In no particular order:

* Mutt doesn't display untagged [ALERT] responses to the user. Pointed out
  by Larry P Schrof 
  19991220

* ~h searches download the entire folder, setting everything to \Seen in
  the process.
  --> Use SEARCH? or at least try to use .PEEK when doing scans. I've been
      thinking of going to always PEEK anyway, but then you'd have to store
      updates for every message you touched. Maybe a config option?

* No checks are performed on long commands to make sure that they are
  still correct after they've been made to fit in their buffers.
  Tagged message sets can exceed the fixed space we've allocated for
  them in large mailboxes. Time to move to dynamic string allocation,
  at least for these commands.

* Mutt doesn't handle timeouts or dropped connections gracefully. You
  may see occasional segfaults. Solving this cleanly requires some major
  architectural updates. Until that happens, I recommend you keep
  backup copies of messages you're composing from within your editor.
  Tommi Komulainen's keepalive patch for edit sessions may work for you,
  though. Committed to CVS late November 1999, I think.

* Sometimes hdr->env seems to be NULL, according to other users' core
  files. I haven't been able to reproduce this, and am not sure how
  to track it down. But it's very naughty indeed. It appears to be the
  result of spurious mx_close_mailbox calls handling the untagged
  BYE response.

* Mutt marks your connection for logout as soon as you say quit, even if
  you bail out with ^G later.
  --> This should be fixed by not using imap_set_logout. Instead there
      should be a function which loops through all connections and logs
      out of them one at a time. Multiple connections haven't gotten much
      testing.

* Mutt is uninterruptible during socket calls. We should handle SIGINT.

* Server copy currently doesn't take into account uncommitted changes in
  messages about to be copied. Sync first.
  --> walk tagged or individual message looking for changes, sync those?
      may require further additions to imap_make_msg_set and breaking up
      imap_sync_mailbox, otherwise shouldn't be too bad. Not currently a
      major priority, though - noone's gotten bit by this.

* Have a hard time when the home namespace isn't default "".

* The mutt_pretty routines don't work well when the delimiter isn't '/'.

Brendan Cully 
Updated 20000304
Hosted by www.Geocities.ws

1