44505 - cell padding depends from table nesting for table-layout:fixed

Status: VERIFIED FIXED

Description

[Matthias Kerkhoff]

From Bugzilla Helper:
User-Agent: Mozilla/4.73 [en] (WinNT; U)
BuildID:    2000070320

Tables using the fixed layout algorithm (table-layout:fixed) seem to ignore the 
padding on all cells in a row (except on the first one).

Reproducible: Always
Steps to Reproduce:
1. Create a table with one row and some cells
2. Set css widths and paddings for each cell
3. Load the table

Actual Results:  The paddding is ignored for all cells except the first one.

Expected Results:  The padding should be applied to all cells where it has been 
specified.							

padding works fine when used with other table layout algorithms.

Comment 1

[Matthias Kerkhoff]

Created attachment 10965 [details]
testcase for 44505 and 44506

Comment 2

[Matthias Kerkhoff]

Bug seems to be related to nested tables. Couldn't reproduce with plain tables.

Comment 3

[Matthias Kerkhoff]

Created attachment 10966 [details]
Corrected testcase

Comment 4

[Bernd Mielke]

Created attachment 11102 [details]
reduced testcase

Comment 5

[Bernd Mielke]

The padding seems to depend on the table nesting (see reduced testcase).
Confirming the bug. Changing summary

Comment 6

[Bernd]

Created attachment 24381 [details] [diff] [review]
patch

Comment 7

[Bernd]

Created attachment 24382 [details]
testcase

Comment 8

[Bernd]

Created attachment 24383 [details]
testcase with image

Comment 9

[Bernd]

Created attachment 25178 [details] [diff] [review]
revised patch (new style handling)

Comment 10

[Bernd]

Finally, I managed to verify that the patch does not not break the table
regression tests. (It would be also very difficult, as we do not have any
testcase in the regression tests that tests for clipping ... this is a ping to
the QA engineer). The patch enables the overflow:hidden from CSS2 for table
cells. I think with the correct handling of the padding we outperfom with this
patch IE5. They have overflow:hidden for the fixed layout (you cannot make it
visible) and they do not handle padding correctly. This patch does not enable
overflow:scroll and overflow:auto, but I think it improves anyway our CSS2
compliance.

Comment 11

[Marc Attinasi]

sr=attinasi - the patch looks fine. Can we add the testcase to the table
regression tests? Generally, Chris K. adds them and not a QA person. As you
mentioned, Bernd, we need a testcase in the regression suite to catch this case
and none currently exist. Thanks.

Comment 12

[Bernd]

> Generally, Chris K. adds them and not a QA person.

But should this prevent [e-mail] to write something comparable to
marvin for CSS2 table stuff, I think we need it desperately.

Comment 13

[Bernd]

The testcase renders now fine. fix checked in

Comment 14

[Bernd]

Please notice that the even the corrected testcase has an omitted padding
attribute for the third cell. 
	<TD nowrap style="text-align:left; 0px 5px 0px 5px; white-space:nowrap;
width:45px">45px</TD>

Comment 15

[Amarendra Hanumanula]

QA contact update

Comment 16

[Amarendra Hanumanula]

 The attached testcase renders fine