229891 - Table cell has inconsistent length during multiple reloads

Status: RESOLVED WORKSFORME

(Core :: Layout: Tables, defect)

Description

[PikeUK]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7a) Gecko/20031231 Firebird/0.7+
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7a) Gecko/20031231 Firebird/0.7+

The cell containing the marquee at http://www.shadowgames.net/ is longer on some
reloads (234px), than on other reloads (204px).

Reproducible: Sometimes

Steps to Reproduce:
1. Visit http://www.shadowgames.net/
2. Reload page until problem appears.

Actual Results:  
After page loading, the cell containing the marquee can have one of two
different lengths.

Expected Results:  
After page loading, the cell containing the marquee should always be the same
length.

Originally reported:
http://forums.mozillazine.org/viewtopic.php?t=43281

The problem is not due to the server, the marquee or any scripting, see reduced
testcase in next comment.

Comment 1

[PikeUK]

Attachment: Test case (1st cell appears in 1 of 2 different sizes on reload)

The first cell sometimes appears to be 0 width, sometimes a few pixels wider.
The wider version appears more often when loading over the Internet than it
does when loading locally.

The title="AAA..." attributes are there to replace the long event handler
attributes in the original page since the bug doesn't seem to occur when they
are removed. I'm not sure if this indicates a problem with parsing large
attributes or if they are just causing the page to take longer to load (however
moving the attributes into comments so the page is the same size but without
the large attributes seems to prevent the bug from occurring).

Comment 2

[Bernd]

pike could you try add <script>var v = document.body.offsetHeight;</script>
somewhere in the script to make the behaviour more reproducible?

Comment 3

[PikeUK]

Attachment: Test case (using JS trick to always force wider 1st cell)

Testcase with <script type="text/javascript">var v =
document.body.offsetHeight;</script> added inside first table cell.

Comment 4

[Greg Campbell]

I can't reproduce this bug. I reloaded the site in the description and both
testcases dozens of times.

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7a) Gecko/20040101
Firebird/0.7+ (aebrahim)

Comment 5

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

This is trivial to reproduce: load attachment 138279 [details], and then text zoom in and
out, and notice the change.

Comment 6

[Andrew Schultz]

Attachment: simpler testcase

with the JS trick, the title attribute isn't needed:

<table border="1">
 <tr>
  <td><script type="text/javascript">document.body.offsetHeight;</script></td>
 </tr>
 <tr><td></td></tr>
</table>

Comment 7

[Andrew Schultz]

confirmed with linux trunk 2003123109

Comment 8

[Bernd]

Attachment: slightly modified testcase

Comment 9

[Bernd]

The testcase triggers the code at
http://lxr.mozilla.org/seamonkey/source/layout/html/table/src/nsTableCellFrame.cpp#917
. This code seems wrong to me and the intentions of this code are mysterious for me.

Comment 10

[PikeUK]

Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.8b2) Gecko/20050524 Firefox/1.0+

Using the above build I can no longer reproduce this issue (with any testcase or
using text zoom as described in comment 5). Can anyone else confirm that this
has been fixed?

Comment 11

[Andrew Schultz]

worksforme too with linux trunk 2005052401
resolving WFM