49794 - NS_ScriptErrorReporter should fall back to message when ucmessage is null

Status: RESOLVED WORKSFORME

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

Description

[Mike McCabe]

From netscape.public.mozilla.dom:
From: Arne de Bruijn <[e-mail]>
Subject: empty javascript errors


Access to properties of documents on another domain sometimes give empty
javascript errors.

If I understand things correctly this is caused by
js/src/jscntxt.c:js_ReportErrorVA not filling in the ucmessage field in
the JSErrorReport structure, but
dom/src/base/nsJSEnvironment.cpp:NS_ScriptErrorReporter expecting it is.

I don't known if this should be fixed in js_ReportErrorVA or in
NS_ScriptErrorReporter, but it worked for me after I changed
NS_ScriptErrorReporter to use the message argument if the ucmessage
field is NULL (very simple patch available on request :-)

Arne

Comment 1

[Rann Bar-On]

if you have a fix in hand, attach it when you file the bug please.  Thanks for
flying bugzilla ;).

Comment 2

[Mike McCabe]

Attachment: Make the DOM error reporter tolerate null ucmessage, filename.

Comment 3

[Mike McCabe]

I've asked Arne to add himself to the CC list...

Under what circumstances does js_ReportErrorVA produces a null ucmessage field? 
We should probably fix that too.

Comment 4

[Mike McCabe]

... So how do we reproduce the empty message problem?

Comment 5

[vidur (gone)]

Inside the firewall, go to http://maps.mcom.com, type something into the locator 
input field and submit.

Comment 6

[Phil Schwartau]

Mass-reassigning mccabe's non-JS, non-Rhino bugs to jband (34 total). 

Would like to cc mccabe; but the mass-reassign page does not allow this. 
I'll leave it up to mccabe to decide if he wants to be cc'ed on these - 

Comment 7

[John Bandhauer]

mass reassign of xpconnect bugs to [e-mail]

Comment 8

[David Bradley]

I'm unable to reproduce this problem.
I figure
1. The map site has changed
2. The problem is covered up, patch was applied and the noise is gone, but the 
problem still exists.
3. The error no longer exists.
I'll keep it, and look into when I have time.

Comment 9

[David Bradley]

This appears to not be a problem any more. The code in question has changed 
considerably so I expect it's not an issue in the current code. I'm marking this 
WORKSFORME.