Re: Courier 20240725 development build
Bernd Wurst <[email protected]> Thu, 12 Sep 2024 17:23:04 +0200
| Newsgroups | gmane.mail.imap.courier.general |
|---|---|
| Message-ID | <[email protected]> |
Am 12.09.24 um 13:23 schrieb Sam Varshavchik: > I don't see why cron has to be involved. It's not like the same .courier > file needs to be created over, and over again. How would you do it without creating the same .courier files in every virtual home dir? > You must have some process in place for provisioning new mailboxes. > Something already has to create that maildir when the record for the new > account gets added to your user database. Well, that something simply > needs to install that .courier file, too, as part of the deal. The account data is set via web. The web process shall not have access to the user account that handles the mailboxes. As our delivery goes to a program, this queries the database and simply creates the maildir as needed. > You are combining bits and pieces of two different things (Courier and > Dovecot) that work in different ways. Yes. But we're unix fellows, for every part of the work I use the tool that does its work best. Courier is great as a mail server but it's webmail is ridiculous nowadays. We have had squirrelmail - not a huge difference - for years but dropped it because customers did not use it. SIEVE filters are supported in thunderbird and roundcube webmail, so this is a huge benefit. Courier IMAP was in place for a long time but as mailboxes had grown a bit, it's a major improvement to get folders with 20k messages opened in no time with dovecot instead of waiting a half a minute for courier-imap. Don't get me wrong, I like the core concepts of courier, some of our customers happily keep their own .courier files. Courier's modular structure make all of this possible, we could not have this setup with any other mail server I'm aware of. These pieces do not really work in different ways, maildrop in delivery mode is just the same as dovecot-mda from the perspective of courierlocal. They're both modular and give an excellent stack of software. > Having said all that, it wouldn't be too complicated to make the > alternative + separator character optional, that just adds a little bit > more complexity. My first intuition is that it's not a good idea to hard code a replacement of a (valid) character in an email address before looking it up in any kind of account database. The plus sign is often used for extension addresses but technically, it's a valid character that one can use in their address. The change makes it impossible to have this char in a regular address. The user now must be aware of courier's replacement and must define the address with a - instead of +. I don't like this in general. _______________________________________________ courier-users mailing list [email protected] Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users