292405 - shorthand closed anchor element slurps in later content

Status: RESOLVED INVALID

(Firefox :: General, defect)

Description

[Zane Rockenbaugh]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.7) Gecko/20050414 Firefox/1.0.3
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.7) Gecko/20050414 Firefox/1.0.3

Using a shorthand closed anchor element (e.g., "<a href="foo" />) erroneously
slurps in later content, making the a part of the link. I found this while
trying to use generated content (using CSS) to create the links and later
content would show up as part of the link. It looks like's it's generating a new
element box because if you create a :before pseudo-element in the CSS, it gets
applied to the later text as well (see http://www.zanecorp.com/bug/index4.html).

Interposing another anchor element makes things act correctly. This is
demonstrated in http://www.zanecorp.com/bug/index3.html.

Reproducible: Always

Steps to Reproduce:
Go to provided URL
Actual Results:  
Page mis-rendered.

Expected Results:  
See http://www.zanecorp.com/bug/index2.html (which uses a non-shorthand anchor
element).

Comment 1

[James Ross]

The example page is being sent as text/html, so this is correct behaviour (in so
much as you aren't closing the tag). Please send the pages as an XML MIME format
if you want XML parsing to occur.

Comment 2

[Zane Rockenbaugh]

Changing the contexnt type to "application/xhtml+xml" does not fix problem; see
http://www.zanecorp.com/bug/index4.html. I have checked with both the latest
released versions of Firefox and Mozilla; both mis-render the page and report
the content type as application/xhtml+xml.

Comment 3

[James Ross]

That file is still being sent as text/html and Gecko (Moz/FF) are still treating
it as HTML only.

Do not forget that <meta> tags are trumped by HTTP headers, and since we have an
HTTP Content-Type header of text/html (to be precice, it is sending
"Content-Type: text/html; charset=ISO-8859-1")...

Comment 4

[Josh Birnbaum]

"View Selection Source" can be used to see what the parser is creating.  In this
case the input of:

<a href="http://www.mozilla.org" />
<div>Foo</div>

becomes:
<a href="http://www.mozilla.org">
</a><div><a href="http://www.mozilla.org">Foo</a></div>

(residual-style here is reopening the anchor tag because the <div> cannot be nested)

This is correct behavior.  Marking invalid.