197022 - inline or table display-model text nested in two postion:absolute container do not wrap properly

Status: RESOLVED FIXED

(Core :: Layout: Block and Inline, defect, P2)

Description

[Michael]

User-Agent:       Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.3) Gecko/20030308
Build Identifier: Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.3) Gecko/20030308

Moz has major alignment problems when displaying
http://www.flightinternational.com/fi_home.asp. First of all the navigation menu
is located at the lower right of the screen (you have to scroll), not centered
at the top like in IE or Safari. The real problem is that a number of links
right to the "This Issue" link and in the row below are unreadably compressed to
the left of the navigation menu. In IE and Safari they appear chart like.

Reproducible: Always

Steps to Reproduce:

Comment 1

[Boris Zbarsky [:bzbarsky]]

Sounds like IE-specific JS.... but someone needs to testcase this.

Comment 2

[Madhur Bhatia]

see it on winXP too 

Comment 3

[Boris Zbarsky [:bzbarsky]]

Please do NOT confirm qawanted bugs without actually doing some QA and producing
a testcase.  I'll bet you money that this is NOT a Mozilla bug.

Comment 4

[Dean Peterson]

I think the biggest problem may be a Mozilla table layout bug.  The dropdown
menus in the center of the page are 1-row tables in absolutely positioned DIV's.
 When the page loads, javascript moves the DIV's with those submenu tables to
the same spot as the menu headings, which they hide behind.  (When the mouse
moves over a menu heading, javascript moves the submenu out from behind the
headings.) 

In Mozilla, when javascript positions the submenu DIV's, the table cell contents
wrap oddly, making the wrapped cell contents stick out from under the menu
headings that should hide them.  This gives the menus the garbled appearance.

The cell wrapping happens when the DIV stylesheet "top" property is set in
javascript.  It only happens if the table is nested more than 1 deep within
absolutely positioned DIV's.  The DIV whose "top" property is set must be the
innermost absolutely positioned ancestor of the table.  If you stick another
absolutely positioned DIV in the tree structure under the positioned DIV but
above the table, the effect doesn't happen.

I will add 3 test cases.  Test case 1 is the main one, showing the above
behavior.  Test cases 2 and 3 show possibly-related behaviors that also happen
only when a table is nested within multiple absolutely positioned DIV's.

There may be other separate problems on the page, too.  I will look at the
little Announcement section in the upper right of the page if I have time.

Comment 5

[Dean Peterson]

Attachment: Test case 1 (see comment 4)

Comment 6

[Dean Peterson]

Attachment: Test case 2 (see comment 4)

Comment 7

[Dean Peterson]

Attachment: Test case 3 (see comment 4)

Comment 8

[Boris Zbarsky [:bzbarsky]]

bernd, what was that bug on tables in nested positioned divs?

Comment 9

[Dean Peterson]

Attachment: Test case 4

The small "Announcements" section toward the upper left of the page has a
formatting problem that also appears to be related to nested absolute DIV's.  

Test Case 4 just shows a long cell continuing off the edge of a fixed width
table instead of wrapping, but only if the table is in multiple absolutely
positioned DIV's.

Comment 10

[Bernd]

bug 201001

Comment 11

[Daniel Wang]

Attachment: testcase, showing table & inline display model problem

display:inline and table elements do not wrap properly in double nested
postion:absolute container.

Comment 12

[Ryan Wood]

I'm pretty sure I have this bug on my page also, but its very intermittent.
Check out http://www.atl3.com/?page=archive for an example of this. If you dont
see anything wrong the first time you load, click refresh and you will see a
problem. At first i thought the problem was php, until i compared the source of
a broken page to that of a normal one. Both pages were exactly the same. Ive
seen this problem on on a few pages.

THe version I've seen isnt always reproducable though.

Comment 13

[Mats Palmgren (inactive)]

-> FIXED (by bug 201897)