--On Tuesday, December 4, 2001 10:44 AM -0800 Randall Gellens <[EMAIL PROTECTED]> wrote:
> At 5:42 PM -1000 12/3/01, Robert Brewer wrote: > >> Unfortunately, users on dialup that receive large attachments or >> have particularly noisy phone lines can have quite long POP >> sessions, longer than one hour. > > That could cause problems with local mail delivery, since the spool would > be locked presumably longer than the specified timeout, causing inbound > mail to bounce. Yeah, that's why we don't think locking the spool for the whole POP session is a feasible solution. >> Even if the POP sessions are short, if they happen frequently >> enough (like every 2 minutes) the odds are that spools will get >> corrupted if there is also a long-term UW-IMAP session going. > > They wouldn't get corrupted if the spool was kept locked. However, either > the POP or the IMAP session would get an error and be unable to proceed > while the other was active. The point of locking the spool is to prevent > simultaneous access. Yes, but if you keep the spool locked then you have the local MDA timeout problem. The solution Clifton mentioned (qpopper and impad checking each other's mutex before starting a session) solves both the simultaneous access problem and the MDA timeout problem.
