Fixed crashes related to incorrect sizing of the Canvas's backing...

Fixed crashes on Windows related to incorrect sizing of the Canvas's
backing store when CSS styles are used to resize the Canvas. Fixed
incorrect rendering output on Mac OS X caused by similar issue.

The deeper issue is why the Canvas's backing store is incorrectly
sized in this case. http://crbug.com/26542 has been filed to track
this remaining problem.

BUG=http://crbug.com/26515
TEST=none (covered by preexisting manual WebGL tests)


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=30859

[Ken Russell (switch to Gerrit), 2009-11-03 01:31:28]

Fixed incorrect rendering output on Mac OS X as well.

[brettw, 2009-11-03 17:28:20]

This seems complicated to avoid a crash in a case that shouldn't occur in the
first place.

How about just doing the copy from GL using the min of the GL and the canvas
size? When this is done, we should focus our efforts on fixing the sizing
problem for real.

[Ken Russell (switch to Gerrit), 2009-11-03 17:42:31]

On 2009/11/03 17:28:20, brettw wrote:
> This seems complicated to avoid a crash in a case that shouldn't occur in the
> first place.
> 
> How about just doing the copy from GL using the min of the GL and the canvas
> size? When this is done, we should focus our efforts on fixing the sizing
> problem for real.

After reviewing the Canvas spec at
http://www.whatwg.org/specs/web-apps/current-work/multipage/the-canvas-elemen...
, in particular:

"The intrinsic dimensions of the canvas element equal the size of the coordinate
space, with the numbers interpreted in CSS pixels. However, the element can be
sized arbitrarily by a style sheet. During rendering, the image is scaled to fit
this layout size."

I think the behavior of the Canvas in this case is actually correct, and that we
should close http://crbug.com/26542 as "not a bug". I believe this is the
correct fix for this issue.

[brettw, 2009-11-03 20:21:18]

LGTM

http://codereview.chromium.org/342087/diff/1002/2001
File webkit/api/src/GraphicsContext3D.cpp (right):

http://codereview.chromium.org/342087/diff/1002/2001#newcode755
Line 755: canvas.drawBitmapRect(*m_resizingBitmap, 0, dst);
Do you want antialiasing? This call will not give you any. If you do, make a
SkPaint and do setFilterBitmap (or something like that) on it before passing it
into drawBitmapRect.

[Ken Russell (switch to Gerrit), 2009-11-03 20:46:19]

On 2009/11/03 20:21:18, brettw wrote:
> 
> http://codereview.chromium.org/342087/diff/1002/2001
> File webkit/api/src/GraphicsContext3D.cpp (right):
> 
> http://codereview.chromium.org/342087/diff/1002/2001#newcode755
> Line 755: canvas.drawBitmapRect(*m_resizingBitmap, 0, dst);
> Do you want antialiasing? This call will not give you any. If you do, make a
> SkPaint and do setFilterBitmap (or something like that) on it before passing
it
> into drawBitmapRect.

Thanks for the hint. Antialiasing doesn't seem to be necessary here. In these
cases the canvas's content is being downscaled fairly drastically before being
scaled back up (with antialiasing) during page composition.