Re: IMAP trouble - and my memory's gone, too...

[Simon Josefsson <simon@josefsson.org>]

"Robin S. Socha" <robin-dated-990090181.ee1ff5@socha.net> writes:

> > The proper solution is to make Gnus wipe out everything (flags, cache,
> > agent etc) it knows about the group and sync it from the server again,
> > but a backend can't tell Gnus to do that today.  So instead nnimap
> > asks the user a cryptic question, and whatever the user answers he
> > will suspect that he created the problem and not nnimap. :) Not good,
> > I know.
> 
> Ok. So what do I do now?

To be really sure, rm the NOV cache file and remove the uidvalidity
value (`G p').

> > > | (2) (memory/critical) Warning: past 85% of memory limit 
> > > | Killing some buffers may delay running out of memory. 
> > > | However, certainly by the time you receive the 95% warning, 
> > > | you should clean up, kill this Emacs, and start a new one. 
> > 
> > Well, how large is your emacsen?  Perhaps the gnus-range functions
> > take a lot of memory if it has to operate on a huge range, like
> > expanding the range (1 . 10000000).  Also, imap.el caches some stuff
> > from the server.
> 
> That was 65k messages. HELP! I'm stuck with mutt and vim. Do something.
> Please!

How large is the memory footprint of emacs?  When does the error
occur, when entering a summary buffer?  Assuming imap.el stores 500
byte for each mail (not unlikely), 65k messages needs 32MB only within
imap.el. And building a summary buffer, with threading and whatnot,
will probably take some memory as well.  Mutt/pine etc only need to
hold in memory what you're currently viewing, while Gnus want to hold
large parts of the group in memory, so there's no easy fix here.

I just entered a nnimap group with 8k messages, and my emacs growed by
10MB.  It took about 50 seconds.  Entering a nnimap group with 16k
messages crashed my emacs (not reproducable though), but on second try
emacs grow by about 15MB.