Re: [Nmh-workers] Locking In Scripts and nmh Locking

[Paul Vixie]

On 4/26/2012 5:42 PM, David Levine wrote:
> The more we can avoid locking, the better. ...

i think locking inside of mh is nec'y, otherwise we'll see two "pick"
processes writing into the same .mh_sequences file at the same time,
corrupting it. (did you know that if there's a corrupt .mh_sequences
file, "rcvstore" won't run? i fought mh locking bugs for years because
my .forward went to a perl script that ultimately ran rcvstore, while
cron was capable of running "pick" at any moment.)

i think that making this locking accessible at a higher level (command
level) makes perfect sense, either because somebody wants to run "find
~/Mail ... -print | xargs ..." from cron or for any other "mh is just
directories and files, i intend to not only search them but also modify
them using normal unix tools" kind of use, or because somebody wants to
run multiple mh commands while preventing others from running
interstitially.

sounds like "mhlock" does both. so i'm fine with that.