Ruud H.G. van Tol wrote:
[...]
That is what you think that you need. I know you don't.
Another few questions/suggestions that come to mind:
1. Is whatever procmail is handing the messages *to* is capable of
handling more than one at a time. Would an optimized "procmaild"
actually be able to hand off the messages any faster? If mhonarc is "a
Perl program for converting mail or news messages into HTML archives",
is it going to be able to handle multiple simultaneous messages?
2. If ALL of the 10,000 messages are ultimately being handed off to the
same receiving program, is procmail really necessary? Could they simply
be handed off by formail? Is there another program that could convert
the existing file to the UUCP mailbox or MH mail folders required by
mhonarc?
3. If only a SUBSET of the 10,000 are to be processed, then perhaps use
formail + procmail to sort them into folders as individual message files
an initial pass, and have the "other thing" running in parallel on the
resulting files as they're sorted? In other words, have the receiving
program run interatively over the sorted folder. (This might have the
additional advantage of being kept running on an ongoing basis. procmail
receives incoming messages and places them in an appropriate folder,
then the other program is run periodically to process them as needed.)
4. Is the need to sort 10,000 messages like this a recurring
requirement, or is this just a one time thing?
I'm in the process of re-training various bayes tools with a collection
of 14,000+ spam and 6,000+ ham, so am sympathetic. However, in my
(admittedly limited) experience, procmail hasn't been the bottleneck
when calling other programs to do such work.
- Bob
____________________________________________________________
procmail mailing list Procmail homepage: http://www.procmail.org/
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail