Closed Bug 378413 Opened 13 years ago Closed 12 years ago

Hang with display: -moz-deck, direction: rtl, fieldset

Categories

(Core :: Layout, defect, critical)

x86
macOS
defect
Not set
critical

Tracking

()

RESOLVED WORKSFORME

People

(Reporter: jruderman, Unassigned)

References

(Blocks 1 open bug)

Details

(Keywords: hang, rtl, testcase)

Attachments

(1 file)

This testcase makes Firefox (Mac trunk debug) hang, spewing the following assertion the whole time:

###!!! ASSERTION: Bogus availSize.width; should be bigger: '!mContentFrame || GetContentMinWidth(aReflowState.rendContext) <= availSize.width', file /Users/jruderman/trunk/mozilla/layout/forms/nsFieldSetFrame.cpp, line 474
I have no idea what the deal is with the width, offhand, but the hang is due to height growth.  nsStackLayout keeps looping while the kid is growing, and each time it sets a computed height on the fieldset, the fieldset reflows at that height, overflows (because the content frame is height: 100% and the legend takes some space), nsFrame::BoxReflow does:

  aDesiredSize.height = aDesiredSize.mOverflowArea.YMost();

which increases aDesiredSize.height from what it used to be (which was equal to the computed height).

So we do this over and over again forever.
Still hangs on trunk.
Is there a reason to not close it?
I haven't tested it on mac and I know that Jesse normally reacts quite quickly to comments (and maybe someone is interested in what fixed it).
Works for me on Mac trunk debug.
Works for me, too.  No crash, no assertion failures.
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → WORKSFORME
Flags: in-testsuite?
Mass-assigning the new rtl keyword to RTL-related (see bug 349193).
Keywords: rtl
Crashtest: http://hg.mozilla.org/mozilla-central/rev/f4622261cce8
Flags: in-testsuite? → in-testsuite+
You need to log in before you can comment on or make changes to this bug.