buzhug is a new, fast, pure-Python database engine, using a pythonic syntax (no SQL). It is published at http://buzhug.sourceforge.net under the BSD licence Here is an overview of the interface : from buzhug import Base my_cds = Base('my_cds') my_cds.create(('artist',str),('title',str),('issued',int)) my_cds.insert(artist='Kaiser Chiefs',title='Employment',issued=2005) my_cds.insert(artist='Rialto',title='Night On Earth',issued=2002) my_cds.insert('Oasis','Definitely Maybe',1994) cd = my_cds.select(artist="Oasis")[0] print cd.title > "Definitely Maybe" new_cds = [ cd for cd in my_cds if cd.issued > 2000 ] Pretty straightforward, isn't it ? As you can see on the last line, the database is an iterator, yielding objects which have attributes of the same name as the fields in the base songs = Base('songs') songs.create(('title',str),('cd',my_cds)) cd = my_cds.select(title="Definitely Maybe")[0] song_id = songs.insert('Supersonic',cd) song = songs[song_id] # lookup by record id print song.cd.artist > "Oasis" A field can be a reference to another database. When you have finished entering all the songs you can get the track listing by [ song.title for song in songs if song.cd.title == "Definitely Maybe" ] A complete documentation, with a tutorial, is available on the web site The implementation has been designed to make all operations, especially selection, as fast as possible, while processing the data on disk (it is not an in-memory database). On a limited set of tests I found that it is much faster than gadfly and KirbyBase, and only less than 3 times slower than SQLite This is still a beta version, so I need feedback on the syntax, performance, bug reports etc. Please send any comment or question to the Google group : http://groups.google.com/group/buzhug?lnk=li Regards, Pierre