Fix lazy namespace resolution of relations.

[juhovh]

Earlier the relations used lazy entity resolving from the current namespace, but they should be resolve the entity from the original namespace instead. Otherwise it is possible that if an entity is created dynamically in a different namespace than where it is used, it fails to resolve the referred entities.

For more information and a test case see the mailing list post at https://groups.google.com/forum/#!topic/sqlkorma/Qfr7szJUFdo

[immoh]

The fix is fine, but please add a failing test case that gets fixed by this. Thanks!

[juhovh]

Added a failing test case, as mentioned this is a bit tricky to test since I need to add an additional namespace for this.

[immoh]

I don't think it is especially tricky, you need another namespace to test this. It could be done in one file but I think it's fine to do it like this.

By the way, I think the issue can be reproduced in a simpler way by using defentity in two different namespaces:

If you have

(ns a
  (:use korma.core))

(defentity foo)

and

(ns b
  (:use korma.core)
  (:require a))

(defentity bar 
  (has-one a/foo))

then

(select bar (with a/foo))

will throw.

Could you squash both fix and test into single commit? Thanks.

[juhovh]

Moved the friend entity to core namespace, changed the test to use the definitions as suggested and squashed it all into single commit.

[immoh]

Thanks!

[juhovh]

There's quite many changes since release 0.4.0 and I'm keeping a custom korma binary in my repository to make the code work, any plans for the next release?

[immoh]

I'm planning to cut a release by the end of the week.

[juhovh]

Cool :+1: