81997 - Changing CSS visibility with DOM causes some visibility problems.

Status: RESOLVED INVALID

(Core :: DOM: CSS Object Model, defect)

Description

[/\/\arcio Galli]

Please go to URL. Tested with 2001051704 and another today build. Win98 but I
believe this is also happening with other platforms/os...

-- the testcase: 

it's a gif image (logo) (visibility:hidden initial state)
with another png image (white) in the foreground. 

--  what's expected? 

When the document is loaded, the logo image is hidden then the foreground image
(png) is placed and you can't see the logo image. The onload event calls the go
JS function that sets the visibility of the logo image to "visible"... 

So, in a perfect world, you still can't see the image because it's behind the
white png. 

-- What's happening here... please doublecheck.. 

I can see the logo image, during some time (less than 1 second)... Seems like we
can see this logo image, exactly in the time  I did the call
document.getElementById('image').style.visibility="visible";.....

It's not happening in Netscape 6.0x ..

Comment 1

[/\/\arcio Galli]

Need help to confirm... marking regression... 

Note: the problem happen in the first time. After this, you need do start the
browser again to see the problem. 

Comment 2

[/\/\arcio Galli]

Adding a new testcase: 

This new testcase, the foreground image have white background... not a png
image...seems like this is working.... :-/

http://geckonnection.com/test-features/visibility/test2.html

Comment 3

[/\/\arcio Galli]

Adding new test:

http://geckonnection.com/test-features/visibility/test3.html

The png image is not in the background-image.. is inside the DIV. This case
works too. 

Comment 4

[Johnny Stenback (:jst)]

Adding z-index: 5; to the style for the logo makes it show up, I don't think
this is a bug, it's probably a bug in Netscape 6.0 that made this work. None of
the testcases in this bug works for me, btw.

Comment 5

[/\/\arcio Galli]

Johnny, 

Maybe I was not so clear... the right behavior is to not display nothing..

adding roger,.. cc.. Roger, can you reproduce this? 

Comment 6

[Johnny Stenback (:jst)]

Oh.

Comment 7

[Jason Johnston]

Ah, it looks like we're not waiting for images specified in CSS background-image
declarations before firing onload for the document.  That would explain the
behavior: document starts to load -> body and its descendants start to load ->
all descendants (html markup and logo.gif) load -> onload event for <body>
fires, n6logo is made visible -> fadevert4.png (specified in style attr)
finishes loading, finally displaying and covering up the N6 logo.

Unsure as to whether this is correct behavior or not.  I'm inclined to say that
it is, because:
The background-image specified through style is not actually a descendant of the
<body> element (it's a property of the View (?)correct me if I'm wrong), so in
that sense <body> is actually being fully loaded before firing onload.

Thoughts from the gurus?
--J

Comment 8

[Johnny Stenback (:jst)]

Post-mozilla1.0, unless someone can convince us we must fix this for mozilla1.0.

Comment 9

[Johnny Stenback (:jst)]

Assuming it really is a bug, that is. Ian, any thoughts?

Comment 10

[Troy Zirk]

This is happening with me. I am redesigning our company's intranet and I've run 
across this. I'm using the following code to view a <div>:

document.getElementById("div").style.visibility = "visible";

This works in NN6.0 but NOT in NN6.1 OR NN6.2.1 (most recent version was 
downloaded and installed 2002-03-05 at 1pm)

Comment 11

[Hixie (not reading bugmail)]

I could reproduce this once. Only once.

Which leads me to suspect that this isn't a bug, it's just the image loading...

Comment 12

[José Jeria]

Check out
http://geckonnection.com/test-features/visibility/test3.html

The png file is not shown. See Page Info / Media.
The png is visible in IE

Comment 13

[Johnny Stenback (:jst)]

Mass-reassigning bugs to [e-mail]

Comment 14

[Boris Zbarsky [:bzbarsky]]

Yeah, this is invalid.  As long as background images don't block onload (which
they should not), it will be possible to write a testcase like this where a
background image load races with onload.