Closed GoogleCodeExporter closed 9 years ago
That's not the best way to thread the messages. The language of the prefix is
not very important (we can count the prefixes and see if they are the same
across messages), if we have for instance:
Subject A date 1
RR: Subject A date 4
RR: Subject A date 2
RR: RR: Subject A date 3
RR: RR: Subject A date 5
We get:
Subject A date 1
RR: Subject A date 2
RR: RR: Subject A date 3
RR: RR: Subject A date 5
RR: Subject A date 4
I'm sure you can find the flaw, the reply made on date 5 can be an answer to
the message from date 2 or 4.
It's best to use the following message headers to calculate the threads
structure:
References: <message_id_1>
In-Reply-To: <message_id_1>
Message-Id: <message_id_2>
You can check RFC5256 (IMAP - SORT and THREAD Extensions) where you can find a
description of the common used threading algorithms.
We have the hlimap lib ready to do the client side threading, but it's not
implemented yet. To sort the messages we have similar infra-structure, this is
already done.
Summary: The prefix language have no influence in the threading algorithm.
Threading by subject is bad. Client side threading will be supported on hlimap.
I prefer to have it done server side.
Original comment by hguerreiro@gmail.com
on 27 Sep 2010 at 7:51
Original issue reported on code.google.com by
akimov.alex
on 27 Sep 2010 at 2:51