exclude const generic arguments from generic_types

[BKSalman]

resolves #626

nvm got help from @aumetra and solved it :)

[juhaku]

Just wondering whether it actually should work with const as well if so, then the fix would be more evolved. Yet at the moment it is not working with consts so I guess it is just fine to ignore them though. But most likely users with a struct having generic const variable it would still fail but with a different error than now.

However that can be another PR later on. I'll create an issue to the kanban board to track this if I ever get to test the cons support.

[BKSalman]

Just wondering whether it actually should work with const as well if so, then the fix would be more evolved. Yet at the moment it is not working with consts so I guess it is just fine to ignore them though. But most likely users with a struct having generic const variable it would still fail but with a different error than now.

However that can be another PR later on. I'll create an issue to the kanban board to track this if I ever get to test the cons support.

sounds good :)

I was actually trying to somehow change the code to make it also support consts (not just filter it out), but I didn't even know what the idea of it would be, or where it can be used in the generated docs

[aumetra]

or where it can be used in the generated docs

Yeah, whatever can be done with const-generics (at the moment) seems somewhat irrelevant to OpenAPI docs (I could also just not be aware of use-cases).

But most likely users with a struct having generic const variable it would still fail but with a different error than now.

Do you mean with a constant associated type? Something like Foo<BAR = 2>? These will receive the same errors presumably, since they fall under the branch GenericArgument::AssocConst, which is covered by utoipa to fail with the same error.

If you mean something like Foo<2>, it would fall under the same branch (GenericArgument::Const) that was patched with this PR and would be ignored.

[juhaku]

Do you mean with a constant associated type? Something like Foo<BAR = 2>? These will receive the same errors presumably, since they fall under the branch GenericArgument::AssocConst, which is covered by utoipa to fail with the same error.

If you mean something like Foo<2>, it would fall under the same branch (GenericArgument::Const) that was patched with this PR and would be ignored.

Either of those, frankly I haven't looked in to the details how or what could be done with a const generic type. For utoipa OpenAPI docs it indeed does not make much sense and even a generic ToSchema type should be used with #[aliases] that tell the actual types of the generic arg anyway.