nndbm

nndbm

[Joe Hildebrand]

I should have a rough first cut of nndbm ready when I get back to
Denver on Wednesday.  It is implemented as a server process, which
looks like an nntp server to your client software, but is actually
saving your mail in a gdbm or berkeley db database.  It should have
the advantage of speed and lack of locking issues, since there is only
one process reading and writing the database.  Oh, and no inode
sucking, like nnml.  It should work very well for large mail groups,
since there is no separate .overview file to munge through.

It should be pretty easy to extend to include pattern matching rules
so that mail gets sorted nicely into groups, so that you wouldn't have
to run procmail, as long as you were just doing simple filtering.

Right now, the only outstanding issue is security.  I'm going to look
into nntp authorization, but I have a feeling that it might not be
adequate.  Anyone have any ideas?

It is written in Perl5, and I'll release it under the GPL.  Is anyone
interested in beta testing?

-- 
Joe Hildebrand                  Fuentez Systems Concepts
hildjj@fuentez.com              Lead Software Engineer
	"Breakfast recapitulates phylogeny" - Spider Robinson

Re: nndbm

[Jason L Tibbitts III]

[CC'd to the list because it's of general interest]

>>>>> "Joe" == Joe Hildebrand <hildjj@fuentez.com> writes:

Joe> It is written in Perl5, and I'll release it under the GPL.  Is anyone
Joe> interested in beta testing?

I'd be happy to put the mailing list archives up in this form, if this is
something that runs over a network.  It would be a good "large scale" test
(~7500 articles and growing).  I have larger archives that I'd be willing
to experiment with, too.

I'll of course leave the old form of the archive around in perpetuity.
(Disk space?  What's that?)

 - J<