Rocco Rutte [Tue, 11 Dec 2007 13:58:31 +0000 (14:58 +0100)]
POP: Name hcache the way IMAP does.
Place the file under $header_cache/<url>/mutt.hcache. For now ignore
that this may break if hcache == bcache and a message's uid is
"mutt.hcache" (extremely unlikely) in which case the user will loose
only the hcache file.
Rocco Rutte [Tue, 11 Dec 2007 12:08:42 +0000 (13:08 +0100)]
Do not trust history file and mutt code about history class.
This prevents crashes as result of programming errors and/or wrong history file
(e.g. negative or too high classes).
Michael Elkins [Sat, 8 Dec 2007 20:54:54 +0000 (12:54 -0800)]
Skip over messages which were previously parsed after sorting the tail of the
list. Avoids a segmentation fault when rescanning the new/ subdir of a maildir
when messages were left over from a previous scan.
Rocco Rutte [Wed, 5 Dec 2007 09:03:59 +0000 (10:03 +0100)]
Improve inode sorting, enable it by default and add MH support.
Sorting is done on-demand only before the first stat() or open()
($maildir_header_cache_verify=yes/hcache miss). Furthermore, only a
partial list of messages starting from that item is sorted reducing
sorting overhead in most cases. For a fully hcache'd folder with
$maildir_header_cache_verify=no, no sorting is needed and no sorting
will be done.
MH parsing is now two-pass, too to simplify the code and remove
duplication.
Rocco Rutte [Thu, 29 Nov 2007 14:24:38 +0000 (15:24 +0100)]
Add a really ugly hack to make sure the plain text manual is us-ascii.
The hack is filtering manual.html to another HTML file but replacing
unicode chars by their ascii equivalents. The intermediate file is used
to dump HTML to us-ascii text (and only for that).
Rocco Rutte [Mon, 19 Nov 2007 14:32:47 +0000 (15:32 +0100)]
Make f=f handler multibyte-aware by not buffering flowed content until it's finished.
We now write out content as we read it (which simplifies the code quite
a bit) and properly flow multibyte paragraphs. This should improve
DelSp=yes handling where a flowed paragraph may not have spaces at all
except trailing ones for flowable paragraphs (after concatenating lines
we didn't have access to the original break points any longer). We still
split content by ASCII space (this should be safe as f=f itself uses
ASCII space to mark flowable lines (even for languages that aren't
expected to use spaces, see DelSp). Closes #862.
Rocco Rutte [Thu, 15 Nov 2007 12:17:16 +0000 (13:17 +0100)]
Fix more documentation build issues.
Define USE_SASL (for $smtp_authenticators) in makedoc-defs.h, include it
in makedoc.c to force doc rebuild if makedoc-defs.h changes.
Rocco Rutte [Tue, 13 Nov 2007 10:50:24 +0000 (11:50 +0100)]
Document that %f in $index_format is either Return-Path: or From:
depending on which parses correctly. Also note that alternates' regex
should be as precise as possible to avoid mismatches. Closes #1894.
Michael Elkins [Fri, 9 Nov 2007 06:13:45 +0000 (22:13 -0800)]
Removed the [!] prefix from the usage from the usage string for the un/mailboxes command since it has no special meaning in this context other than referring to $spoolfile.
Brendan Cully [Thu, 8 Nov 2007 19:48:21 +0000 (11:48 -0800)]
Add $time_inc to control progress update frequency.
I found 250ms to be noticeably different from the old behaviour, and
it ruins the rounding of the other _inc variables.
Rocco Rutte [Tue, 6 Nov 2007 17:23:06 +0000 (18:23 +0100)]
Add percentages to progress bar optionally supplied by caller.
This is necessary to let the mbox driver print useful percentages. If there's no
percentage given and we have a size, calculate on it on our own. Closes #2929.
Rocco Rutte [Mon, 5 Nov 2007 16:47:26 +0000 (17:47 +0100)]
Save value of errno across mutt_endwin() calls.
Mutt_endwin() is sometimes called after noticing an error but before
printing the message via perror(). Closes #2965.
Ralf Wildenhues [Mon, 5 Nov 2007 11:11:21 +0000 (12:11 +0100)]
Optionally define $datarootdir on our own for autoconf <2.60 compatibility.
While I'm at it, fix two warnings in other files about
ignoring datarootdir with autoconf >2.60. Closes #2905.
cypher [Sun, 4 Nov 2007 17:27:24 +0000 (18:27 +0100)]
Use strtok_r() to parse mailto: links, not strtok().
In case a headers needs to call mutt_parse_references() which uses
strtok(), too, later headers will be silently discarded. Closes #2968.
Rocco Rutte [Sun, 4 Nov 2007 17:14:25 +0000 (18:14 +0100)]
Prevent mailto parsing buffer overflow by ignoring too long header.
If they're longer than our buffer, we can't turn it into a header to
be parsed by mutt_parse_rfc822_line() anyway, so we bail out in this
case. Also make main() catchup mailto parsing errors. Closes #2980.
Rocco Rutte [Thu, 1 Nov 2007 20:21:33 +0000 (21:21 +0100)]
Unbreak exact address feature.
The problem was that when reading back tempfiles for pager display, trailing
\n were not removed from the input so the pager displayed 2 instead of 1.
Brendan Cully [Wed, 31 Oct 2007 21:04:48 +0000 (14:04 -0700)]
Fix the test for flushing queued STATUS requests in imap_buffy_check.
This should fix the bug that caused every other IMAP connection to
fail to get the new message counts. Thanks to exg for his generous
help debugging this one.