Removing unnecessary assert from BitmapImage::draw

Removing unnecessary asserts from BitmapImage::draw

BUG=496605
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=197066

[Justin Novosad, 2015-06-12 16:38:11]

junov@chromium.org changed reviewers:
+ chrishtr@chromium.org

[Justin Novosad, 2015-06-12 16:38:11]

PTAL

[chrishtr, 2015-06-12 16:45:12]

https://codereview.chromium.org/1181963006/diff/1/Source/platform/graphics/Gr...
File Source/platform/graphics/GraphicsContext.cpp (left):

https://codereview.chromium.org/1181963006/diff/1/Source/platform/graphics/Gr...
Source/platform/graphics/GraphicsContext.cpp:949: image->draw(m_canvas,
imagePaint, dest, src, shouldRespectImageOrientation,
Image::ClampImageToSourceRect);
Why is the rect of negative width or height? Can it be fixed earlier in the call
stack?

[Justin Novosad, 2015-06-12 17:06:49]

On 2015/06/12 16:45:12, chrishtr wrote:
>
https://codereview.chromium.org/1181963006/diff/1/Source/platform/graphics/Gr...
> File Source/platform/graphics/GraphicsContext.cpp (left):
> 
>
https://codereview.chromium.org/1181963006/diff/1/Source/platform/graphics/Gr...
> Source/platform/graphics/GraphicsContext.cpp:949: image->draw(m_canvas,
> imagePaint, dest, src, shouldRespectImageOrientation,
> Image::ClampImageToSourceRect);
> Why is the rect of negative width or height? Can it be fixed earlier in the
call
> stack?

It could, assuming this is the only case.  Let me check that out...

[Justin Novosad, 2015-06-12 19:18:14]

After some digging... the right fix is to just remove the asserts. Skia handles
the negative size just fine.

[chrishtr, 2015-06-12 19:44:27]

On 2015/06/12 at 19:18:14, junov wrote:
> After some digging... the right fix is to just remove the asserts. Skia
handles the negative size just fine.

Can you point at the code in skia? thanks.

[Justin Novosad, 2015-06-12 20:14:35]

Sure. Skia just rejects the draw command when dest rect is inverted (in
SkCanvas::quickReject), which is actually what we want in these case where the
layout calculations cause an inversion due to inset self-overlaps. the rectangle
normalization that existed before was futile because it was ending up issuing
draws that were going to be clipped-out anyways.

[chrishtr, 2015-06-12 20:19:32]

The CQ bit was checked by chrishtr@chromium.org

[chrishtr, 2015-06-12 20:19:33]

lgtm

[commit-bot: I haz the power, 2015-06-12 20:19:57]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1181963006/40001

[commit-bot: I haz the power, 2015-06-12 22:30:29]

Committed patchset #3 (id:40001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=197066