[sqlalchemy] semantics of tometadata

2010-09-16 Thread Chris Withers
Hi All, As part of looking into #1919, I see that if a table of the same name as the one passed to tometadata already exists in the destination metadata, then the table object passed in is ignored and the one already there is returned. That feels wrong to me. In the event there's already a

Re: [sqlalchemy] semantics of tometadata

2010-09-16 Thread Michael Bayer
On Sep 16, 2010, at 2:56 PM, Chris Withers wrote: Hi All, As part of looking into #1919, I see that if a table of the same name as the one passed to tometadata already exists in the destination metadata, then the table object passed in is ignored and the one already there is returned.

Re: [sqlalchemy] semantics of tometadata

2010-09-16 Thread Chris Withers
On 16/09/2010 20:26, Michael Bayer wrote: As such, I'd expect an exception to be raised rather than the other table object being returned. What do people feel about this? Im fine with tometadata raising for 0.7. a warning for 0.6 perhaps. Cool, done for 0.6. Where should I make the