Add check to ensure generics for sorts are correct

[vrindisbacher]

Closes (#820)

[vrindisbacher]

Checking works, but I'm not sure this error is particularly helpful...

Code:

use flux_rs::*;

#[refined_by(n: int)]
struct S {
    #[field(i32[n])]
    f: i32,
}

defs! {
    fn foo(s: S<int>) -> int {
        s.n
    }
}

Output:

error[E0999]: cannot instantiate base generic with opaque type or a type parameter of kind type
 --> ../sort_gen.rs:8:1
  |
8 | struct S {
  | ^^^^^^^^
-Ztrack-diagnostics: created at crates/flux-fhir-analysis/src/wf/sortck.rs:394:18

error: aborting due to 1 previous error

It would be nice if we could point to the flux def itself and say something like:

Argument int supplied but definition for S has no generics. Either remove the generic type supplied to sort S or add missing generic type to S

[vrindisbacher]

Example input:

#![allow(unused)]

use flux_rs::*;

#[refined_by(n: int)]
struct S {
    #[field(i32[n])]
    f: i32,
}

defs! {
    fn foo(s: S<int>) -> int {
        s.n
    }
}

Output:

error[E0999]: sorts associated with this struct should have no generic arguments but 1 generic argument supplied
  --> ../sorts.rs:12:15
   |
12 |     fn foo(s: S<int>) -> int {
   |               ^ expected no generic arguments on sort
-Ztrack-diagnostics: created at crates/flux-fhir-analysis/src/conv/mod.rs:2017:40

error: aborting due to 1 previous error

Another example:

#![allow(unused)]

use flux_rs::*;

#[refined_by(n: int)]
struct S {
    #[field(i32[n])]
    f: i32,
}

defs! {
    fn foo(x: int<bool>) -> int {
        1
    }
}

And output:

error[E0999]: primitive sort int expects no generics but found 1
  --> ../sorts.rs:16:15
   |
16 |     fn foo(x: int<bool>) -> int {
   |               ^^^ Incorrect generics on primitive sort
-Ztrack-diagnostics: created at crates/flux-fhir-analysis/src/conv/mod.rs:2043:17

error: aborting due to 1 previous error

[nilehmann]

Cool, this looks good. I like having separate messages for primitive sorts and adt sorts.

There are a couple of extra places where we should report an error. I suppose we can have a different message for each of them but I'm fine with lumping them together in a generic "this sort doesn't have the right number of arguments" message.

Please also add some tests inside tests/tests/neg/error_messages/conv/

Feel free to merge after fixing the comments or ping me if something is not clear.

[vrindisbacher]

@nilehmann, I think this is almost good to go, but I'm confused by what's going on here tests/tests/neg/error_messages/conv/mismatched_generics.rs

EDIT: Seems like the same thing is happening with Self<int>...

[nilehmann]

@vrindisbacher I think you put the wrong #[diagnostic] attributes. Both the error structs for self and opaque have

#[diag(fhir_analysis_generics_on_type_parameter, code = E0999)]

[vrindisbacher]

doh! thanks... can't believe i didn't catch that...