175070 - faulty table cell resize after background style attribute change

Status: RESOLVED WORKSFORME

(Core :: Layout: Tables, defect, P2)

Description

[Herman van Rink]

When using statements like below, the table cell is resized to the minimum
required by the contained text.

document.getElementById('foo').style.background='#fcc';
document.getElementById('foo').width='110';

<table>
<tr><td id=foo>Bar</td></tr>
</table>

Comment 1

[Phil Schwartau]

Attachment: Reporter's testcase

Comment 2

[Phil Schwartau]

Browser, not engine ---> DOM Style

The testcase WORKSFORME with Mozilla trunk 20021015xx on WinNT, Linux.
I cannot confirm this bug. The behavior is identical to IE6, as well.

Herman: what build ID of Mozilla are you using? Please include that
in any bug report; thanks -

Comment 3

[Herman van Rink]

Attachment: Not working situation

After tairing my page appart this is the minimum where is still goes wrong.
Removing the align=right in the inner table tag solves the problem.
The fully implemented page is http://www.wijnstravastgoed.nl/beta

Comment 4

[Herman van Rink]

Using:
Mozilla 1.1
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.1) Gecko/20020826

Comment 5

[Phil Schwartau]

Confirming bug with Mozilla trunk binary 20021015xx on WinNT.
OS: Linux ---> All.

When I click the button in Herman's testcase, the table remains
the same width in IE6, but collapses in Mozilla to fit the text.
All that the JavaScript is changing is the background color:


<script>
    function test()
    {
        document.getElementById('foo').style.background='#fcc';
    }
</script>

<table border=1>
    <tr><td>
        <table border=1 align=right>
            <tr><td id=foo width=600 height=40 class=all>Bar</td></tr>
            <tr><td>&nbsp;hhhh</td></tr>
        </table>
    </td></tr>

</table>
<input type="button" value="set background and width" onClick="test();">

Comment 6

[Boris Zbarsky [:bzbarsky]]

This is a table reflow problem.

Comment 7

[Bernd]

Attachment: reduced testcase

Comment 8

[Bernd]

The change of the background color causes a reframe because the
background-attachment for the table cell has the initial value 0 css_unit_null
and it changes to 0 css_enumerated. This throws in a new anonymous colframe
which blows up the table width calculations.

Comment 9

[Andrew Schultz]

regression between linux trunk builds 2002050902 and 2002051010
potential candidates are bug 120107 and bug 141021

Comment 10

[Bernd]

Attachment: testcase with background-attach

while attachment 107246 [details] can be fixed by the style system bug 171830 this
testcase causes a reframe inside the style system. Due to the reframe a
colframe will be initialized,( placing a break point at
http://lxr.mozilla.org/seamonkey/search?string=nsTableColFrame%3A%3AResetSizing
shows that perfectly ) There may be several possibilities to fix the bug a) fix
the reframe which seems to me the source  b) fix the reflow as it should be
able to handle this.

Comment 11

[Kevin McCluskey (gone)]

Bulk setting karnaze's P2-P5 bugs to future

Comment 12

[Boris Zbarsky [:bzbarsky]]

This should now be fixed... please retest (modulo that last testcase; can't tell
what that's testing).

Comment 13

[Daniel Wang]

Comment on attachment 107280 [details]
testcase with background-attach

2nd and 3rd testcase (attachment 103241 [details] and attachment 107246 [details]) still
show problems on 2003030609 win build. 4th testcase (attachment 107280 [details]
is worksforme)

Comment 14

[Andrew Schultz]

all 4 testcases workforme with linux trunk 20030307 (gotta test with a build
after 20030306)

Comment 15

[Daniel Wang]

all testcases worksforme moz 2003031408/win98