354458 - Hang with -moz-column-*, height, and float

Status: RESOLVED WORKSFORME

(Core :: Layout, defect)

Description

[Jesse Ruderman]

Loading this testcase hangs Firefox.

Comment 1

[Jesse Ruderman]

Attachment: testcase

Comment 2

[Jesse Ruderman]

Still hangs on trunk (after reflow branch landing).

Comment 3

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Attachment: another testcase

Just ran into this bug independently (while trying to find a testcase that splits images).

Comment 4

[Mike Schroepfer]

Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9a6pre) Gecko/20070613 Minefield/3.0a6pre

Confirmed still affects trunk..

Comment 5

[Eli Friedman]

WORKSFORME?

Comment 6

[Jesse Ruderman]

Worksforme (no hang) with both testcases, Mac trunk debug.

The second testcase gives me a warning the first time I load it, or when I shift+reload it:

WARNING: This situation currently leads to data not printing: 'NS_FRAME_IS_TRUNCATED(reflowStatus) && aForceFit', file mozilla/layout/generic/nsBlockReflowState.cpp, line 1007

Comment 7

[Eli Friedman]

I added that warning pretty recently.  It indicates a place where we could improve the pagination of floats.  (As you can see from viewing the testcase, the float is getting cut off.)

Comment 8

[Ryan VanderMeulen [:RyanVM]]

The second testcase still hangs me for a short time when I load it on a current trunk build

Comment 9

[Daniel Holbert [:dholbert]]

WORKSFORME -- I get no hang from either testcase using a current build.  Tested both on Windows and Linux.  

Talked to RyanVM on IRC, and he said he also no longer gets the hang.

Jesse, can I (or whoever) close this as WORKSFORME?


One other interesting point, possibly deserving of a separate (not blocking1.9) bug:

(In reply to comment #6)
> The second testcase gives me a warning the first time I load it, or when I
> shift+reload it:

Strange -- with a current build, I get the opposite behavior. On first load / shift+reload, I get no warning, but on subsequent reloads, I get the warning.  
Also (more importantly), the image doesn't appear on the first load / shift+reload.   I observed this behavior on both Linux and WinXP SP2.

Comment 10

[Jesse Ruderman]

WORKSFORME.  I see the same thing as Daniel.

I think the "This situation currently leads to data not printing" warning is covered by other bugs.

I filed bug 387075 for the layout inconsistency.

Comment 11

[Jesse Ruderman]

Crashtests checked in.