The way to hook most parts of the ORM is by creating a MapperExtension <http://www.sqlalchemy.org/docs/adv_datamapping.html#advdatamapping_exte nding>. Basically, define a subclass of MapperExtension, overriding whichever method you are interested in (possibly append_result or create_instance in your case), and pass an instance of this class as the 'extension' parameter to the mapper.
However, I think it would be worthwhile trying to understand exactly why the current recipe isn't working for you. (I see someone has updated it to put the session in the hashkey - I'm not sure I would have been brave enough to do that until at least one other person had said it was the right thing to do. I don't really know what I'm talking about ;-) ) If you have any places in your code where you create UniqueName objects without explicitly saying UniqueName(<whatever>), then they won't pass through the metaclass __call__ method, and therefore won't get added to the cache. This also includes situations where the object is loaded from a relation on another class. But I don't understand why that would be a problem, as that would only affect instances that already exist in the database, and because of the identity_map magic, you will only get one object in the session representing that row anyway. I think a testcase is probably needed - you could try and make a single script that starts two threads and performs operations on UniqueName obects, and see if you can get the same failure. Simon -----Original Message----- From: sqlalchemy@googlegroups.com [mailto:[EMAIL PROTECTED] On Behalf Of kris Sent: 18 June 2007 20:48 To: sqlalchemy Subject: [sqlalchemy] Re: UniqueObject recipe and turbogears I agree with your summary, I also noted that sqlalchemy doesn't really like to have objects that are linked together and in different sessions. I tried using hash_key = (session.context, name), but this failed in the same way. Loading the code with log statement, I note that some object are loaded not with the UniqueName loader, but probably by sqlalchemy itself. I believe it is those objects that are creating the conflict. Is there a way to hook the sqlalchemy loading system, so I can place the object in the cache? --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "sqlalchemy" group. To post to this group, send email to sqlalchemy@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sqlalchemy?hl=en -~----------~----~----~----~------~----~------~--~---
