95124 - Mail news interface for Simple MAPI

Status: VERIFIED FIXED

(MailNews Core :: Simple MAPI, defect, P1)

Description

[Tiantian Kong]

This is to track the implementation for the mail news interface for Simple MAPI.

Rajiv has mailed the request to Jean-Francois. Here's a copy of the mail.

"Hi David and JF,

Thank you very much once again for ur feedback. Based on your answers 
two enhancements need to be done for supporting the MAPI Logon, Logoff, 
SendMail and SendDocs APIs. 

1. define an API in the nsIMsgComposeService for registering a 
SendListener which will be called/notified on completion of the send.

2. figure out a way, either add a param to existing API or define 
another one, to pass the password that is recieved during MAPI Logon at 
the time of send message. This password would then be used to verify 
with the SMTP server, if there is a password specified for the sender. 
If the password is wrong prompt the user for it again and if it is still 
incorrect return failure.

Please let us know if there is anything else that you feel might be 
added / modified in the mailnews API.

Also I will open a separate bug for these changes to mailnews module.

Thank you,

- Rajiv."


Tiantian has discussed with Scott about the delivery schedule and our
expectation for the delivery of these interfaces before the end of this week. (8/17)

Comment 1

[Jean-Francois Ducarroz]

patch coming...

Comment 2

[Hong Kwon]

mass qa assigning all simple MAPI bugs to olgac 

Comment 3

[Jean-Francois Ducarroz]

Attachment: proposed fix, v1

Comment 4

[Jean-Francois Ducarroz]

oops, I forget to extend the compose service to let you pass the listener and
the password...

I'll create a new API that let you Open a Compose Window from a compose Param.

Comment 5

[David :Bienvenu]

NS_IMETHODIMP nsMsgCompose::GetExternalSendListener(nsIMsgSendListener **_retval)
+{
+  NS_ENSURE_ARG(_retval);
+

should be NS_ENSURE_ARG_POINTER. And is there a reason this is [noscript] in the
idl?

Comment 6

[Jean-Francois Ducarroz]

rigth, should be NS_ENSURE_ARG_POINTER. it's a noscript api because it's used
only by the listener internally and all other internal use only api are noscript.

Comment 7

[Jean-Francois Ducarroz]

Attachment: proposed fix, V2

Comment 8

[Jean-Francois Ducarroz]

If V2, I added nsMsgComposeService::OpenComposeWindowWithParam and used
NS_ENSURE_ARG_POINTER instead of NS_ENSURE_ARG.

Comment 9

[David :Bienvenu]

r=bienvenu

Comment 10

[Scott MacGregor]

sr=mscott. I'm still fuzzy as to why we need the no script tag on the the
interface  method though. Just because it currently is only used internally
doesn't mean the no script tag should be there. If there's a security reason why
we don't want it exposed via xpconnect then you should ue the no script tag.
Otherwise, I don't think we need it.

Comment 11

[Jean-Francois Ducarroz]

The only reason I use noscript for the new interface is to be consistent with
the others api. In this case there is no security issue. Also I don't see any
reason to be scriptable but that doesn't mean it should not.

I'll break the old pattern an make this API scriptable.

Comment 12

[Jean-Francois Ducarroz]

Fixed and checked in.

Comment 13

[Brendan Eich [:brendan]]

Lots of tabs in the file, and some in the + lines of the patch -- could you
please respect the indent-tabs-mode: nil of the Emacs modeline comment on line
1, in the future?  If you use vi, switch to vim and :set expandtabs.  If you use
MSVC, try Options / Tabs / Expand Spaces (from memory, should be close).  If you
use CodeWarrior, I hear the latest version has an expand-tabs-to-spaces option,
but don't know for sure.

/be

Comment 14

[David :Bienvenu]

it looks to me like it's the old, existing, context lines with tabs, not the
lines that were added.

Comment 15

[Jean-Francois Ducarroz]

Right, we still have early code which uses TABS. I tried to fix them when I have
to go back to those source files but I did not changes it this time. Maybe next
time...

Comment 16

[Jean-Francois Ducarroz]

You can now specify an external send listener as well a smtp password when you
open a compose window through the interface nsIMsgComposeParams:

1) Create a nsIMsgComposeParams, set the various field you need.
2) Open a compose window using nsIMsgComposeService::OpenComposeWindowWithParams

Also, in the case you need to do a blind send (without a compose window), you
can now pass a smtp password through the interface nsIMsgSend::CreateAndSendMessage

PS: I have not tested those new interfaces, I am just crossing my fingers.

Comment 17

[Rajiv Dayal]

The usage of nsIMsgCompose::createAndSendMessage doesnot seem to work in case of
send without a UI. Seems like an smtp connection is not made, although the smtp
servers etc is specified correctly. am in touch with mscott but looks like it
could be (not very sure though) because the notification callbacks specified by
nsMsgSend::DeliverMsg is null since no mailnews window exist. This problem needs
to be looked into.

Comment 18

[Rodney Velasco]

QA over to trix, please verify.

Comment 19

[Trix Supremo]

reopening bug, since no fix has been checked in yet.  not clear why this was
marked resolved fixed in the first place.

Comment 20

[Scott MacGregor]

trix this was marked fixed because we checked in the fix for this bug. =) You
are looking for eclient's simpla mapi bug which uses the apis we exposed in this
bug. 

You really can't verify this bug except by either code inspection or by waiting
for eclient to land simple mapi and then verifying that simple mapi works. This
bug was to add some apis for the mapi team to use and we did that. reclosing. 

Comment 21

[Trix Supremo]

marking verified with 2001-10-22-18-0.9.4

Comment 22

[Trix Supremo]

changing component from Composition to Simple MAPI