Re: That newfangled IMAP thing...

[Lars Magne Ingebrigtsen <larsi@gnus.org>]

Ted Zlatanov <tzz@lifelogs.com> writes:

> The storage is easy if the backend supports it (which is why it's been
> done like that so far).  The value is when we reuse the logic for
> merging and resolving conflicts and we let users store marks outside of
> the newsrc.eld (for NNTP access from home vs. work for instance).

My vision is that Gnus in Emacs 24 should be so snappy and usable that a
normal nerdy programmer user would prefer using Gnus to, er, what,
Mozilla?  Or whatever the kids these days are using.

So a typical setup with a couple of IMAP servers and a couple of NNTP
servers should use less than 1 second on a `g', preferably much less
than that.  For NNTP that's easy enough, since it just gives you all the
data you need on a platter.  For IMAP, it's more difficult, since
there's a lot more data to transfer (i.e., all the flags per message).

I think I'm on the right track there with the streaming flags and stuff,
and a general gnus-sync thing, while sounding very useful, I can't
really see achieving that speed.  I'm cutting all the corners I can to
get Gnus snappy.  :-)

But that doesn't mean that Gnus couldn't use gnus-sync stuff in addition
to doing it in as sloppy manner as possible with "native" nnimap sync.

Of course, after getting nnimap up to speed, and ironing out the rest of
the problems that turned up after I rewrote the `g' function (I'm sure
there are more even though Katsumi has fixed a bunch), is writing an
HTML renderer in Emacs Lisp.  Then people can read their HTML email from
their IMAP servers without any problems, and Gnus will be
feature-complete.  :-)

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen