Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!samsung!zaphod.mps.ohio-state.edu!tut.cis.ohio-state.edu!mailrus!ncar!tank!abbo
From: abbo@tank.uchicago.edu (Laurie Abbott)
Newsgroups: comp.mail.mh
Subject: Re: Why doesn't "dist" use aliases??
Message-ID: <6723@tank.uchicago.edu>
Date: 14 Dec 89 05:35:45 GMT
References: <8912131727.AA02600@hub.ucsb.edu> <1989Dec14.043413.13647@haven.umd.edu>
Organization: The University of Chicago
Lines: 23
In-reply-to: louie@haven.umd.edu's message of 14 Dec 89 04:34:13 GMT

In article <1989Dec14.043413.13647@haven.umd.edu> louie@haven.umd.edu (Louis Mamakos) writes:

>  It looks to me that two processes
>  fiddling with a folder simultaneously will hose each other, since it
>  seems that no locking is done.
>
>  This is the reason that I don't use the slocal program to
>  automagically distribute mail to folders as it arrives.  How could you
>  safely modify folders when mail could be asynchronously stuffed into
>  them?

I've been using slocal for some time and have never had any problem
with concurrent processes axing each other (though we have other
mysterious mh bugs %:)).  My automagic mail goes into certain discreet
files in certain folders, nothing elaborate, but it is serviceable.  I
don't think I am willing to give up the magic even if it's dangerous.
Two automagic files are dumped into ~/Mail, which I modify often; only
problem I've ever had was a couple of core dumps, where were (as
usual) operator instigated because of a faulty .maildelivery file.

I cannot, however, convince .maildelivery to pipe any receipts
through shell commands, though I've decided it's probably the fault of
the shell I run (a modified tcsh) rather than mh.