aboutsummaryrefslogtreecommitdiffstats
path: root/TODO
diff options
context:
space:
mode:
Diffstat (limited to 'TODO')
-rw-r--r--TODO29
1 files changed, 29 insertions, 0 deletions
diff --git a/TODO b/TODO
index 4fd4b99..7008962 100644
--- a/TODO
+++ b/TODO
@@ -1,3 +1,32 @@
+LOCKING & Co:
+* switch to or add fcntl() locking; when combining with flock, be careful not to
+ break on Solaris/FreeBSD, where flock() is just fcntl() so using both would
+ block.
+* Ensure that we don't accidently lose fcntl locks when closing some file
+ descriptor; this applies even for flock, since again, it might be emulated
+ with fcntl
+* Block signals while writing changed mailbox back. Also, we probably shouldn't
+ use shutil.copy2() for this; at least we have to handle symlinked targets in a
+ sane way, see Debian bug #349068. mbox_sync_mailbox() in the mutt code might
+ be an example how to write back a changed mailbox.
+* Double-check the entire locking code.
+
+IMAP SEARCH BEFORE disregards time and timezone information. This should at
+least be documented. E.g. I've found that '-d 0' didn't match all messages in
+an IMAP mailbox. This is because the SEARCH key is (BEFORE 14-Nov-2007) on 14
+November, not matching messages that arrived today. Do we need an '--all'
+option to fix this?
+
+Document mbox format issues: link to
+http://homepages.tesco.net/~J.deBoynePollard/FGA/mail-mbox-formats.html,
+qmail mbox manpage, Debian manpage, RFC 4155. Document what mbox format we can
+read, and what we write.
+FIXME: we cannot yet parse rfc 2822 addr-spec stuff like quoted local-parts in
+return-path addresses.
+
+Minor annoyance: when a From_ line is generated, guess_delivery_time() reports
+the used date header a second time.
+
Check sf.net and Debian BTS for new bugs. Again.
IMAP: ensure mailbox archives are properly named. Currently imap folder names