Has anyone ever experienced this theoretical corruption? I haven't. I
use procmail's locking, but then other nmh programs don't use that same
locking mechanism.
So it depends what you mean by procmail's "locking". If you're using
locking in procmail so only one rcvstore is running at once, then no,
you wouldn't have seen it.
Obviously procmail doesn't know how to lock mh sequences files; it doesn't
even know they exist. However, people DID get corrupted sequence files
before Paul Vixie's code. That went into the tree in 2003, and that was
during the long dark teatime of the soul between 1.0.4 in 2000 and 1.2
in 2005 (1.1 never really saw a real release). So anything 1.2 or newer
shouldn't produce corrupted sequence files, even without procmail
locking.
Now it's clear that there is a race condition if you don't use locking
in procmail; if two messages come in to two rcvstore processes at the
same time, then the loser of the race won't have his message on the
unseen sequence. The sequence file will still be readable by nmh, it
will just be missing a message (the same thing can happen if you run
"show" or other nmh programs when messages come in). That's what we've
been talking about fixing. But like everything else in nmh, a seemingly
simple thing is actually a whole pile of work.
--Ken
_______________________________________________
Nmh-workers mailing list
Nmh-workers(_at_)nongnu(_dot_)org
https://lists.nongnu.org/mailman/listinfo/nmh-workers