You need to
before you can comment on or make changes to this bug.
Setting visibility:hidden and position:fixed to an outer element then overriding the properties to visible and relative/absolute respectively breaks rendering. This seems to be a regression that occurred in the 534 - 535 range (Safari is not affected Chrome 15+ is) and can be reproduced as early as r103795 running under Safari 5.1.2 (didn't try other revisions or browsers).
Regression range: <http://trac.webkit.org/log/trunk/?rev=94017&stop_rev=94008>, meaning it's <http://trac.webkit.org/changeset/94016>.
Created an attachment (id=122290) [details]
Same test as an attachment.
Created an attachment (id=122758) [details]
Proposed fix: correct the optimization to not forget some cases.
Is this different to the fix I made yesterday, http://trac.webkit.org/changeset/105471 ?
(In reply to comment #4)
> Is this different to the fix I made yesterday, http://trac.webkit.org/changeset/105471 ?
Yes. This bug does not involve composited layers at all and is about scrolling.
Simon, could you look at this patch when you have a minute? I really would like to knock this regression out.
(From update of attachment 122758 [details])
View in context: https://bugs.webkit.org/attachment.cgi?id=122758&action=review
> +<div style="overflow: hidden; visibility: hidden;" id="outerDiv">
> + <div style="position: relative; background: red; visibility: visible;"></div>
> + <div style="position: relative; overflow: hidden; visibility: visible;">
> + <div style="background: green"></div>
> + </div>
I'd prefer the style to not be inline, and use descriptive class names.
Created an attachment (id=124438) [details]
Patch for landing
(From update of attachment 124438 [details])
Clearing flags on attachment: 124438
Committed r106191: <http://trac.webkit.org/changeset/106191>
All reviewed patches have been landed. Closing bug.