112015 - Absolute position span inside relative position span not painted correctly

Status: RESOLVED DUPLICATE of bug 79315

(Core :: Layout, defect, P3)

Description

[Stephen Sottong]

The eval() function in javascript does not work in Netscape 6.2.  It worked in 
earlier versions but seems to have stopped in 6.2.  eval does not generate a 
javascript error, it just does nothing.

Comment 1

[Boris Zbarsky [:bzbarsky]]

works fine here... (linux build 2001-11-23-08).  How are you using it, exactly?

Comment 2

[Phil Schwartau]

Stephen: without an example, testcase, or URL to go to, we'll have
to mark this one as invalid. Note the "Create a New Attachment"
link above if you have a (reduced) testcase. Otherwise, a URL to 
a testcase or website can be provided in the "URL" textbox above. 

Comment 3

[Phil Schwartau]

Stephen has provided me with the URL above.

At this site, eval() is used to dynamically assign the CSS visiblity
of <SPAN> elements, as in:

  eval("document.getElementById(\"inner6\").style.visiblity=\"hidden\"")


and this is not working in Mozilla. Confirming with Mozilla trunk binary
200111265xx on WinNT. If you load the site in IE6, notice what happens
when you click the option buttons, like "Title" and "Author". Each time
you click one, look at "Enter search terms(s)". To the right of it,
you see dynamically assigned text, e.g. "example shakespeare,william".

That is, in IE6 you see it. In Mozilla you don't. No errors appear
in the JavaScript Console. The key functions using eval() are :


function ShowEx(iVis)
{
 for (i=1;i<=5;i++)
 {
  eval(layerRef + "inner" + i + styleRef + "visibility=\"hidden\"")
 }
 eval(layerRef + "inner" + iVis + styleRef + "visibility=\"visible\"")
}


function TabShow(iVis)
{
 for (i=1;i<=6;i++)
 {
  eval(layerRef2 + "tabinner" + i + styleRef + "visibility=\"hidden\"")
 }
 eval(layerRef2 + "tabinner" + iVis + styleRef + "visibility=\"visible\"")
}



And the eval strings here are like the one I gave above:

  eval("document.getElementById(\"inner6\").style.visiblity=\"hidden\"")

Comment 4

[Phil Schwartau]

Reassigning to DOM Style - 

Comment 5

[Boris Zbarsky [:bzbarsky]]

Phil, thanks for the data!

DOM style is working correctly.  Layout is not.  The problem is an absolutely
positioned span (block) inside a relatively positioned span (inline).

Attaching a minimal non-dynamic testcase.

Comment 6

[Boris Zbarsky [:bzbarsky]]

reassign for real

Comment 7

[Boris Zbarsky [:bzbarsky]]

Attachment: You should see text ("This is a sentence.")

Comment 8

[Boris Zbarsky [:bzbarsky]]

Attachment: giving the outer span a border makes things work for some reason...

Comment 9

[Johnny Stenback (:jst)]

Attachment: Simplified testcase

Comment 10

[rubydoo123]

setting to 9.9

Comment 11

[Stephen Sottong]

I can confirm that giving the outer layers a border:

#outer {position: relative; vertical-align: top; border-color: #FAF0E6;
border-style: solid;}

#tabouter {position: relative; vertical-align: top; border-color: #FFCE63;
border-style: solid;}

makes the page work.  The border-style cannot be "none".

Comment 12

[Stephen Sottong]

Congrats.  Version 0.9.9 seems to have fixed it.

Comment 13

[Boris Zbarsky [:bzbarsky]]

Sorry, reopening.  The testcases in this bug still fail for me with a current
tip build on Linux....

Comment 14

[Stephen Sottong]

After additional testing, I'm in agreement with Boris -- the problem is NOT fixed.

Comment 15

[Stephen Sottong]

If I change the outer, container (#outer and #tabouter in the example) from a
<span> to a <div>, this also makes it work with Mozilla, so the problem must
have something to do with handling of <span>'s.  I'd rather not have to do this
since it screws up the spacing.

Comment 16

[Stephen Sottong]

The problem is still present in 1.0 RC1.

Comment 17

[Stephen Sottong]

It's disappointing, the final release of 1.0 is out and it still has the bug.

Comment 18

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

Maybe a duplicate of bug 79315.

Comment 19

[Benjamin Smedberg]

Dup, and this was fixed. Yay!

*** This bug has been marked as a duplicate of 79315 ***