MacNN's new front page features a JS powered navbar. In Safari 2.0.2 (416.13), it renders correctly, with the moused-over sections changing graphics while the other sections remain in their non-moused-over state. However, in TOT powered Safari, once a section has been moused over it remains depressed until the user mouses over the next section in line, which returns the previous section to its un-moused-down state. However, once the user moves to a third section, the first section returns to its mouseover state. Eventually, by scrubbing up and down the navbar, all the sections will retain their mouseover state except the one immediately before (direction dependent) the currently moused over section. When no section is moused over (after all have been moused over at least once) all of the sections will retain their moused over state. I'm not sure that's very clear, but the behavior is obvious when you visit the site and try to use the navbar.
The regression happened before 2005-10-01 according to nightly builds, and it (or some version of it) is already present in WebKit-417.9
<rdar://problem/4403730>
I can't repro with the latest sources. Can anyone else?
I also can't reproduce. Did the site change? We probably need to close this bug.
I cannot reproduce this with WebKit-CVS-2005-12-14 20-06-29 GMT. Apparently, the site has indeed changed.
Me neither, closing this, no use to all look at it :)
Removing NeedsReduction. Alice, don't know if you want to remove the "InRadar" too?
(In reply to comment #4) > I also can't reproduce. Did the site change? I can't reproduce on the site either, but a copy I made of the site on Jan 4 (and slightly reduced) still looks buggy. Since this bug is a regression and the site always worked in Firefox, I think it's important to look at it. It's very likely that the site was changed soon after WebKit 417.9 was released, since the release contained this regression.
Created attachment 5788 [details] Reduced testcase When an element changes to display:none, it maintains (some?) of its hover state (and doesn't call its onmouseout handler). I think this is a regression from the 2005-07-29 checkin that changed :hover and :active behavior.
Good testcase Mitz! Upping severity to Major and changing NeedsReduction to HasReduction.
And forget to add regression keyword.
Created attachment 5801 [details] Update hover info when a renderer goes away
Comment on attachment 5801 [details] Update hover info when a renderer goes away Looks great, r=me.
Removing keyword(s) cause bug is closed.