[sqlalchemy] Re: transaction control when mixing SQL and sqla

2010-12-17 Thread Laurence Rowe
See http://pypi.python.org/pypi/zope.sqlalchemy (quoting): By default, zope.sqlalchemy puts sessions in an 'active' state when they are first used. ORM write operations automatically move the session into a 'changed' state. This avoids unnecessary database commits. Sometimes it is necessary to in

[sqlalchemy] Re: transaction control when mixing SQL and sqla

2010-12-17 Thread Kent
I believe zope.sqlalchemy has no group/mailing list, so this message is specifically intended for zope.sqlalchemy developers (whom I'll point to this thread): Please read the above. Here is a script to demonstrate the behavior zope.sqlalchemy is causing. I believe is not an intended behavior (i.e

[sqlalchemy] Re: transaction control when mixing SQL and sqla

2010-12-17 Thread Kent
I had mistaken it for a SQLA flaw, but, from pdb, it appears the behavior I'm observing is being caused by zope.sqlalchemy. Thanks for your input. On Dec 16, 7:19 pm, Michael Bayer wrote: > That doesn't sound right.   Session.commit() issues the commit regardless of > whether or not flush() h