Mutt now has the strange habit of asking "Append messages to ...?"
when Fcc: is set and the confirmappend option is set. Answering "n"
goes back to the main loop of ci_send_message. (So the question should
be asked differently -- otherwise you'd expect that the message
will be sent, but you won't keep a copy. Also the plural doesn't make
sense.)
If a message has been PGP encrypted, things are not the way the main
loop apparently thinks they should be. The function definition
lacks appropriate assert()s, so this shows in a segment violation when
check_attachments tries to strfcpy idx[i]->content-filename,
which then is a NULL pointer.
A related feature request: I often would prefer to skip the first
round of editing, i.e. go immediately to the menu and only then into
the editor. One reason is that I want to check the Cc: entries or
might want to add Bcc: entries before I forget it; another reason is
that I want to set the PGP settings to "encrypt" before even typing my
message.
Finally, I'd like to have the possibility to assign OpenPGP key IDs to
recipient's e-mail addresses -- so I could have something like
key-id <[EMAIL PROTECTED]> 0x5a6a9b85
in my .muttrc and wouldn't have to look through the matching entries
of the selection menu (which is typically not complete for some
reason -- the correct key shows up only after leaving the initial key
menu with "q" and then typing the key-ID --, but I understand that
GnuPG support cannot yet be considered complete, so I don't worry too
much about that).
