204283 - UMR in CNavDTD.cpp - enabled not set by nsHTMLFragmentContentSink::IsEnabled()

Status: VERIFIED DUPLICATE of bug 230536

(Core :: DOM: HTML Parser, defect)

Description

[Andrew Schultz]

CNavDTD::WillBuildModel calls mSink->IsEnabled and expects it to set |enabled|,
and while nsHTMLContentSink::IsEnabled will always set it,
nsHTMLFragmentContentSink::IsEnabled never sets it (it just returns NS_OK). 
Valgrind complains about this.

http://lxr.mozilla.org/mozilla/source/htmlparser/src/CNavDTD.cpp#427

I'll attach a simple testcase that triggers this.  This occurs with today's
linux CVS trunk.

Comment 1

[Andrew Schultz]

Attachment: testcase

Comment 2

[Kathleen :Brade]

hmmmm....
I tried this fix but it didn't fix bug 228688 so maybe this isn't the cause of
228688 (directly) [the timing doesn't seem right anyways]:

Index: nsHTMLFragmentContentSink.cpp
===================================================================
RCS file: /cvsroot/mozilla/content/html/document/src/nsHTMLFragmentContentSink.cpp,v
retrieving revision 1.79
diff -r1.79 nsHTMLFragmentContentSink.cpp
102c102
<   NS_IMETHOD IsEnabled(PRInt32 aTag, PRBool* aReturn) { return NS_OK; }
---
>   NS_IMETHOD IsEnabled(PRInt32 aTag, PRBool* aReturn) { *aReturn = PR_FALSE;
return NS_OK; }

Comment 3

[chris hofmann]

so if not related to the problem in bug 228688 , should the 1.6 nomination hang
around...  other than Valgrind errors are we seeing places where bad things are
happinging...

also not clear from the comments...  did errors just show up on 2003-05-02 or
were they around before?

Comment 4

[Andrew Schultz]

> other than Valgrind errors are we seeing places where bad things are
> happinging.

I don't know of any other symptom for this bug other than the Valgrind error
(there are no user-visible symptoms).

> also not clear from the comments...  did errors just show up on 2003-05-02 or
> were they around before?

I only noticed the problem on 5/2/03.  I don't know if it started before that or
not (valgrind doesn't give much useful info from a stripped (default nightly)
build).  That is before bug 228688 started happening (7/1/03).

Comment 5

[chris hofmann]

ok, thanks. sounds like 1.6-

Comment 6

[Andrew Schultz]

*** This bug has been marked as a duplicate of 230536 ***

Comment 7

[Kathleen :Brade]

v