48256 - Incorrect CSS1 color inheritance [CASCADE]

Status: VERIFIED FIXED

(Core :: DOM: HTML Parser, defect, P3)

Description

[Waldemar Horwat]

The attached HTML page renders as shown in the attached GIF using the M18 
apprunner.  The three links on the first line have the wrong color; they should 
appear as they do on the second line.  On both lines the innermost element 
enclosing the links' text is a SPAN with a class that sets the color to light or 
dark red.  The only difference between the two lines is the presence of an extra 
outermost SPAN on the first line that also sets the color to red.

Comment 1

[Waldemar Horwat]

Attachment: HTML source code

Comment 2

[Waldemar Horwat]

Attachment: Screen shot

Comment 3

[Marc Attinasi]

I think the markup is either illegal or the parser is doing something wrong. 

When I dump the content I see that the first case (with the outer SPAN) has 
Anchor -> span -> anchor -> text whereas the second case, without the outer 
span, has what I'd expect, anchor -> span -> text

Content Dump of the links =>

first case:
        a@026BE1F0 href=#Link1 refcount=3<
        >
        span@026BF7B8 class=type-name refcount=3<
          a@026BF660 href=#Link1 refcount=3<
            Text@026BF4F0 refcount=3<Type1>
          >
        >

Second case:
      a@026C6740 href=#Link1 refcount=3<
        span@026C6628 class=field-name refcount=3<
          Text@026C4F20 refcount=3<field2>
        >
      >

When I change the outer SPAN to a DIV it all works as the reporter expects. I'm 
not sure, but maybe the SPAN cannot contain another SPAN so the parser is 
reconstituting the markup to be legal (closing and reopening tags)? 

CC'ing harish for a sanity check on the content. Can a SPAN contain a SPAN? Is 
this a residual style problem?

Comment 4

[Hixie (not reading bugmail)]

According to the specs, a span can contain a span.

Comment 5

[Marc Attinasi]

Then this looks like a parser / residual style problem. Assigning to Harish to 
check out the way the content model is being formed for the two cases.

Comment 6

[Marc Attinasi]

Actually, looking at the markup some more I realize I asked the wrong 
question... The problem occurs when a SPAN contains an anchor. It looks like the 
A is closing the outer span, and then that span is being opened again inside of  
the A. In any case, it looks like a case of the parser trying to massage the 
content, and the style inheritance is changing from what was authored. 

Comment 7

[Nisheeth Ranjan]

Marking nsbeta3- and futuring.  Is this bug affecting high visibility sites?  If 
not, we can probably live with it till 6.1.

Comment 8

[Braden]

This appears to be a parser problem with <SPAN><A><SPAN></SPAN></A></SPAN>.
Adjusting component.

Note <news://news.mozilla.org/39EF5A4E.BF0605BF%40alias.it> for another
report/test case.

Comment 9

[Waldemar Horwat]

Nominating for 6.5.

Comment 10

[Heikki Toivonen (remove -bugzilla when emailing directly)]

r=heikki for the patch harishd showed me. Please add the fix here as well.

Comment 11

[harishd]

Attachment: Proposed patch v1.1

Comment 12

[harishd]

Fix landed. Marking FIXED.

Comment 13

[harishd]

marking fixed for real.

Comment 14

[bsharma]

updated qa contact.

Comment 15

[bsharma]

Verified on:
build: 2001-06-04-09-Mtrunk
platform: WinNT

The attached html and the screen shot are similar.