226277 - regression in CSS rendering of overflow:hidden? -- pure CSS menus no longer work when there is layer behind with overflow:hidden

Status: RESOLVED WORKSFORME

(Core :: Layout, defect)

Description

[[e-mail]]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031007 Firebird/0.7
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031111 Firebird/0.7+

I'm using the Pure CSS menus with the IE workaround from
http://www.howtocreate.co.uk/tutorials/testMenu.html (based on Eric Meyer's Pure
CSS menus) to do a menu, but it's not compatible with recent builds of Mozilla
Firebird when there is another layer overlapping/underneath it.

Based on discussion at http://forums.mozillazine.org/viewtopic.php?p=260521 it
appears to be an issue with the overflow:hidden rule - removing that rule
restores the menu to its full glory.  however, there is no reason why
overflow:hidden should be related to the problem - hence there must be a bug
introduced sometime between 0.7 and the recent nightlies.

Reproducible: Always

Steps to Reproduce:
1.On a recent nightly, go to the page. (The test page is in the URL.)
2.On the menus/navigation (that's the blue thing!), go to projects --> web design.
3.Oops, you can't get there!
Actual Results:  
On a recent nightly build (such as Mozilla/5.0 (Windows; U; Windows NT 5.1;
en-US; rv:1.6b) Gecko/20031111 Firebird/0.7+), it's impossible to reach that
menu item becaues the menu won't remain open once it passes on top of the other
layer. 


Expected Results:  
On the released Mozilla Firebird 0.7 (Windows; U; Windows NT 5.1; en-US; rv:1.5)
Gecko/20031007 Firebird/0.7), it works fine - the menu remains open.

Comment 1

[Boris Zbarsky [:bzbarsky]]

We actually create a scrollframe for overflow:hidden now....  I bet that's
causing spurious mouseout events.  I further bet that using overflow:auto
instead of hidden won't work with the older builds _or_ the newer ones..

Comment 2

[[e-mail]]

What is the rationale behind the change in behaviour?
Will it be fixed?

Removing overflow:hidden -or- changing it to overflow:visible works - on both
Mozilla Firebird 0.7 and recent nightlies.  (But, obviously one wouldn't change
it to overflow:visible if he/she wanted it hidden in the first place)

I haven't tried overflow:auto though.

Comment 3

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

'overflow:hidden' was intenionally changed to be more like 'overflow: auto' and
'overflow:scroll' because that's what other browsers do and is thus the
direction in which the CSS spec has been clarified.  That change will not be
undone because the new behavior is correct.

If there were existing bugs that showed up with 'overflow: auto', they'd now
show up for 'overflow: hidden' as well.

Comment 4

[[e-mail]]

I see.  However, the effects it cause (mouseout/focus being lost) seems, to me,
to be a casualty of the change - it is affecting something that is unrelated and
therefore seems like a bug to me.

Will the effects on focus being lost be corrected?  Or should I be editing my
CSS styles to match the new behaviour?

Thanks.

Comment 5

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Why does it seem that way to you if you haven't actually tested whether the bug
was present with 'overflow: auto' in 1.5?

Comment 6

[[e-mail]]

Okay, I've tested it now.  (I'm not a software developer so I don't really know
how this works......)

Yup, problem in overflow:auto and overflow:hidden in 1.6b
And problem in overflow:auto only in 1.5
No problem with either when set to overflow:visible.
(Tested on Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031007)

See:
http://www.sichunlam.com/mozillaauto.html
http://www.sichunlam.com/mozillavisible.html

So, I guess this means it is part of the "existing bugs that showed up with
'overflow: auto'"?

Comment 7

[Boris Zbarsky [:bzbarsky]]

Attachment: Minimal-ish testcase

Note that the onmouseover handler of the overflow:auto div does NOT fire when
the onmouseout of the positioned div fires... Futher, the relatedTarget for the
mouseout is null.

Comment 8

[[e-mail]]

This is from Firebird 0.7...  It is most definitely fixed now.

Comment 9

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Nope.

Comment 10

[Eli Friedman]

The testcase seems to work fine for me (FF 2.0 and trunk); is there still an issue?

Comment 11

[Mike Shaver (:shaver emeritus)]

WFM.