Stop using GrDrawState to track stencil state

[TomH, 2012-05-21 20:53:17]

LGTM.

Do you believe in deja vu?
Although you didn't document which melody you're singing "lala" to.

[robertphillips, 2012-05-21 21:00:17]

LGTM

http://codereview.appspot.com/6227047/diff/2001/src/gpu/gl/GrGpuGL.h
File src/gpu/gl/GrGpuGL.h (right):

http://codereview.appspot.com/6227047/diff/2001/src/gpu/gl/GrGpuGL.h#newcode281
src/gpu/gl/GrGpuGL.h:281: kIgnoreClip_StencilClipMode,
From the code it looks like we only really need one of Ignore and Invalid.

[bsalomon, 2012-05-21 21:11:04]

http://codereview.appspot.com/6227047/diff/2001/src/gpu/gl/GrGpuGL.h
File src/gpu/gl/GrGpuGL.h (right):

http://codereview.appspot.com/6227047/diff/2001/src/gpu/gl/GrGpuGL.h#newcode281
src/gpu/gl/GrGpuGL.h:281: kIgnoreClip_StencilClipMode,
On 2012/05/21 21:00:18, robertphillips wrote:
> From the code it looks like we only really need one of Ignore and Invalid.

I think we could remove invalid because whenever we set fHWStencilClipMode to
invalid we also invalidate fHWStencilSettings, meaning we will fully reconfigure
the stencil state on the next draw. We could leave fHWStencilClipMode set to
whatever value we wanted at that point. But I'd feel compelled to document that
and it just seems easier to have an explicit invalid state.

[TomH, 2012-05-21 21:18:05]

On 2012/05/21 21:11:04, bsalomon wrote:

> ... But I'd feel compelled to document that
> and it just seems easier to have an explicit invalid state.

Agreed.

[bsalomon, 2012-05-21 21:29:36]

Closed with r4021.