86096 - Rendering not reflecting DOM under certain circumstances

Status: RESOLVED WORKSFORME

(Core :: Layout, defect)

Description

[Stephen Harris]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux 2.4.2-2smp i686; en-US; rv:0.9.1+)
Gecko/20010613
BuildID:    2001061308

The page has some JavaScript which changes the DOM when one method is called and
changes it back when another is called. After the DOM has been changed and
changed back the browser no longer renders the DOM correctly.

Reproducible: Always

Steps to Reproduce:
1. Click on "Highlight"
2. Click on "Unhighlight"

The grey box at the bottom of the page is supposed to show the DOM in its
current state.

Actual Results:  After the yellow highlight is removed the portion of the FONT
element after the word "Sun" refuses to render. Resizing the window etc. has no
effect.

Expected Results:  The word "Sun" should be highlighted when Highlight is
clicked by adding a SPAN arround the word and the SPAN should be removed when
Unhighligh is clicked.

After one cycle the browser should look as it did at load time. The
Highligh/Unhighlight cycle should be repeatable with no side effects.

This has been broken since 0.7 or 0.8, but I have only just managed to repoduce
it reliably.

Removing the FONT element or replacing the call to normalize() on line 28 of
xuldlstest6.js with some equivalent javascript avoids the bug or hides the
symptoms, eg. http://inanna.ecs.soton.ac.uk/~swh/xuldlstest7.html

I have tested the code on IE5.5 and it works there (with some modifications).

Comment 1

[Johnny Stenback (:jst)]

I dug into this a bit and it turns out to be a problem with layout, and not with
the DOM code, the content model is correct after clickin Highlight and
Unhighlight, but layout isn't showing the correct text, the word "Sun" is
incorrectly duplicated and " Microsystems, call:" is not displayed at all.
AFAICT the DOM code is giving the correct notifications when the modifications
are made to the content model.

Waterson, is this something you wanto have a look at? If not, reassign to
whomever this belongs to.

Comment 2

[Stephen Harris]

I was wrong about the normalize code, replacing the call to normalize with some
javascript doesn't help, neither does replacing the FONT element with a SPAN for
e.g.

Comment 3

[Jason Johnston]

The testcase links in this bug report are broken.  Stephen, could you please
check if this bug still occurs in a recent build?

If it still occurs and you still have the testcases could you please attach them
to this bug report using
http://bugzilla.mozilla.org/attachment.cgi?bugid=86096&action=enter

Comment 4

[Stephen Harris]

Sorry, I accidentally deleted the testcase and removed this code from my software.

The testcase isn't in google's cache either :( 

Comment 5

[Sivakiran Tummala]

reassigning to default owner/QA

Comment 6

[Chris Waterson]

Taking back.

Comment 7

[Boris Zbarsky [:bzbarsky]]

Marking worksforme due to inability to reconstruct the testcase.... if there is
ever a testcase that does show this bug, please attach it directly to the bug.