Re: [Nmh-workers] Front-end census

[Ken Hornstein]

>I'm not sure how this would be an issue, unless you had multiple
>processes accessing the file at once, which is never the case for
>me. (I don't have any cron jobs fetching email. I simply fetch it when
>I want to read it.)

A number of people have fetchmail retrieving email, or procmail feeding
messages into rcvstore ... it comes up.  If you're using mutt to retrieve
your email then it's not an issue.  Again, my only point was that depending
on the front-end program you cannot assume that every nmh command will
interoperate with every front-end program.  It depends on individual usage.

>> Also ... a glance at the
>> code suggests to me that it will wipe out any sequences that are not
>> "unseen", "flagged", or "replied".  Although it seems like what mutt
>> does is write the sequences to a temporary file and then rename()s it to
>> the sequence file, so I suspect there won't be any sequence corruption
>> (but it's vulnerable to losing sequence information).
>
>I have not seen any issues with this, but again, this may be dependent
>on individual usage.

Do you not use custom sequences at all?  If not, then it wouldn't be
an issue.  It also occurs to me (from looking at the code) that mutt
does not interoperate with the standard nmh way of marking a replied-to
message; nmh puts that as an annotation (and that's what scan looks for)
but mutt stores that in a sequence.

--Ken