368603 - JS marquee on page keeps expanding

Status: RESOLVED WORKSFORME

(Core :: Layout: Tables, defect, P3)

Description

[Stephanie Daugherty [:sdaugherty]]

Not sure exactly when this regressed, but in Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a2pre) Gecko/20070128 Minefield/3.0a2pre, and according to Aleksej on IRC, Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9a2pre) Gecko/20070123 Minefield/3.0a2pre as well, the Javascript marquee expands constantly on http://www.powerball.com/pb_info.asp making the page unusable.
This doesn't happen in 2.0.0.1, and seems to be a recent regression. I'll try to narrow down the regression window later.

Comment 1

[Martijn Wargers (dead)]

Attachment: partly minimised testcase

Comment 2

[Martijn Wargers (dead)]

This regressed between 2006-12-07 and 2006-12-08, so somehow a regression from the reflow branch landing.

Comment 4

[Jesse Ruderman]

The dup was blocking1.9+ and had a pair of reduced testcases.

Comment 5

[Daniel Holbert [:dholbert]]

Attachment: More minimized testcase

Further minimized version of a testcase taken from the duplicate Bug 380861.

I've stripped out the JS marquee functionality and replaced it with something simpler that has the same bad vertical-growth effect.  Namely, I use a button to append the empty string to a div's HTML.  This shouldn't cause any visible change, but in fact it seems to trigger a reflow that makes the div grow vertically.

Interestingly, if you change "tweak()" in the test case to append a visible character instead of the empty string, the problem doesn't occur.

Comment 6

[Daniel Holbert [:dholbert]]

The 2 posted testcases seem to have become WorksForMe between the nightlies for 2007-06-24 and 2007-06-26.  (probably fixed by patch for bug 370525 landing, since this bug seems to be related to special height reflows)

However, the original page ( http://www.powerball.com/pb_info.asp ) still shows buggy behavior *while it's loading*, if your browser is wider than 800px.  (buggy behavior =  marquee grows vertically.)  It also seems to show the behavior if you resize the browser from below 800px to over 800 px while the marquee scrolls.  

(800px is rough -- I'm not sure if that's the exact threshold)

Working on getting a new reduced tescase that shows the buggy behavior.

Comment 7

[Daniel Holbert [:dholbert]]

Comment on attachment 254939 [details]
partly minimised testcase

Obsoleting because this testcase was fixed by patch for bug 370525

Comment 8

[Daniel Holbert [:dholbert]]

Comment on attachment 266960 [details]
More minimized testcase

Obsoleting because this testcase was fixed by patch for bug 370525

Comment 9

[Daniel Holbert [:dholbert]]

Attachment: new testcase

(In reply to comment #6)

Posting a new reduced testcase of the remaining buggy behavior from the powerball website.

Comment 10

[Daniel Holbert [:dholbert]]

Note: The new testcase causes lots of these assertions:

---- 
###!!! ASSERTION: invalid divisor: 'Error', file /scratch/work/builds/trunk.07-07-02.13-20/mozilla/layout/tables/nsTableFrame.cpp, line 3279
----

This assertion already has a bug filed: bug 284844.

Comment 11

[Andrew Smith]

Ah why not, taking bug.

Comment 12

[Andrew Smith]

It doesn't look like I'll have the time to work on this after all. Sorry.

I haven't found anything useful during the time I did spend on it.

Comment 13

[Bernd]

wfm? with current trunk?

Comment 14

[Ryan VanderMeulen [:RyanVM]]

WFM as well. I guess Dave should be the one to make the final decision since he wrote the testcases. Powerball site works fine too.

Comment 15

[Ryan VanderMeulen [:RyanVM]]

ugh, s/Dave/Dan (sorry, just woke up when I posted that)

Comment 16

[Daniel Holbert [:dholbert]]

WFM too, on all testcases and on the Powerball page.  Great!

Comment 17

[Daniel Holbert [:dholbert]]

oops, wrong resolution flag.
 --> WFM

Comment 18

[Daniel Holbert [:dholbert]]

I just noticed that "new testcase" is broken (grows on button-click) in my mozilla-central nightly, but is fine in Firefox 3.0.9.

The original powerball website and first two testcases are fine, though, so I think I'll file a new bug on the final testcase, rather than reopening this bug.

Comment 19

[Daniel Holbert [:dholbert]]

filed bug 489906 for comment 18.