On Friday, 09 December 2011 at 16:52, Victor Sudakov wrote: > Brendan Cully wrote: > > > > > > > > > > > > After upgrading to mutt-devel-1.5.21_3 from ports (FreebSD > > > > > > 7.4-STABLE) > > > > > > the mark_old=yes option ceased working in IMAP folders (it still > > > > > > works > > > > > > in local mboxes). > > > > > > > > > > > > Now, when I exit an IMAP mailbox without reading some new messages, > > > > > > and then open it again, those messages again show as "N" instead of > > > > > > "O". Even if I explicitly mark some messages as "O", they become > > > > > > "N" > > > > > > on the second visit to the mailbox. > > > > > > > > > > Have I asked something weird or does nobody use IMAP? > > > > > > > > > > I clearly remember IMAP messages becoming "O" in mutt-1.4.2.3_4, and I > > > > > miss the feature very much. I wonder, where mutt could have stored the > > > > > "O" flag on the IMAP server and why it doesn't do it now. > > > > > > > > It is supposed to work (it works for me against a Cyrus IMAP > > > > server). Perhaps your server doesn't allow custom flags to be stored? > > > > We'd have to inspect the debug transcript to know for sure. I'd > > > > suggest you file a bug. > > > > > > Here is a complete IMAP session: http://zalil.ru/32233113 > > > Can you please look what could be wrong with it so that a bug can be > > > filed? > > > > I can't tell from that. It is suspicious that there's no "Old" in the > > FLAGS or PERMANENTFLAGS sections, but they seem to contain other > > custom flags. It would be useful if you could generate a trace over > > three sessions (preferably with a smaller mailbox): > > > > session 1: mark some messages as unread, sync the mailbox (you can > > skip this if you already have a mailbox with new messages to use) > > session 2: open the mailbox, don't read the messages, quit (this > > should mark the unread messages as old) > > session 3: open the mailbox (this will test whether the old flag is > > preserved). > > Here you are: http://zalil.ru/32234313 > Three IMAP sessions as you requested and the complete packet dump. > > I expect that mutt should try to save some flags in the second > session to mark the unread messages as old, but I don't see it happening.
I've realised I didn't have to put you through that, because the answer was in the first transcript: Communigate doesn't appear to support custom flags (there's no '*' in PERMANENTFLAGS), so mutt isn't trying to set the Old flag. This changed for you because mutt stopped relying on the Recent flag to distinguish between new and old messages (the flag is very unreliable in many contexts). I think we could fix this case in mutt if you have hcache enabled by storing the Old flag in the header cache.
