On Sat, Oct 10, 2009 at 07:47, Iwan Vosloo <[email protected]> wrote:

>> Well that global (ie provided by default) collection, session and
>> metadata was a poor design decision and this is one of the numerous
>> problems it brings. At some point, I should really think things
>> through and see how to get rid of them with minimal upgrade hassle.
>
> I'd be interested to see what you come up with...

Don't hold your breath though, it might take a while before I fix that
issue, if ever. Of course, a patch in that direction would help ;-).

>> > Neither do we want to be
>> > dependent on when what gets imported...
>>
>> Well, that's the only semi-clean (IMO) option currently available.
>
> The problem is that we don't always have control over that. And, it is
> such an unintuitive place to look for problems. Such a secondary
> impact of import order is almost invisible in a way.  For me the
> logical place to do something about it feels like setup_all (or
> something analogous to it)?
>
> We're doing the following now (before setup_all):
>
> elixir.session.configure(autoflush=True, transactional=True)
> elixir.session.registry.scopefunc = custom_scope_func

Well, I didn't dig that far into SQLAlchemy's code. But I thought this
would work until I saw your post on SQLAlchemy's list... Note that a
session is *not* thread-safe, so whatever you are trying to do needs
to take that into account (and that *might* be the reason your test is
failing).

-- 
Gaëtan de Menten
http://openhex.org

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"SQLElixir" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/sqlelixir?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to