Bug 50967 – Reflection does not render properly when -webkit-transform is removed and re-added

• Status: RESOLVED
• Resolution: FIXED
• Component: CSS

[Simon Fraser (smfr)]

Created attachment 76431 [details]
Testcase

Steps to reproduce:
1. Load reflection.html (attached) in Safari
2. Click on "Toggle Reflection" to show reflection
3. Click again to hide
4. Click again to show

Results:
The reflection has an incorrect offset of 100px (see attached screenshot) in Desktop Sarfari Version 5.0.3 (6533.19.4).

[Simon Fraser (smfr)]

<rdar://problem/8729148>

[Simon Fraser (smfr)]

I filed bug 50968 to remove layer clones when we can (which would be another way to fix this bug). That's not completely trivial, though.

[Simon Fraser (smfr)]

Created attachment 76432 [details]
Patch

[Eric Seidel (no email)]

Comment on attachment 76432 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=76432&action=review

> LayoutTests/compositing/reflections/remove-add-reflection.html:38
> +        window.setTimeout(function() {

timeouts often make for flaky tests :(

> LayoutTests/compositing/reflections/remove-add-reflection.html:59
>  \ No newline at end of file

EOF newlines are always nice. :)

[Simon Fraser (smfr)]

(In reply to comment #4)
> (From update of attachment 76432 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=76432&action=review
> 
> > LayoutTests/compositing/reflections/remove-add-reflection.html:38
> > +        window.setTimeout(function() {
> 
> timeouts often make for flaky tests :(

Unfortunately we have to use them for things that requires compositing layer pixel tests, because CA has built-in delays.

A future solution would be to build in some DRT callbacks.

> > LayoutTests/compositing/reflections/remove-add-reflection.html:59
> >  \ No newline at end of file
> 
> EOF newlines are always nice. :)

That's fixable :)

[Simon Fraser (smfr)]

http://trac.webkit.org/changeset/74155