* [9fans] nupas imap4d
@ 2009-05-19 13:52 erik quanstrom
0 siblings, 0 replies; 2+ messages in thread
From: erik quanstrom @ 2009-05-19 13:52 UTC (permalink / raw)
To: 9fans
On Tue May 19 08:25:47 EDT 2009, kokamoto@hera.eonet.ne.jp wrote:
> What is the most important difference between the two?
> Would you please post it to 9fans?
i sent a more complete answer off list (which i've lost).
the main difference is that with nupas you don't have to
load the entire mailbox into ram if you use mdirs. at
coraid we have >3gb of mboxes opened multiple
times on a 2gb machine. the largest active mbox is
598mb. also using mdirs has reduced the size of the
dump due to mailboxes by 95%. i'm pretty sure there
would be very little block sharing in venti, due to
how messages move around wrt venti blocks, so the venti
savings would be similar.
the paper is http://www.quanstro.net/plan9/nupas.pdf.
- erik
^ permalink raw reply [flat|nested] 2+ messages in thread
* [9fans] nupas imap4d
@ 2009-05-19 2:04 erik quanstrom
0 siblings, 0 replies; 2+ messages in thread
From: erik quanstrom @ 2009-05-19 2:04 UTC (permalink / raw)
To: 9fans
i put a new version of imap4d up on sources that should
eliminate uid sequence error messages with outlook.
(and the oh-so-helpful dialog boxes that go with them.)
thunderbird and apple mail don't seem to care about this
problem.
for those who care, the problem is ...
each imap message is identified by an ephemerial sequence
number and a perminant uid. imap4 takes the nonsensical
position that while result sets may be returned in any order,
uid_a < uid_b <=> seq_a < seq_b; being able to return
things in any order is no help at all if you have to sort the
messages to sequence them anyway.
so that's what i did. but at least i didn't like it.
- erik
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2009-05-19 13:52 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-05-19 13:52 [9fans] nupas imap4d erik quanstrom
-- strict thread matches above, loose matches on Subject: below --
2009-05-19 2:04 erik quanstrom
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).