On 2000-03-15 15:25:56 -0500,
[EMAIL PROTECTED] wrote:
> Yes, mh has a problem with that. The simple solution
> is just to avoid any form of concurrent access to
> folders. If you only use mh commands from the shell
> (i.e. no rcvstore), that's the case. Mutt can't be
> faulted by mh users for doing what mh does in similar
> circumstances; conversely, you shouldn't make it worse
> by ignoring mh_sequences altogether.
Frankly, I don't like the idea to put in code which either
goes to lengths for working around a broken folder format,
or is bound to fail, corrupt files, and the like.
Maybe the MH community will do their homework at some
point of time and specify a locking mechanism for the
.mh_sequences file, which would certainly be possible (and
be it that they just flock, fcntl-lock, or dotlock it).
--
http://www.guug.de/~roessler/
