collerek

That is fine to set as query params, easy to change if needed, one remark is that #37 should allow to not only filter but also sort by multiple columns...

Ok, fair enough. So when you have relation `A-B-C-D` all of them should be loaded by default when we apply your approach of `pydantic` model defining the relations to load?...

There are related queries for functions and subqueries in #242 and #227. That seems like a nice feature, but as you wrote that might be challenging to properly parse to...

Hi, sorry for the late answer. Unfortunately, right now it's not supported to order by aggregated functions, it can be an enhancement. Right now you have to reorder them in...

Looking at starlette docs it seems that graphene can also run async: https://www.starlette.io/graphql/ There is also: https://github.com/mirumee/ariadne https://github.com/strawberry-graphql/strawberry and (most basic but more pythonic?) https://github.com/ethe/pygraphy

Yeah it's more like a separate package that will ease the graphql queries construction. Not a part of the core. I have some outstanding tasks like changing the type hints...

Yes, the schema is not supported at that moment, related to #287 Can you check after installing from git with the branch from https://github.com/collerek/ormar/issues/287#issuecomment-1007319230

Hmm, what changed is that I no longer stringify the queries as that didn't work with complex types (i.e. postgres arays). Now I pass sqlalchemy query and bindparams to databases...

Yes, unfortunately it would break complex types, it was also using not really safe literal binds [docs](https://docs.sqlalchemy.org/en/13/faq/sqlexpressions.html?highlight=literal_bind#rendering-bound-parameters-inline). Although I was re-parsing it to sqlalchemy text clause you couldn't inject SQL...

The error is caused by using two separate metadata that don't know about each other. So each table is registered in a separate namespace. To achieve what you want you...