Query for non-primary mapper returns cached object for primary mapper

[Former user]

I use non-primary mapper with additional count field to gather statistics. This additional field is available when I query with clear session, but isn't when the object created using primary mapper exists in session cache (either just created and saved or selected with primary mapper).

[Mike Bayer]

theres not really any way to avoid this behavior since a non-primary mapper is typically used to select rows in a different way, but not to add any new properties to a class. if you used entity_name mappers, the loaded classes would have an identity distinct from those already loaded, but in that case the alternate entity_name mapper is also responsible for re-persisting the instance so is again probably not what you want.

to add a count field, the two easiest ways are either to create your primary mapper against, the select, or add the column at the query level via

session.query(MyClass).add_column(func.count(whatever)).filter_by(foo).list()