by I need something that can hardly be called a database, but this seems
like a good group to ask in anyway.
What I want is a b-tree or similar storage scheme for variable-size
(usually small) records. The library need not have much more
functionality than good ol' dbm, but it must be fast and it must support
multiple concurrent accesses under Unix. "Fast" means a couple hundred
updates/sec on a several-hundred-meg database using a small workstation
or fast PC. And I want it to be thoroughly debugged and bulletproof ...
otherwise I'd just roll my own.
Any recommendations? Commercial product is OK, but I'd like to have
source code.
thanks for any info, tom lane