set_at_reference type catching

[JacobHast]

Currently set_at_reference allows you to overwrite an attribute with a wrong type in a way that cannot be caught by type-checkers. Could we have, e.g., a disallow_type_mismatch bool kwarg which checks for matching types and raises an error if a non-matching type is provided as value?

[nulinspiratie]

I'm unsure whether to go ahead with this one. It's similar to Pydantic's approach, but this brings about a few additional considerations:

• Should type matching only be enforced for exact types, or should it try to convert types? If not implemented properly, this can result in unexpected results.
• Should it only happen for set_at_reference, or also in general when trying to set an attribute's value? I would argue the latter, but this would be a pretty major breaking change.

In general Pydantic's approach of typecasting makes sense, but it's a rather strict rule, and has significant code related to validation and casting. If it were to be implemented, maybe a good approach would be to have it as a global optional setting.
Any thoughts on this @JacobHast and @natangqm ?

[JacobHast]

For me it would be fine if types are not converted. And for my use-cases it's enough if it only happens at set_at_reference, as type-checkers can check the case of directly setting the attribute.
I'll leave it to you if you want to go full Pydantic at some point :)