Table reflection broken with InformixDB

[Former user]

The Informix dialect references the attribute ''owner'' in sqlalchemy.Table (''lib/sqlalchemy/databases/informix.py:256''), which does not seem to ever get set in the Table constructor (the other dialects use ''schema'' instead). There also appears to be a typo in the type mapping dictionary (''lib/sqlalchemy/databases/informix.py:164'') causing an exception to be raised when reflecting tables with SMALLFLOAT columns.

I've attached a patch against 0.5.6 that works for me in the hopes that it will be useful. Note though this is the first time I've ever looked at sqlalchemy internals, so I'm not at all sure I'm doing the right thing.

[Former user]

informix dialect table reflection patch

[Mike Bayer]

the informix dialect was donated somewhat anonymously some years ago and has never been tested, since nobody has access to an informix test environment. If you have the resources to help, we mostly need the dialect fixed up for 0.6, which is the current SQLAlchemy trunk. The structure of dialects has changed somewhat between 0.5 and 0.6 (not too much though and lots of folks have been contributing dialects for 0.6). If you can spare some cycles that would be a huge help.

[Former user]

I can't promise I'll have the time to do this, but I'll do what I can. I should also say that I'm a sqlalchemy newbie. Anyway, I'm currently using the informix dialect behind a web-service (the db is non-transactional), so it will get some testing that way. I'll submit additional patches as needed. Regarding 0.6: is there documentation on how to implement dialects (and related sqlite internals) somewhere, or should I just use an existing dialect as a model? I assume the informix dialect test cases would need fleshing out; can you clarify what a dialect test case should cover?

[Mike Bayer]

thanks for the patch ! this is in the 0.5 branch in f9487913f860049a7c7801793f5e88224a6b3748. Its good to know the informix dialect still works, will see if I can find time to to the 0.6 port which is straightforward - discussion there is #1499.

[Mike Bayer]

Removing milestone: 0.5.7 (automated comment)