Database experiences in Python: Good or Bad?

[Jeff Schedin]

I am thinking of getting into Python but before I take the plunge I
wanted to get some opinions on how the database connectivity is
actually working in python since the work I will be doing will be
heavy on getting data from Oracle, MSSQL and DB2 databases off of all
sorts of OS'.

(Currently I am programming on an AS/400 in RPG which due to the
integrated nature of OS/400 DB2 is native - no drivers I just declare
the file and read or embed the SQL directly into the code. Needless to
say all of this monkey business with drivers makes me a little
nervous.)

I have looked at the DB SIG and think that if the drivers actually
implement all of the features I will be fine.  Some of the drivers
look a little scary to me - this could just be my perception of course
- while others look like a lot of work (mxODBC - which looks good
requires some compiling ?)  Some of my main concerns are that metadata
be available and that the drivers are activly maintained either
privately or by an active community.

It seems that the Open Source databases are fine in this area:
PostgreSQL, MySQL etc  but what about DB2/Oracle/MSSQL?  I don't mind
having to do a little extra coding I just want to be able to download
a driver (I will even compile if the instructions are decent) that I
know will work and that will evolve as the Database Vendors do.

Java's JDBC api seems to be well supported by most databases so would
Jython be a better approach?  Since databases are the heart of many
projects I must assume that things working fine but from my basic
research (ok it was just ala google) I don't seem too confident in
Python's DB-API compliance or support.

Perhaps I am just thinking about the whole topic incorrectly (it sure
feels like it).  I mean don't most DBs expose an API which these
drivers are then just coded against?  All wisdom appreciated.


Opinions?  

Thanks, 

Jeff