Standalone client database for Python?

[Aaron Watters]

Harry George <harry.g.george at boeing.com> wrote in message news:<xqxfzfjlpfg.fsf at cola2.ca.boeing.com>...
> Gustavo Campanelli <birdiepageANTI at SPAMciudad.FILTERcom.ar> writes:
> For "most maintained", mysql and postgresql would be hard to beat in
> client/server space.  For embedded dbms's, I'd guess Berkeley DB would
> be the winner (http://www.sleepycat.com/)

Mysql and postgressql are also huge and complicated and not
easily distributed (eg if you want to install on both windows
and unix), unless something has changed.  Berkeley DB is more
correctly called a btree and indexing library rather than a
true database in the tradition of SQL databases.

My favorite of courseis gadfly http://gadfly.sourceforge.net 
which is

- small
- 100% python (with an optional add on C module for speed)
- SQL based, with recovery and transaction support
- in memory (so it's not appropriate for huge databases,
  but it's pretty fast for small ones)
- originally written by me :)

I've also heard good things about pySQLLite and metakit although
they both require installing extension modules and supporting
libraries.

Lately I've been working on http://xsdb.sourceforge.net which is
also 100% python but more appropriate for larger databases.
Just added support for using berkeley db as a back end (in CVS).
I'm hoping to bridge xsdb and gadfly together at some point :).

   -- Aaron Watters

===
nothing exceeds like excess