403837 - Using window close button does not invoke the shutdown service [but not Mac], so roaming service never sees the "quit-application-requested" notice.

Status: RESOLVED DUPLICATE of bug 459376

(MailNews Core :: Backend, defect)

Description

[Nick Kreeger]

The shutdown service listens to "quit-application-requested" to determine when to run the shutdown tasks. This event works perfectly on the Mac platform. However, in Windows and Linux - it only works if a file->exit or cmd-q event is submitted to the application. Simply closing the last open window just shuts the app down and the roaming service never sees the "quit-application-requested" notice.

Comment 1

[Nick Kreeger]

NOTE: Simple solution would be to add a window listener to the windows that are open...

Comment 2

[Wayne Mery (:wsmwk)]

could this potentially lead to dataloss?

Comment 3

[Wayne Mery (:wsmwk)]

I typically close via quit or just power off the PC.  But most business users I know, and casual users as well, are fanatics about using their big red X and nothing else.  So there surely bugs reported for things which should happen at shutdown but don't. Like expunge. It will be interesting to see what happens when this gets fixed.

Bug 345067 may be of interest - Issues with prompt service's confirmEx - confirmEx always returns 1 when user closes dialog window using the X button in titlebar

Comment 4

[Dan Mosedale (:dmosedale, :dmose)]

I wonder the same thing as Wayne; marking as blocking-thunderbird3?

Comment 5

[Magnus Melin [:mkmelin]]

blocking‑thunderbird3-, but wanted P1, as this blocks some neat extension work.

Comment 6

[Wayne Mery (:wsmwk)]

dmose, kreeger, this sorta got lost in the runup to 3.0.  Assuming this is still be true gecko 1.9.2 and higher, this probably deserves serious thought going into 3.next.

Magnus, what types of extension work this might block?


(In reply to comment #0)
> The shutdown service listens to "quit-application-requested" to determine when
> to run the shutdown tasks. This event works perfectly on the Mac platform.
> However, in Windows and Linux - it only works if a file->exit or cmd-q event is
> submitted to the application. Simply closing the last open window just shuts
> the app down and the roaming service never sees the
> "quit-application-requested" notice.

Comment 7

[Wayne Mery (:wsmwk)]

(In reply to comment #6)
> dmose, kreeger, this sorta got lost in the runup to 3.0.  Assuming this is
> still be true gecko 1.9.2 and higher, this probably deserves serious thought
> going into 3.next.

we probably shouldn't let this slip further, no??


> Magnus, what types of extension work this might block?

can someone elaborate on what "neat extension work" Magnus is hinting at?

Comment 8

[Dan Mosedale (:dmosedale, :dmose)]

It's a little hard to guess how severe this is in reality, but getting a fix would be a fine thing.  Adding Standard8 for his thoughts...

Comment 9

[Magnus Melin [:mkmelin]]

I believe i was referring to some sync extension Nick was doing at the time (which needed to sync stuff on shutdown).

Comment 10

[Ludovic Hirlimann [:Usul]]

(In reply to comment #8)
> It's a little hard to guess how severe this is in reality, but getting a fix
> would be a fine thing.  Adding Standard8 for his thoughts...

Comment 11

[Mark Banner (:standard8)]

I think what Nick was working on, either didn't get mainstream or stayed as an extension.

The only thing we'd really want it for currently is send in background - which also the only thing that is using the shutdown service currently.

I'm not sure if we should dupe this to bug 459376 or leave it open, but in any case, bug 459376 would be the fix that's actually required.

Comment 12

[antistress]

Maybe that https://bugzilla.mozilla.org/show_bug.cgi?id=455600#c3 is related ?

Comment 13

[Wayne Mery (:wsmwk)]

(In reply to Mark Banner (:standard8) from comment #11)

I think what Nick was working on, either didn't get mainstream or stayed as
an extension.

The only thing we'd really want it for currently is send in background -
which also the only thing that is using the shutdown service currently.

Nick's project was Bug 377242 - [SoC] Enable Roaming Support in Thunderbird. So there is that

I'm not sure if we should dupe this to bug 459376 or leave it open, but in
any case, bug 459376 would be the fix that's actually required.

Magnus, do you agree we can just dup this to bug 459376? (and make that block bug 377242 )

Comment 14

[Magnus Melin [:mkmelin]]

Let's dupe to bug 459376 yes.