Just for fun, here are screenshots of what a 20-minutes-to-develop OPAC might look like. It just uses the Django project's built-in admin screens over a simple mysql db and offers simple access to about 250,000 records from the medical and historical medical libraries I work for.

This was literally a 20-minute project. You can search for records by title, author, or imprint information, and filter by local location codes. This screenshot is the first result page from a search of all locations for "darwin":

And this is the same search, restricted to the location code for "historical medical, locked stacks" (i.e. the really cool but complicated stuff):

The data is from an extract from voyager generated using a tool written by staff at the Yale Libraries' Integrated Library Technology Services group.

suki, the repository into which I'm putting the full bib records in multiple formats for all this data, will soon actually own a copy of all this data and I can hook a more-than-20-minute search interface up to it for real. Still, cool, eh?