On Sun, Jul 03, 2005 at 02:14:45PM -0400, Andrew Athan wrote: > >>I'm investigating embedded databases for an upcoming project, and I came > >>upon this thought: Wouldn't an SQLite pager that uses > >>Sleepycat/BerkleyDB be quite interesting?
Maybe you could clarify a bit more what you are proposing. Your subject line says "BerkeleyDB Pager", but in the text you ask about using the SQLite pager with a BerkeleyDB (backend?). Which one would be grafted on to which one, and what might be benefit be? > Basically I am thinking that some of BerkleyDB's locking granularity and > distribution/replication/etc. features might be interesting within the > context of SQLite -- and conversely, it would provide an SQL compliant > interface to what is otherwise a "B-Tree" style database. I'm guessing this means you are proposing to write a parallel to SQLite's btree.c that acts as a wrapper for the BerkeleyDB functions. If so, I would find this idea interesting, not so much for the specifics of BerkeleyDB, but as I've been thinking about how to backend SQLite to some other custom btree implementations. > Clay Dowling writes: > >It would be of fairly limited value for a lot of us however. For projects > >which don't release source code Sleepycat is rather expensive. Speaking > >as a program it has also not been my favorite database for administration > >issues. > > > Are we talking about the same sleepycat? BerkleyDB (Sleepycat) is open > source, available here: http://www.sleepycat.com/products/db.shtml I think Clay's point here is that while SQLite can be redistributed in closed source systems, BerkeleyDB cannot be unless one buys a license: http://www.sleepycat.com/download/licensinginfo.shtml --nate