551917 - Leak regression following bug 496019 landing in m-c (mozilla::storage::Connection::Close can spin a nested event loop)

Status: RESOLVED FIXED

(MailNews Core :: Database, defect)

Description

[Mark Banner (:standard8)]

I'm not quite sure what's happening here... when we landed the storage-backport on comm-central, it had a Mail Lk regression which Andrew fixed.

Now that bug 496019 landed on m-c, we've got a similar regression.

What I have also realised is that the storage-backport isn't being built with trunk builds (mail/build.mk needs adjustment as m-c changed the tier names).

Passing to asuth to let him work out what's going on here.

Comment 1

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

(already stated on IRC, but for posterity...)
This will be fixed by the landing of bug 507414 on mozilla-central today; the gloda detection logic checks for the interface provided by 507414 but the change in behavior is to use the changes of both bug 496019 and bug 507414.

We definitely want to use trunk storage on trunk, so no need to change mail/build.mk.  I might have forgotten to add a comment to the file to explain what's going on, so I'll be sure to add that factor into the comment too.

Comment 2

[Mark Banner (:standard8)]

This was fixed by bug 507414 landing in mozilla-central.