[orm-devel] post 1.0 and metaclasses

[Charles Brandt]

> This toppic being raised, I've decided to call the next stable orm
> release 1.0, bacause after that I'll change the infrastructure to use
> metaclasses. Also a lot of identifyers are going to change, too
> (practically all used by orm in dblcasses. I'll switch to Python's
> naming convention and use variable method names that look
> __like_this__ .) Wrapper classes for the old format will be provided
> and your current datamodels need not to be modified.

I'm curious to hear what benefit you see in metaclasses vs the current
approach.   It seems to me the added complexity detracts from the idea of a
thin membrane.  It also raises the barrier of entry for others to contribute
to the code. (or just follow along)  When you add this with some of the
limitations that (i think) are imposed by metaclasses (i.e. globally unique
class names) I'm not sure if there is a compelling case.  To me, this was
one of the main distinguishing *features* of Object Membrane from SQLObject.
Maybe features like automatic schema generation, or schema to class
translators would attract more users instead?    I'm interested to hear more
thoughts on the topic if anyone has them.

-charles.

relevant links:
http://mail.python.org/pipermail/python-list/2003-May/161119.html
http://www-106.ibm.com/developerworks/linux/library/l-pymeta.html