On Tue August 21 2007 12:35:52 pm Daniel Jacobowitz wrote: > On Tue, Aug 21, 2007 at 12:33:37PM -0500, John Goerzen wrote: > > > > I would think that Courier would *have* to write to disk in that > > > > situation. > > > > > > Maybe it has to, but it doesn't. I can provoke an IMAP protocol > > > violation by trying this. I would still be happy if offlineimap > > > > I assume you used strace or something on it? > > I looked at the source, found the conditions for updating the cache, > and saw a very suspicious if (!ro). I found your cloned bug > immediately after that, so I've added the details there.
Heh, when I wrote "down this path lies madness", I did not expect that Courier has actually already gone down that path! The relevant RFC3501 section is 2.3.1.1. The changing of a message's UID without changing UIDVALIDITY violates a MUST as you rightly noted. The changing of the UIDVALIDITY after EXAIME violates a "SHOULD" and "STRONGLY ENCOURAGES" (heh, I hadn't seen that one before) as well. The mutating UID without a change in UIDVALIDITY is a potential data loss or corruption issue. I think that #438993 may warrant a severity higher than Important. Thank you very much for all your help tracking this down. -- John -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
