Re: [PATCH 2/3] rust: implement wrapper for acpi_object

[Gladyshev Ilya]

On 1/8/26 21:46, Miguel Ojeda wrote:
> > Using an enum would require keeping Rust's enum synced with the C side,
> > as well as implementing some simple but non-trivial checks that the
> > `type_` field is a valid enum value (and the valid range isn't
> > continuous). I think that keeping it as an integer is simpler and better
> > matches C side.
>
> If this refers to the `ACPI_TYPE_*` constants, there seems to be a
> comment in the C docs that requires it to be kept in sync already with
> elsewhere, so perhaps it could be reasonable to add one more place to
> sync? (Though I don't see the mentioned arrays from a quick grep?)
>
>       * NOTE: Types must be kept in sync with the global acpi_ns_properties
>       * and acpi_ns_type_names arrays.
>
> [...snip about bindgen...]

If I understand correctly, acpi_object is something untrusted in general 
and broken hardware can provide arbitrary _type value. So ergonomics of 
enum solution would be kinda strange:

```
type_id() -> Result<Enum> // Result because it can be invalid

// ...

// '?' here makes it look like non-trivial operation
if x.type_id()? == Enum::Buffer
```

And it's unclear should ACPI_TYPE_INVALID be Ok(INVALID) or Err...

Also users of AcpiObject generally doesn't care if received object has 
valid, but unexpected type or invalid type, so this conversion int->enum 
will introduce (small but) unneeded overhead.

That's why I preferred int-style approach, matching C side. Here is 
alternative proposal from my side:

1. Enum with type values like you want
2. This Enum implements .id() method that returns underlying <int>
3. AcpiObject::type_id() still returns <int> value
4. There is method to convert int into enum (like TryFrom)

Probably you can 'newtype' <int> and implement TryFrom AcpiTypeId -> 
Enum and write something like

```
match x.type_id().try_into()? /* or as_enum() or smth */ {
  Enum::Integer => /* ... */,
  Enum::String => /* ... */
}

```


---
Sorry for my chaotic explanations...