<<< Date Index >>>     <<< Thread Index >>>

Re: imap/2182: segFault on imaps folder change with mailboxes=.



The following reply was made to PR imap/2182; it has been noted by GNATS.

From: Brendan Cully <brendan@xxxxxxxxxx>
To: Paul Walker <paul@xxxxxxxxxxxxxxxxxxxxx>
Cc: bug-any@xxxxxxxxxxxxx
Subject: Re: imap/2182: segFault on imaps folder change with mailboxes=.
Date: Mon, 3 Apr 2006 17:24:43 -0700

 On Tuesday, 04 April 2006 at 01:17, Paul Walker wrote:
 > On Mon, Apr 03, 2006 at 05:09:22PM -0700, Brendan Cully wrote:
 > 
 > > = and + always expand to $folder, so I don't think this is
 > > inconsistent.
 > 
 > What I was seeing on traces through was that imap_parse_path set mx.mbox to
 > NULL. Your code does a strdup of "INBOX" if mx.mbox is NULL. So "=" does not
 > expand to $folder in this particular case with the code in CVS.
 
 = still expands to folder. The IMAP URL convention (at least in mutt)
 is that an empty path component is equivalent to INBOX when referring
 to a mailbox - you'll see similar code already in
 imap_fix_path. Probably I should have found a nicer way to consolidate
 things.
 
 > So, logically, if you test CVS mutt with a mailboxes line such as 
 > 
 > mailboxes = ! foo
 > 
 > you'll get INBOX and INBOX polled twice, and foo once.
 
 it's possible. the mailboxes parser has some rudimentary duplicate
 filtering, but I'm not sure whether imap_fix_path gets invoked
 anywhere in there. the nice thing to do would probably be to make the
 duplicate filter smarter if necessary.