[tpop3d-discuss] Tpop3d over NFS with Maildir: New mail files not immeditately seen?

Chris Lightfoot chris at ex-parrot.com
Fri, 22 Nov 2002 09:26:37 +0000


On Fri, Nov 22, 2002 at 02:10:07AM +0000, Paul Makepeace wrote:
> On Thu, Nov 21, 2002 at 06:09:07PM +0000, Chris Lightfoot wrote:
> > If Paul M. is reading this -- did you put in the
> > check-for-newer-files thing, or is it mine? I can't quite
> > think why it's there....
> 
> I was asking myself that same question! Looking back, it's something
> that happens in qmail-pop3d whence I drew some inspiration. I vaguely
> recall understanding why at the time but don't recall any more. To add
> further mystery, no-one on this thread knows either,
> 
> http://www.geocrawler.com/mail/thread.php3?subject=Vpopmail%2Bqmail+pop3+has+lost+it%27s+mind%21&list=513
> 
> So in summary: I have no idea.

Hmm. Nor do I, thinking about it any more. It can't be to
prevent partial delivery, given the semantics of
delivering to a maildir.

> It's a toss-up then I suppose between adding an entry in the FAQ saying,
> 
> Q. tpop3d allows a user to log in fine, the MTA is delivering
>    successfully, and indeed the messages are there. So why is tpop3d not
>    showing messages?
> 
> A1. Make sure the times are synchronised for the server that tpop3d is
>     running on and the server that the MTA is delivering on,
>     presumably over NFS.
> 
> A2. Because of something in the code no-one understands :-)
> 
> ...and commenting it out and seeing what happens.


Heh. I guess we could have tpop3d log a warning message...
(maildir $foo contains messages FROM THE FUTURE(TM); check
your clock synchonisation or fix your time machine).

I guess I should do a survey of some other maildir
implementations and see what they do....

> PS Incidently, the *only* comment in all of qmail/maildir.c is on this
>    exact subject. /* don't want to mix up the order */ ...

Lovely.

    /* You are not expected to understand this. */


-- 
If you would like to buy marijuana, please press the `#' key now....
(answering machine message)