Further cleanup in FilterEffectRenderer.

Further cleanup in FilterEffectRenderer.

Change FilterEffectRenderer not to be a subclass of Filter anymore, and instantiate a ReferenceFilter as a parent for all shorthand filters instead. Remove FilterEffectRendererHelper entirely, and put its remaining functionality into FilterEffectRenderer.

Replace the AffineTransform in Filter with a float scale parameter, since that's all it does now anyway. Note that applyHorizontal/VerticalScale() still need to remain separate, since SVGFilter applies further, possibly non-uniform scaling in its overrides of those functions.

prepareFilterEffect() and beginFilterEffect() have been merged. This removes an incorrect early-out, which fixes http://crbug.com/254449.

BUG=240827, 254449
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=183676

[Stephen White, 2014-10-10 15:45:32]

senorblanco@chromium.org changed reviewers:
+ fmalita@chromium.org

[Stephen White, 2014-10-10 16:16:29]

fmalita@: PTAL. Thanks!

[Stephen White, 2014-10-14 14:03:59]

Florin: friendly ping. :)

[f(malita), 2014-10-14 14:17:16]

On 2014/10/14 14:03:59, Stephen White wrote:
> Florin: friendly ping. :)

Whoops, started looking and never finished :)

LGTM

[f(malita), 2014-10-14 14:18:23]

https://codereview.chromium.org/644713004/diff/410001/Source/platform/graphic...
File Source/platform/graphics/filters/Filter.h (right):

https://codereview.chromium.org/644713004/diff/410001/Source/platform/graphic...
Source/platform/graphics/filters/Filter.h:61: float m_inverseScale;
If this is always 1/m_scale, can't we just compute it on the fly?

[Stephen White, 2014-10-14 14:26:53]

https://codereview.chromium.org/644713004/diff/410001/Source/platform/graphic...
File Source/platform/graphics/filters/Filter.h (right):

https://codereview.chromium.org/644713004/diff/410001/Source/platform/graphic...
Source/platform/graphics/filters/Filter.h:61: float m_inverseScale;
On 2014/10/14 14:18:23, Florin Malita wrote:
> If this is always 1/m_scale, can't we just compute it on the fly?

Will do. Divisions still take double-digit CPU cycles, but this one is probably
not a bottleneck.

[Stephen White, 2014-10-14 15:16:24]

The CQ bit was checked by senorblanco@chromium.org

[commit-bot: I haz the power, 2014-10-14 15:17:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/644713004/430001

[commit-bot: I haz the power, 2014-10-14 15:41:22]

Committed patchset #7 (id:430001) as 183676