17445 - String-ID error on sending

Status: VERIFIED FIXED

(MailNews Core :: Backend, defect, P3)

Description

[Ben Bucksch (:BenB)]

After compsing a msg and hitting "Send", I get a warning msg-box with no title
and the content "[String-ID: 2147467259?]". After clicking OK, the composer
re-appears.

On the console, I see the following:

SendMessage from XUL
GenericSendMessage from XUL
Identity = [nsIMsgIdentity: id1]
attachments =
THE CLEANUP ROUTINE FOR nsMsgComposeAndSend() WAS CALLED
Message Delivery Failed!
[Gtk-WARNINGs]
commonDialogOnLoad
Move window by 107.5,96
screen x 0screen y 0
[warning msg appears and waits for user action]
nsMsgComposeSendListener: the message send operation failed!

This bugs exists since at least a month and was seen by [e-mail]
(IRC name, no email address), too. He reported to have seen this on 2 machines,
both using the same SMTP setup.

I'm using smail from Debian 2.1.

Comment 1

[Phil Peterson]

What build are you running? ducarroz just fixed something like this, although I
can't put my hands on the bug number right now.

Comment 2

[Jean-Francois Ducarroz]

My previous fix was to insure that we display an error (text or string ID) in every case.

We still have several case where we display a generic string ID error instead of a nice human language string. We
still have to cleanup the way we manage errors receive form low level modules like netwerk.

For that we need to have a better way to report errors that just returning a nsresult. We need to be able to carry the
low level error as well when it append during the multiple send message phases (preparing the message, sending the
email , posting to newsgroup, copy the message, etc).

Comment 3

[lchiang]

The bug Phil was thinking about is:
http://bugzilla.mozilla.org/show_bug.cgi?id=14591

Comment 4

[Ben Bucksch (:BenB)]

From the bug lchiang gave me (thanks!), I guess, this is because of bug #2734.

Comment 5

[Ben Bucksch (:BenB)]

This was caused by buggy migration code, that is (mostly) fixed now. The
error-msg is now in clear text, too. FIXED.

Comment 6

[Ben Bucksch (:BenB)]

BTW: that "mostly" is the dependency on bug #20878.

Comment 7

[esther]

Since the "mostly" part is covered in the dependency bug, I will verify this 
bug per the reporter and resolver are one in the same person and have stated it 
is fixed. Re-open if you feel this is verfied in error.  If you reopen give test 
scenario for reproducing.  I know we have had bad string-id's in the past but 
have not run into them since the specifice bugs for reproducing them have been 
fixed.