On Mon, Jun 27, 2011 at 06:13:25PM +0200, Sébastien Michel wrote: > Really! I will read deeper the code of the master branch to learn more on > that :) > Do you add a new per-user database for ANNOTATE extension or reuse the > database annotation_db ?
We will be adding a per-user database for ANNOTATE. Actually, per mailbox I think. Greg is working on that right now (well, he's asleep RIGHT NOW, but he's working on it this week) > > Cross-folder conversations is very popular. I would be delighted to see > > master > > > / master replication in Cyrus. Is there any way to contribute to these > > > efforts? > > > > Sure! If you have someone who wants to learn about how the replication > > system works now and what's necessary to get master/master working I > > would love to chat to them. > > There is a nice documentation provided by David Carter to start learning > about Cyrus replication. Secondly deep into the code should be enough. It's changed a bit since David wrote it - I integrated it a lot more closely with the mailbox layer, so that the consistency logic is only in one place now. But yes, the code is king :) > > The main thing isn't mailbox replication, it's side things like sieve > > script replication, ACLs, etc. Stuff that doesn't have a history or > > change number on it at the moment, so you can't reconcile which end is > > correct if they don't match. > > Yes. And another issue is race condition on multi-master Races are a concern, but hopefully a single user's connections are only at one end at a time. At least that's how we're planning to use it, have users 'mastered' at one end, but just switch the DNS and away you go - instead of the current need to gate the switchover. > > Excellent. I'm sorry I haven't got back to you on your features. I know > > you have a lot of interesting things to add, and I'd love to chat to you > > more about them. It's good to see other groups than just CMU and FastMail > > working on Cyrus! > > It's OK. Some of us are able to chat on these features. Excellent :)
