问题
I'm trying to implement the proper architecture for multiple databases under Python + Pylons. I can't put everything in the config files since one of the database connections requires the connection info from a previous database connection (sharding).
What's the best way to implement such an infrastructure?
回答1:
Pylons's template configures the database in config/environment.py, probably with the engine_from_config method. It finds all the config settings with a particular prefix and passes them as keyword arguments to create_engine.
You can just replace that with a few calls to sqlalchemy.create_engine() with the per-engine url, and common username, and password from your config file.
来源：https://stackoverflow.com/questions/2205047/multiple-database-connections-with-python-pylons-sqlalchemy