118229 - Trunk Crash in nsAtomTable.cpp when destructing the compose window upon send [@ AtomImpl::~AtomImpl]

Status: VERIFIED WORKSFORME

(MailNews Core :: Composition, defect, P1)

Description

[lchiang]

Crash in nsAtomTable.cpp when destructing the compose window upon send

Win32 build 2002-01-03-06-trunk
IMAP, HTML compose window, signature file used
I am able to reproduce this crash with a particular msg.  I will send the msg in
.eml format to you, JF.  Once I get the "ok" from the senders of the email since
it references various conversations, I will post the .eml file to this bug report.

1.  Select this particular message
2.  Forward inline
3.  Send the message
4.  When the status shows "delivering mail", a crash occurs.

The talkback ID is:
http://climate/reports/SingleIncidentInfo.cfm?dynamicBBID=1243157

I will attach stack trace shortly.

Comment 1

[lchiang]

Attachment: stack trace

Comment 2

[lchiang]

Interesting note:
The message does get sent and when viewed, some extraneous text gets added to
the end of the message after my sig file which should end with my "4420" phone
number.

-- 
Lisa Chiang
Netscape, Client QA Director
[e-mail] | [e-mail]
IM: LisaChiang | 650.937.4420< @netscape.com="">< @netscape.com=""><
@netscape.com="">< @netscape.com=
"">< @netscape.com="">< @netscape.com="">< @netscape.com="">< @netscape.com=""><
@netscape.com="">

Comment 3

[lchiang]

Attachment: email msg to reproduce the problem.

Comment 4

[Sheela Ravindran]

I did crash trying to forward this message inline using build 2002-01-07-06 on 
win98.  However, this message was sent and received.  After looking at the 
message I see the message is quoted with '>' like when replying to a plain text 
message.
The crash occurs even when you reply to this email as well.

Comment 5

[lchiang]

Right, the message gets sent.  The crash is on the destruction of the compose
window, I think.  

I think it may be the different levels of > that could be causing the problem?  

Comment 6

[lchiang]

This is happening more frequently to me.  I have another msg which this bug 
occurs on.  Note to myself: Message from administrator stating that my message 
to  selmer failed.  I will flag this msg:  dated today at 8:38pm.

I'm cc'ing Tom Greer. Tom, is this a topcrash bug?

Comment 7

[greer]

Lisa,
This one is a little difficult to nail down. All of these crashes come under 
varying "0x0..." signatures with a "real" signature of AtomImpl::~AtomImpl in 
the second frame. 

Since this crash does not fall under a specific signature (or range of 
signatures) our mechanism isn't spotting it. However, with the 13 crashes that 
I have found it would make the Trunk topcrash list for windows.

Marking topcrash.

Comment 8

[greer]

Here are four incidents with comments that may help to affirm previous theories.

Build: 2002011006
(1493935) Comments: sending an email... there was a report of a problem with 
character sets  something like 'your mail contains characters not int the 
character set so it will be converted' then boom. I think the mail got sent.
 
Build: 2002011406
(1688463) Comments: Sending a mail message.  With a curious font insert from 
Garber travel that looked bold but according to font menu items was just normal 
text.   Hmmmm.....
 
Build: 2002010706
(1361556) Comments: Replied to incoming mail. Observed there was no scrollbar  
although the text for the original mail was there (could move down with arrow 
keys). Hit "send". App crashed.
 
Build: 2002010706
(1361688) Comments: Second occurrenceReplied to incoming mail. Observed there 
was no scrollbar  although the text for the original mail was there (could move 
down with arrow keys). Hit "send". App crashed.additional note: Selected default 
radio button for 'text and html' This is repeatable.

Comment 9

[Jean-Francois Ducarroz]

I tried the test case with a release build from 12/17 and was able to reproduce
the crash. However, using a recent debug build, I was not able to reproduce it.
My guess is that I can close it WFM?

Comment 10

[varada]

Using Win2K 2002021203 and daily debug builds I was unable to reproduce this crash.
Suggest closing as WFM.

Comment 11

[Jean-Francois Ducarroz]

WFM.

Comment 12

[lchiang]

I no longer get the crash on 2002-02-21-05 commercial build with the same
message I had in my Inbox.

I'll mark verified.

Should confirm that this is no longer a topcrash on the trunk builds though. 
I'm trying to search through Talkback now for AtomImpl::~AtomImpl stack
signature to see when the last submittals to Talkback were.  It's taking some
time....  Will post when I get more results.

Comment 13

[lchiang]

Verified on the recent trunk reports.  AtomImpl::~AtomImpl is not a topcrash any
logner.