Using multiple ORMs? - And SQLalchemy vs Pony vs Peewee vs stdnet vs …

[Alec Taylor]

Investigating possible using multiple ORMs in my project.

Toy project, want to make it as generic as humanly possible; whilst
still exposing abstract pythonic interfaces.

E.g.: support most number of backends, including SQL ones like:
Postgres, SQLite, MySQL, … and NoSQL ones such as Redis (using
python-stdnet).

One way of doing this is to write generic Python libraries; with
different models, insert and query lines for each of the backend ORMs.

What are your thoughts on this?

Additionally, there are a variety of new ORMs popping up, what are
your thoughts on them; and is there something other than the 4
mentioned in the subject which I should look into?

Thanks for all suggestions,

Alec Taylor

[Wolfgang Keller]

> Thanks for all suggestions,

Two essential criteria:

If an ORM only allows 1:1 mapping between classes and tables à la
"active record", then it's entirely pointless.

And if an ORM allows only surrogate keys, then its developers don't
have a clue of databases or they don't give a darn. Or both.

Sincerely,

Wolfgang