499064 - need way for pending listener to know when a db has been opened

Status: RESOLVED FIXED

(MailNews Core :: Database, defect)

Description

[David :Bienvenu]

Attachment: wip

Pending listeners are a handy way for listeners to deal with db's getting closed and re-opened, except that there's no way to know when the db has been opened. So I'm going to add a db change listener notification for when a pending listener becomes an actual listener.

I've also made nsMsgDBView use pending listeners, as a test case, and I believe asuth needs it anyway.

As a test, I faked out an imap uid validity roll and verified that the view gets told about the new database.

The attached patch works, but I need to clean it up a little - get rid of duplicate code, basically.

Comment 1

[David :Bienvenu]

Attachment: patch for review

Asking asuth for review since he knows the relevant code really well, and he needs the patch...

Comment 2

[Andrew Sutherland [:asuth] (he/him)]

You never UnregisterPendingListener, yet the view somehow doesn't leak in the DBViewWrapper unit tests.  Is it getting cycle collected?

Comment 3

[Andrew Sutherland [:asuth] (he/him)]

Comment on attachment 383942 [details] [diff] [review]
patch for review

r=asuth with UnregisterPendingListener

Comment 4

[David :Bienvenu]

Attachment: unregister pending listener

good catch on the unregister pending listener...

I'm not sure why the unit test leak logs wouldn't have picked that up. We don't cycle collect.

Comment 5

[Mark Banner (:standard8)]

(In reply to comment #4)
> I'm not sure why the unit test leak logs wouldn't have picked that up. We don't
> cycle collect.

Yes we do: http://mxr.mozilla.org/comm-central/source/mailnews/test/resources/mailShutdown.js#34

Comment 6

[Mark Banner (:standard8)]

Comment on attachment 384009 [details] [diff] [review]
unregister pending listener

>+    nsCOMPtr<nsIMsgDBService> msgDBService = do_GetService(NS_MSGDB_SERVICE_CONTRACTID, &rv);
>+    NS_ENSURE_SUCCESS(rv,rv);

nit: add a space after the comma please (2 places in this patch).

sr=Standard8 with that fixed.

Comment 7

[David :Bienvenu]

fix checked in, with nits addressed.

Comment 8

[David :Bienvenu]

I meant to say that the db doesn't use the cycle collector, afaik.