On Fri, Jun 30, 2006 at 05:55:53PM -0700, Tracy R Reed wrote:
> 
> Neat idea but as this guy points out it is hard to make these two 
> different ways of design and thinking (relational vs object) mesh 
> nicely. This is one reason why I have been studying ZODB.
> 
> http://blogs.tedneward.com/PermaLink,guid,33e0e84c-1a82-4362-bb15-eb18a1a1d91f.aspx
> 
> His Vietnam analogy is quite strained but after that he gets into good 
> details about the difficulties.

I generally find ORMs rather disappointing.  They are either featureful and
slow (hibernate, sqlalchemy) or they just get in the way of making full use
of the sql engine (SQLObject).  I find that it's usually efficacious to just
do a straight sql call for reports (like a big master list) and instantiate
objects as needed (e.g. for a detail form.)

ZODB is great if startup time is not a big factor.  In particular, if you
have to traverse a big object graph because you need to sort it first, it
can be quite costly (the same issue that the ORMs have, I guess, but you
don't have the option of going directly to a relational engine to get around
it).  Once your objects are loaded into memory it's pretty fast.  

With an OODB you'll have to do all the indexing and integrity checks
yourself.

There's a new Berkeley DB backend for the Durus OODB that claims to have
a fast startup time, but I haven't had a chance to try it:

http://www.argo.es/~jcea/programacion/durus-berkeleydbstorage.htm

You might also want to check out buzhug, but it also has the same startup
time issue.

Dave

-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-lpsg

Reply via email to