Re: [Nmh-workers] Locking (specifically, sequences)

[David Levine]

Ken wrote:

> - The "complex" case, where the sequence file could
>   potentially be changed from the time that it's first
>   read until the sequence file needs to be manipulated.
>   "inc" is an example; "pick" is another.

Is sortm in the same category as folder -pack?

> The basic issue there is if you add a new message to a
> folder, the long-running command wouldn't necessarily know
> the new message existed and thus when it wrote the new
> sequence file out it wouldn't write out any sequences that
> these new messages are in.

I'm worried about something else:  concurrent read/write of
the sequences file.  If two or more processes read it and
write it concurrently, the last one to write will destroy
everything that the other processes wrote.  Locked-read-
write will detect that, but I'm not sure we can always
correct it.  I don't have a concrete example of where we
can't, but it feels analogous to a merge conflict.

> adding new messages to a folder that you're currently
> inc'ing into is already not safe (messages could get
> overwritten).

How so?  It depends what "adding" means, but I would limit
that to just nmh commands that can add a new message.  I
think that's inc, refile, rcvstore, burst, and thing that
fcc's (comp, dist, forw, repl, mhmail).  And I don't think
they'll overwrite a message that any other, even non-nmh,
program had created.

David