Sieve

[ganjo]

Hi

Mail filters are slow, and become quite unusable when I got more than 20 rules.
I found that filter are sieve scripts stored on ldap.
Instead, should it not be better to make db's request like dbmail do (and I think, like tag and searches) ? It could be really faster to load and to execute, and should offer some possibilities like moving already received messages, or make filter using, for example, sender's tags.

Just an idea

[KevinH]

When is it slow? To edit the rules in the options UI? Or are you timing the speed at which the filters run somehow? Filters in general run as the mail arrives to the user will never see or notice if there is 1 filter or 100 to run.

[ganjo]

To edit the rule in the UI

I agree that it's not really a big prob as I make my filter rules only one time, and then use tags or searches. But I was only thinking that its strange to store sieve rules into the ldap. IMHO it can be interesting to store a filter as a request to the database, it should be easier easier to add rules as "if the sender is tagged" "if the sender is in this distribution list"... Things that I can do with a search, but searches don't allow to move mails in a specific directory

My 2 cents

[KevinH]

Where the filters are stored doesn't really matter. LDAP is perfect for data that is updated once/rarely and read often. We cache the lookup so it's not even looking in LDAP much.

Some good ideas for new filter options. Bugzilla is the place for those. We also have some open issues to let you manually run your filters across a folder so that should help with updating a filter and then wanting to re-filter old mail.