Re: Patch: rework IMAP IDLE locks behavior

Review is for tomorrow or this weekend, if you are lucky ;-)

Thanks, though.

On Wed, 2008-10-29 at 17:20 +0100, José Dapena Paz wrote:
> 	Hi,
> 	I'm attaching a complete rework of the lock behavior of Tinymail
> (camel) imap IDLE implementation.
> 	The problem we want to fix is:
> 	* Sometimes IDLE thread causes deathlocks (connect lock, idle lock,
> g_thread_join, etc). This happens easily when there are disconnects, or
> when we switch fast among folders.
> 	The main ideas with this implementation:
> 	* Provided a count of reasons to stop idle because of connect lock.
> 	* If a caller has taken connect lock, then it should do the send done
> stuff itself.
> 	* Idle thread has the connect lock always when it's in the body of the
> loop.
> 	* In any point, if we want to  take the connect lock, then we should
> stop idle. No partial idle can be alive when we want to connect.
> 	* Camel queue enables/disables the possibility to start idle. That's
> for avoiding running IDLE / SEND DONE in the middle of queued
> operations.
> 	With these changes, the behavior is good, and I didn't get any new
> hang. I tested this both from Modest and Tinymail Demo UI, for 4 days,
> and still didn't get hangs.
> 	Anyway, as the patch is really complex and long, I would ask for
> careful review). It's one of the most annoying bugs we have in Tinymail
> and it may be a solution for this :).
> Changelog... the description of the patch in this mail would be a good
> explanation ;). I could also add a deep explanation of the way this is
> working in docs directory, for developers reference.
> _______________________________________________
> tinymail-devel-list mailing list
> tinymail-devel-list gnome org
Philip Van Hoof, freelance software developer
home: me at pvanhoof dot be 
gnome: pvanhoof at gnome dot org

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]