145572 - [ABS] abs. pos. table does not honour the right: xxpx

Status: RESOLVED FIXED

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

Description

[Bernd]

A absolute positioned table with a percent width does not honour the right
position description. win98 2002051508

Comment 1

[Bernd]

Attachment: testcase with left:200px

Comment 2

[Bernd]

Attachment: testcase with right: 200px

Comment 3

[Bernd]

the issue has nothing to do with percent tables. It's a consequence of our
bizarre style context relationship between outertable and table. At
nsHTMLReflowState::InitAbsoluteConstraints we look up the containing block left
offset and not the ParentStyleContextProvider
(http://lxr.mozilla.org/seamonkey/source/layout/html/base/src/nsHTMLReflowState.cpp#1025)

As the left offset is not specified it has the value auto, at html.css we have
left:inherit for the outer-table so it should inherit the value from table but
nsHTMLReflowState.cpp the containing block is looked up.

Comment 4

[Amarendra Hanumanula]

 right:200px in the testcase is not honoured even on WIN2K. Adding testcase 
keyword.

Comment 5

[karnaze (gone)]

mass reassign to default owner

Comment 6

[Melvyn Sopacua]

This can be easily worked around with divs around the table, but it's annoying
extra markup, especially when creating right aligned menu's or RSS data dumps.
Also applies to 'fixed' positioning.

Comment 7

[Ger Teunis]

This bug looks fixed, but as you can see, the second div (blue foo) is not align
200px from the right, but less. We are almost there!

Comment 8

[Ger Teunis]

*** Bug 220906 has been marked as a duplicate of this bug. ***

Comment 9

[Bernd]

Attachment: patch

width:0 is a hack
see
http://bonsai.mozilla.org/cvsview2.cgi?diff_mode=context&whitespace_mode=show&file=html.css&branch=&root=/cvsroot&subdir=mozilla/layout/html/document/src&command=DIFF_FRAMESET&rev1=3.69&rev2=3.70

this comment was removed during the cleanup in revision 109 but is IMHO very
true.

Comment 10

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

What about removing the 'height: 0'?

Comment 11

[Bernd]

Attachment: patch

This patch does not generate block size errors in the layout regression tests
(to my big surprise)
It generates however style mismatches, but thats what I would expect.
With this patch moz will render
http://www.hixie.ch/tests/adhoc/css/box/absolute/008.html correctly.
I did test the xul tree problems within a sidebar in bug 37643 and bug 37663
that led to this hack and did not see a cpu cycle eating.

Comment 12

[Ger Teunis]

Is this bug fixed now? I will be very happy when this bug is fixed!

"It generates however style mismatches, but thats what I would expect."
Does this result in incorrect rendering pages?

Comment 13

[Bernd]

Ger this has not checked in yet as the tree is closed. I will check it in in the
1.7alpha and bake it a little bit. If it doesn't cause regressions I will ask
for a 1.6 permission. In the meantime you can apply this patch to your local
copy of mozilla or fb by modifing the ua.css as shown in the patch.

The width of the outer table frame has changed from 0px to auto this is a
stylistic change but it doesnt create changes in the box parameters as tables
never even looked what width has been specified for the outer table frame which
is more a placeholder than a real thing.

Comment 14

[Bernd]

taking care

Comment 15

[Brendan Eich [:brendan]]

Comment on attachment 137002 [details] [diff] [review]
patch

Sure, what can go wrong?

/be

Comment 16

[Bernd]

fix checked in

Comment 17

[Bernd]

*** Bug 229088 has been marked as a duplicate of this bug. ***