703400 - We should remove DBConnection from nsIFormHistory.idl.

Status: RESOLVED WONTFIX

(Firefox :: Sync, defect)

Description

[Rafael Ávila de Espíndola (:espindola) (not reading bugmail)]

And use a higher level function to create the statements so that they can be finalized.

Comment 1

[Justin Dolske [:Dolske]]

I think this is likely a WONTFIX... Bug 566746 is going to add async APIs that Sync should be able to use instead of the raw DB connection. I'm not planning on removing the DB connection (since it's often useful for tests), but I could be convinced otherwise if there's a reason!

Comment 2

[Marco Bonardo [:mak]]

(In reply to Justin Dolske [:Dolske] from comment #1)
> I think this is likely a WONTFIX... Bug 566746 is going to add async APIs
> that Sync should be able to use instead of the raw DB connection. I'm not
> planning on removing the DB connection (since it's often useful for tests),
> but I could be convinced otherwise if there's a reason!

The reason behind removing the connection is that, until we implement bug 702559, we don't have a "safe" connection to expose to add-ons.
Practically now we could fix all internal Storage users so that they don't cause thread contention (and Bug 566746 is going toward that direction), then a single add-on using your connection handle from the outside may reintroduce contention and thread-locking by using your internally-pure-async connection with the Storage synchronous API.
Surely, removing the connection may have a cost in terms of tests and existing add-ons using it, that I've not yet evaluated.

Comment 3

[Rafael Ávila de Espíndola (:espindola) (not reading bugmail)]

The patch in 702848 has been changed to use a "will close" callback. Closing this bug as Wontfix. We can reopen if something changes in 702848.