438927 - ###!!! ASSERTION: XPConnect is being called on a scope without a 'Components' property! (Thunderbird mail windows)

Status: RESOLVED FIXED

(Thunderbird :: Mail Window Front End, defect, P2)

Description

[Mark Banner (:standard8)]

When we close one of:

main mail window, message compose window, standalone message window

we get two copies of the assertion:

###!!! ASSERTION: XPConnect is being called on a scope without a 'Components' property!: 'Error', file /Users/moztest/mozilla/hg/mozilla/js/src/xpconnect/src/xpcwrappednativescope.cpp, line 765

This never sounds good and we need to fix it, although I'm never quite sure what the consequences actually are.

Comment 1

[Mark Banner (:standard8)]

Attachment: Part 1 - balance insertControllerAt with removeController [Checkin: Comment 5]

When looking at how to fix this, I stumbled across bug 415495, as David Baron said on that bug:

> This assertion happens because we execute JS code in the controller
> implementation late during nsGlobalWindow's cleanup (in ClearControllers). 
> This is, it appears, after we've cleaned up all the properties in the window --
> after JS should be executing in that window.

This seems to relate to the cases where we use insertControllerAt(0, ...), therefore with this patch I'm only fixing those cases.

The patch reduces the number of assertions from two to one for all of the windows above. I think the second needs more debugging.

Comment 2

[Mark Banner (:standard8)]

The other assertion is caused by msgReadSMIMEOverlay.js, I'll look at fixing that soon. Interestingly it uses appendController, so I guess this assertion really depends on what is being called in the observer/controller.

Comment 3

[Mark Banner (:standard8)]

(In reply to comment #2)
> The other assertion is caused by msgReadSMIMEOverlay.js, I'll look at fixing
> that soon. Interestingly it uses appendController, so I guess this assertion
> really depends on what is being called in the observer/controller.
> 
Oh and for the reference, it was the information/stack on bug 413296 that helped me find out it was msgReadSMIMEOverlay.js

Comment 4

[Serge Gautherie (:sgautherie)]

(In reply to comment #2)
> The other assertion is caused by msgReadSMIMEOverlay.js, I'll look at fixing that soon.

You'll want to port bug 456860 msgCompSMIMEOverlay.js part to TB.

Comment 5

[Serge Gautherie (:sgautherie)]

Comment on attachment 324855 [details] [diff] [review]
Part 1 - balance insertControllerAt with removeController
[Checkin: Comment 5]

http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=ThunderbirdTinderbox&branch=HEAD&branchtype=match&sortby=Date&hours=2&date=explicit&mindate=2008-06-13+01%3A53&maxdate=2008-06-13+01%3A55&cvsroot=%2Fcvsroot

Comment 6

[Mark Banner (:standard8)]

(In reply to comment #4)
> (In reply to comment #2)
> > The other assertion is caused by msgReadSMIMEOverlay.js, I'll look at fixing that soon.
> 
> You'll want to port bug 456860 msgCompSMIMEOverlay.js part to TB.

Except as far as I can tell that is just tidy up and won't aid resolving this
bug, so therefore I'm not going to do it as part of this bug. Please feel free
to raise a separate bug though.

Comment 7

[Martin Mokrejs]

Still happens with

Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1b2pre) Gecko/20081023 SeaMonkey/2.0a2pre



###!!! ASSERTION: XPConnect is being called on a scope without a 'Components' property!: 'Error', file /home/mmokrejs/proj/comm-central/mozilla/js/src/xpconnect/src/xpcwrappednativescope.cpp, line 761

Comment 8

[Mark Banner (:standard8)]

(In reply to comment #7)
> Still happens with
> 
> Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1b2pre) Gecko/20081023
> SeaMonkey/2.0a2pre

This isn't a surprise considering this bug isn't closed. If you're referring to the SeaMonkey part, please try reading the dependent bug 456860.

Comment 9

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Mark, could you please provide a stack. There are a couple of other bugs around with this assertion and I want to trim the list down. Thanks.

Comment 10

[Mark Banner (:standard8)]

(In reply to comment #9)
> Mark, could you please provide a stack. There are a couple of other bugs around
> with this assertion and I want to trim the list down. Thanks.

Henrik, this happens whenever you close one of the main Thunderbird windows, e.g. Standalone Window, 3-pane window, I can't remember offhand about compose & addressbook. Its an issue in the js code and it is fixed in the patch attached to bug 456860 that is waiting review.

Comment 11

[Henrik Skupin [:whimboo][⌚️UTC+1]]

I don't have any debug build of Thunderbird. That's why I've asked. The assertion message only isn't helpful at any time. There are different code paths which could raise it. If you would have time to get the stack and post the first 5 frames here it would be great. But at least there will be no backend changes checked-in so other open bugs with this assertion shouldn't change their status. Thanks Mark.

Comment 12

[Serge Gautherie (:sgautherie)]

Attachment: (Bv1) Copy bug 456860 assertion fix [Checkin: Comment 14]

(In reply to comment #6)
> (In reply to comment #4)
> > You'll want to port bug 456860 msgCompSMIMEOverlay.js part to TB.
> 
> Except as far as I can tell that is just tidy up and won't aid resolving this
> bug

No, more than only tidy up.

Comment 13

[Mark Banner (:standard8)]

Comment on attachment 398866 [details] [diff] [review]
(Bv1) Copy bug 456860 assertion fix
[Checkin: Comment 14]

Looks good. r+a=Standard8

Comment 14

[Serge Gautherie (:sgautherie)]

Comment on attachment 398866 [details] [diff] [review]
(Bv1) Copy bug 456860 assertion fix
[Checkin: Comment 14]


http://hg.mozilla.org/comm-central/rev/82bceb6fbeab

Comment 15

[Mark Banner (:standard8)]

With Serge's patch being completed and checked in, this is now fixed.

http://tinderbox.mozilla.org/showlog.cgi?log=Thunderbird3.0/1253198230.1253200046.27718.gz

Comment 16

[Serge Gautherie (:sgautherie)]

(In reply to comment #6)
> Except as far as I can tell that is just tidy up [...]
> Please feel free to raise a separate bug though.

I'll file a new bug to port further cleanup I'll do for SeaMonkey.