Bug 16334 – REGRESSION (r28299): Homepage of any DotMac Web Gallery won't load completely

• Component: New Bugs
• Status: RESOLVED FIXED

[Vincent Gasseling]

The homepage of any gallery.mac.com-site won't load completely, only a couple of placeholders do appear. If you click on a placeholder and then after the next page has loaded go back via 'Emily's Web Gallery' (left top) to the homepage it will load, but then after a reload again only a couple of placeholders will appear.

[David Kilzer (:ddkilzer)]

Confirmed with a local debug build of WebKit r28483 with Safari 3.0.4 (523.12) on Mac OS X 10.4.11 (8S165).

This is a regression from Safari 3.0.4 (523.12) with original WebKit on Tiger.

[David Kilzer (:ddkilzer)]

Searching internal builds, this regressed from r28298 to r28299.
http://trac.webkit.org/projects/webkit/changeset/28299

[David Kilzer (:ddkilzer)]

<rdar://problem/5634923>

Using Safari 3.0.4 (523.12) on Mac OS X 10.4.11 I get a page that tells me that my browser is incompatible with .Mac Web Gallery :-\

[David Kilzer (:ddkilzer)]

(In reply to comment #4)
> Using Safari 3.0.4 (523.12) on Mac OS X 10.4.11 I get a page that tells me that
> my browser is incompatible with .Mac Web Gallery :-\

I saw that intermittently when testing nightly revisions.  Hmmm...guess it happens all the time with Safari 3.0.4 on Tiger.  Works okay with Safari 3.0.4 on Leopard.

(In reply to comment #5)
> (In reply to comment #4)
> > Using Safari 3.0.4 (523.12) on Mac OS X 10.4.11 I get a page that tells me that
> > my browser is incompatible with .Mac Web Gallery :-\
> 
> I saw that intermittently when testing nightly revisions.  Hmmm...guess it
> happens all the time with Safari 3.0.4 on Tiger.  Works okay with Safari 3.0.4
> on Leopard.
> 

I got it intermittently on some .Mac galleries regardless of the browser I was using. I've found another gallery where I could reproduce the problem and I know what's happening - something I've convinced myself couldn't happen when I wrote the original patch :-)

Created attachment 17771 [details]
Test case

You should see a green square (without having to resize the window or otherwise force relayout after the page loads).

Created attachment 17772 [details]
Avoid re-entering scheduleRelayoutOfSubtree()

No layout test regressions.

[Darin Adler]

Comment on attachment 17772 [details]
Avoid re-entering scheduleRelayoutOfSubtree()

r=me

I am getting frustrated with the difficulty of understanding things like the second boolean to setChildNeedsLayout and the various constraints on what is and is not safe to call. We need to make some progress in clarity in the render tree code when viewed up close -- ideally we will get to the point where we can reason about the correctness of a fragment of code without having to constantly consider the big picture.

Landed in <http://trac.webkit.org/projects/webkit/changeset/28519>.