[webkit-reviews] review denied: [Bug 47868] Pixel cracks when using background image sprite on transformed element at certain scales. : [Attachment 71117] Change to sub-image path when CGContext has scale and sourceRect is not the same as destRect.

[bugzilla-daemon at webkit.org]

mitz at webkit.org has denied Yongjun Zhang <yongjun_zhang at apple.com>'s request
for review:
Bug 47868: Pixel cracks when using background image sprite on transformed
element at certain scales.
https://bugs.webkit.org/show_bug.cgi?id=47868

Attachment 71117: Change to sub-image path when CGContext has scale and
sourceRect is not the same as destRect.
https://bugs.webkit.org/attachment.cgi?id=71117&action=review

------- Additional Comments from mitz at webkit.org
View in context: https://bugs.webkit.org/attachment.cgi?id=71117&action=review

> WebCore/platform/graphics/cg/ImageCG.cpp:189
> +	   bool hasScale = transform.a != 1.0f || transform.d != 1.0f;

WebKit style is to omit the “.0f”. It serves no useful purpose whatsoever.

This test criteria seem to be wrong. On the one hand, the (very common) flip
transform is included, and on the other hand, a shear or skew transform, which
is likely to require interpolation and therefore introduce bleeding, is
excluded. AffineTransform::isIdentityOrTranslationOrFlipped() may be the right
test here.