I've taken some time to think about this, and I'm still not sure of what the best solution could be (for my use case, at least), but I like the configurability of 5 too.
On Mon, Oct 14, 2013 at 11:21 AM, Benny Kjær Nielsen <mailingl...@freron.com > wrote: > On 14 Oct 2013, at 9:33, Alberto Caporro wrote: > > So, please, consider a more general From-derivation a feature request :-) > > This is one of those things that could be simple to implement when I've > decided how to do it, but it's hard to decide. As usual, allow me to think > out loud and a decision might follow :-) > > The problem: Identify all sent messages in all mailboxes of all accounts > with no false positives/negatives. Do it as quickly as possible. > > Current solution: Simply use all messages located in “Sent Messages”. > > Ideas: > > 1. > > Identify sent messages by looking for messages with no Receivedheaders. > This triggers loading database indexes for > Received headers which I would currently like to avoid (it's slow). > Also, it could involve a large number of false positives, for example, for > me it includes all messages in “[Gmail]/Chats” and there are probably users > with similar issues. An alternative header is the Return-Path header, > but this is also not 100% reliable. > 2. > > Identify sent messages by searching for any messages from email > addresses listed in the account settings. The obvious problem is that this > does not include messages which only match an “Address Pattern” of an > account -- and the query system does not allow searches based on regular > expressions. > 3. > > Identify sent messages by searching for any messages from names listed > in the account settings. This can include too many messages for users with > common names, but the from derivation could subsequently verify addresses > by checking the email address (also with Address Pattern). > 4. > > Introduce an IMAP keyword for marking sent messages. Automatically > mark messages when sending and use it to locate sent messages for the > From-derivation. Some servers (MS Exchange) do not support this and it > would only work if only sending using MailMate. > 5. > > Make “Sent Messages” configurable such that users can choose to > configure it as a smart mailbox instead of a universal mailbox (or maybe > allow it to behave like both). This doesn't work 'out-of-the-box' and for a > user with many email addresses it could be tedious to setup conditions > (most often it would work using names only). > > I think 3 and 5 are the most interesting. I like 3 because it would work > out-of-the-box and I like 5 because it's highly configurable. It could be > both, but I guess I'm still undecided (leaning towards 5). > > Note that this is closely related to the hidden preference > MmAddressCompletionMailbox and a solution to the above should also work > as a solution to that problem. > > -- > Benny > > _______________________________________________ > mailmate mailing list > mailmate@lists.freron.com > http://lists.freron.com/listinfo/mailmate > >
_______________________________________________ mailmate mailing list mailmate@lists.freron.com http://lists.freron.com/listinfo/mailmate