Embedded models have to be defined before any normal models that reference them about odmantic

[rollebolle123]

No, sorry, I am not trying to be condescending.
And maybe I did not understand the original problem. Aaaaand, I had no idea you could put quotes around types to forward declare them. Then it makes sense to me :D

[rollebolle123]

But, that makes sense? :O The embedded model should be defined first so a definition exists?

Also:
class Bar(Model):
bar: "Foo" <------ should lose the quotation marks?

aaaand
[Index(asc(Bar.bar.foo)] <--- missing a parentheses :D :
[Index(asc(Bar.bar.foo))]

[duhby]

Well as I mentioned, it doesn't matter if the quotations are there or not for both cases, and I didn't test the example code specifically so I missed the parentheses. I just think it should be specified in the documentation because it's not obvious that it should be defined before (other odm libraries don't have this behavior).

When you say "the embedded model should be defined first so a definition exists," that should (intuitively) only be necessary for the type hint if you want to define it without quotations. The reason I say it's not intuitive or obvious is because by the time any code is run regarding creating indexes or creating search queries, both of the models have already been defined.

I'm not saying the behavior is bad, it just needs to be clarified somewhere, and it might be unintended.

[duhby]

Also, what's with the unnecessary condescension? You clearly didn't take the time to fully understand the original post.