer, though that may change as we explore
> shadow APIs.
> 
> Also, we do bind to the depth attachment when using depth-stencil formats.

I forgot about depth-stencil. But it looks like the existing code doesn't use
that anyway?
https://skia.googlesource.com/skia/+/refs/heads/chrome/m52/tools/gpu/gl/debug....
Does anything actually use depth-stencil with skia?

This only makes things slightly easier for implementing glGet(GL_SAMPLES). I
mostly did it because I thought it was not being used.

[bsalomon, 2016-06-21 15:22:22]

On 2016/06/21 14:49:59, csmartdalton wrote:
> On 2016/06/21 14:04:14, bsalomon wrote:
> > Does this make something easier for a coming refactor?
> > 
> > We don't currently use the depth buffer, though that may change as we
explore
> > shadow APIs.
> > 
> > Also, we do bind to the depth attachment when using depth-stencil formats.
> 
> I forgot about depth-stencil. But it looks like the existing code doesn't use
> that anyway?
>
https://skia.googlesource.com/skia/+/refs/heads/chrome/m52/tools/gpu/gl/debug....
> Does anything actually use depth-stencil with skia?

We use depth-stencil formats exclusively for the stencil buffer. Some drivers
are unhappy if you attach a depth stencil format to GL_STENCIL_ATTACHMENT but
not GL_DEPTH_ATTACHMENT. So in GrGLRenderTarget::completeStencilAttachment() we
attach to both if the format is depth-stencil.

> 
> This only makes things slightly easier for implementing glGet(GL_SAMPLES). I
> mostly did it because I thought it was not being used.

[csmartdalton, 2016-06-21 15:24:40]

Got it. Dropping the issue.