> Would it be beneficial to get rid of MAXUSERS and use dynamic tables?
> Where need be a dynamic pool could be backed up with an index array
> which would rarely be resized... (maybe growing ^2 when required)?

I started writing some of them earlier in the year.

I have a version of the pid allocater and an associated pfind
and pgfind that doesn't rely on a hash table.
No searches are ever done, so the code is (should be) faster
than the current one.

I'll repost it if anyone is interested.
Unfortunately I've also removed the p_cred structure which
makes separating out some diffs rather tricky :-(

	David

-- 
David Laight: david@l8s.co.uk