Re: That newfangled IMAP thing...

[Daniel Pittman <daniel@rimspace.net>]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
>
>> Next I'll do group entry and article selection, and then it's time to
>> implement the new group info function that will return real ranges of
>> unread/read/unseen articles directly to Gnus without going via the *nntpd*
>> buffer.  It might be called...  er...  `nnimap-request-group-data'?
>
> Eurhm.  I'm reading the RFC, and I'm wondering whether I'm seriously
> missing something.
>
> Is the client supposed to do a
>
> EXAMINE foo
> FETCH 1:* (FLAGS)
>
> on each single mailbox to get all the article data?

Kind-of.  Plain-jane IMAP is really bare about these things; what you want to
do is query capabilities from the server at startup, then select the
appropriate mechanisms for action from that.

For example, the EXPUNGE after sync discussed elsewhere probably means you
want to demand (unconditionally, because nothing doesn't implement) UIDPLUS
from the server.

That adds a bunch of UID versions of commands that otherwise need to reflect
the id in the current fetch, making life *much* faster for you.


Anyway, to sync flags you want this:

    https://tools.ietf.org/html/rfc5162#section-5

That would be the official upstream documentation about the QRESYNC variant of
the "flag sync" process, and includes how to handle less capable servers.
Given that "less capable" means "most" — since only Zimbra has QRESYNC that
I deal with regularly...


Anyway, CONDSTORE capability is pretty routine, and most things will have
that; I would suggest you stash away a pointer to the right routine somewhere
in the server thing every login, after doing a capabilities query.

> If you have a million emails on the server (and who hasn't?), that'll like
> take forever.  Fer sure.

Yah.  CONDSTORE makes it much faster.
        Daniel

-- 
✣ Daniel Pittman            ✉ daniel@rimspace.net            ☎ +61 401 155 707
               ♽ made with 100 percent post-consumer electrons