[webkit-reviews] review denied: [Bug 95240] RenderLayer doesn't clean up its transparentPaintingAncestor's transparencyLayer : [Attachment 161054] Patch

[bugzilla-daemon at webkit.org]

Simon Fraser (smfr) <simon.fraser at apple.com> has denied Dana Jansens
<danakj at chromium.org>'s request for review:
Bug 95240: RenderLayer doesn't clean up its transparentPaintingAncestor's
transparencyLayer
https://bugs.webkit.org/show_bug.cgi?id=95240

Attachment 161054: Patch
https://bugs.webkit.org/attachment.cgi?id=161054&action=review

------- Additional Comments from Simon Fraser (smfr) <simon.fraser at apple.com>
I think this fix is taking the wrong approach.

The actual bug is that 'outer' is not becoming a stacking context when the mask
is applied in the keyframes, yet our transparency layer code assumes that
things with masks are stacking context. Here's the layer tree:

(lldb) expr showLayerTree(this)
layer 0x109b02798 at (0,0) size 1231x1175 (composited, bounds=at (0,0) size
1231x1175, drawsContent=1, paints into ancestor=0)
  RenderView 0x109b01fa8 at (0,0) size 1231x1175
 positive z-order list(1)
  layer 0x109b31f18 at (0,0) size 1231x1175
    RenderBlock 0x109b31be8 {HTML} at (0,0) size 1231x1175
      RenderBody 0x109b327e8 {BODY} at (8,8) size 1215x1159
	RenderBlock (anonymous) 0x109b4eb58 at (0,0) size 1215x0
	  RenderInline 0x109b4dc08 {DOCTYPE} at (0,0) size 0x0
	    RenderText 0x109b56af8 {#text} at (0,0) size 0x0
	    RenderText 0x109b56a78 {#text} at (0,0) size 0x0
	    RenderText 0x109b59f28 {#text} at (0,0) size 0x0
	    RenderText 0x109b49af8 {#text} at (0,0) size 0x0
	RenderBlock (anonymous) 0x109b4ea18 at (0,0) size 1215x100
	RenderBlock (anonymous) 0x109b4fa68 at (0,100) size 1215x0
	  RenderInline 0x109b4fbc8 {DOCTYPE} at (0,0) size 0x0
   positive z-order list(2)
    layer 0x109b4e8e8 at (8,8) size 1215x100
      RenderBlock 0x109b54688 {DIV} at (0,0) size 1215x100 class="outer"
    layer 0x109b504b8 at (8,8) size 100x100
      RenderBlock 0x109b503e8 {DIV} at (0,0) size 100x100 [bgcolor=#FFFFFF]
class="inner"

This is similar to the problem described in bug 66341.