Mark Crispin <[EMAIL PROTECTED]> writes: > Since it seems that Eudora wants to search 60 messages at a time, with a > progress bar and stop button, it should be using message sequence numbers. > Remember my repeated harping on how even UID clients should use sequence > numbers? This is the solution for the "it tries all possible UID values > even if they don't exist" problem. > > In other words, it should do: > 00015 UID SEARCH 1:60 BODY target > 00016 UID SEARCH 61:120 BODY target > etc. until it reaches the number of messages in the mailbox: > 00020 UID SEARCH 300:343 BODY target > > Doing this will search each and every message in the mailbox, without > taking weeks to do so.
This isn't a complete example -- the client will want to have the UID of the matched messages as well (since local article caches can only usefully be based on UIDs) and will have to send commands to the server to retrieve the UID of the matches sequence numbers: perhaps it sends FETCH 4:7,9,... (UID). Depending on the number of matches, this will waste much more bandwidth than the Eudora method. So there isn't a generic recommendation possible to give here, which I believe is symptomatic of something missing in the protocol.
