Brendan Cully [Tue, 13 Dec 2005 06:52:23 +0000 (06:52 +0000)]
Restore IMAP postponed check in the wake of the buffy pipelining patch.
Fall back to NOOP if IDLE poll fails.
Allow for untagged messages when starting IDLE.
Free idata->cmdbuf in imap_free_idata.
Brendan Cully [Mon, 12 Dec 2005 17:24:11 +0000 (17:24 +0000)]
. Add $imap_idle boolean to disable the IDLE extension (may be needed with some
servers like the particular version of dovecot I tried yesterday).
. Don't attempt to poll mailboxes with connections that haven't yet
authenticated.
. Recognize NO responses again (silly typo, spotted by Charlie Allom)
Brendan Cully [Sun, 11 Dec 2005 22:51:29 +0000 (22:51 +0000)]
IMAP command batching code, used to pipeline mailbox poll requests.
Up to 10 poll commands will be sent at a time (tunable in imap_private.h).
This is a huge win on my currently awful wireless link. It takes a knife
to a lot of fundamental IMAP code (mostly for the better), so it may
have destabilised things. Time for some brave (or lazy non-Changelog-reading)
testers to report... next up, IDLE support.
Brendan Cully [Sat, 10 Dec 2005 07:06:44 +0000 (07:06 +0000)]
Set up a command pipeline. Currently everything still runs a single command from
start to finish before moving on, but this creates the infrastructure to pipeline
mailbox polling calls. Bitter experience at the end of a flaky wireless link
teaches me that this can be useful.
Thomas Roessler [Thu, 24 Nov 2005 12:50:27 +0000 (12:50 +0000)]
mutt/2142 suggests that the change from pattern.c,3.14 -> 3.15
causes inconsistency and confusion. Reverting the change;
adding a helpful message to the status line.
Brendan Cully [Mon, 10 Oct 2005 18:26:31 +0000 (18:26 +0000)]
Make PGP decode failure non-fatal when displaying messages (as opposed to
decode-saving them). I think it would be nicer to include the original text
when decryption fails though...
David Champion [Tue, 4 Oct 2005 06:05:39 +0000 (06:05 +0000)]
Attachment counting for index display (patch-1.5.11.dgc.attach.6).
Modifications: attach_recurse and attach_ignore_fundamental stripped,
some debugging code removed, some bones thrown to check_sec.sh.
The following patch has an effect only when the braille_friendly
variable is set. When this is the case, the cursor is put on the first
line of the message body, when viewing it with the pager.
Ensure that Content-Length printf format specifier matches off_t size, to
avoid mail-losing Content-Length: 0 (bug introduced by large file patch a couple
of days ago). Closes: #2096.
Thomas Roessler [Thu, 22 Sep 2005 09:06:13 +0000 (09:06 +0000)]
When composing a reply, there might be degenerate cases in which a
CC header is created, but no To header. In this case, we now use
the auto-generated CC header as the To header.
Don't allow imap URLs with empty hosts. Also don't attempt to complete
on open connections when there aren't any (diagnosed by Nico Golde). Closes: #2079.
Use ctx instead of Context when checking whether or not to do IMAP search.
Resolves a possible segfault in color index ~b operations. I can't remember
whether Context vs ctx was intentional - I doubt it.