Respect `#[allow(clippy::inconsistent_struct_constructor)]` on adt definition.

[rzvxa]

Description

Is it possible to add support to inconsistent_struct_constructors so it respects the allow attribute on the adt definition as well as on construction?

We are using a proc macro to reorder fields in repr C structs; Which means we have inconsistency between field definitions and their ABI; It would be really convenient for us to be able to mark the types with the allow attribute instead of having to use constructor methods for everything(or force this attribute to our downstream with isn't a possibility).

So are there any other rules that work in this way(respecting the attribute on both call site and definition)? I'm happily willing to create PR for this issue if I get the green light.

I'd appreciate any suggestions to resolve my issue, There might be some way of doing it which I'm missing here.

Version

No response

Additional Labels

No response

[Jarcho]

There are a few lints that do this (e.g. len_without_is_empty). Just needs a use of is_lint_allowed and fulfull_expectation on the right HIR node.

[rzvxa]

There are a few lints that do this (e.g. len_without_is_empty). Just needs a use of is_lint_allowed and fulfull_expectation on the right HIR node.

Thanks, I'd like to give it a try(It seems to be the fastest way to get it in). Is it OK if we assign this issue to me?

[Jarcho]

Just use @rustbot claim in a comment on the issue you wan to claim (without the backticks). You can read https://rustc-dev-guide.rust-lang.org/rustbot.html for more info.

[rzvxa]

@rustbot claim

[rzvxa]

@Jarcho Is it possible to use is_lint_allowed on an AdtDef where it is from a foreign crate? I can't find a way to get hir from non-local AdtDef of the struct type.

            && !expr.span.from_expansion()
            && let ty = cx.typeck_results().expr_ty(expr)
            && let Some(adt_def) = ty.ty_adt_def()
            && adt_def.is_struct()
            && !is_lint_allowed(cx, INCONSISTENT_STRUCT_CONSTRUCTOR, adt_def.to_hir_somehow?()) // what should go here?

If is_lint_allowed only operates on local types, How can we achieve this when the inconsistent type is marked with allow and exists outside of our crate?

[Jarcho]

You can't get lint attributes from other crates. I would probably just restrict the lint to only work with types from the current crate while you're at it. Upstream crates don't have a definition order from the point of view of the current crate (with the exception of #[repr(C)] structs).

You can use adt_def.did().as_local() to get an Option<LocalDefId> and cx.tcx.local_def_id_to_hir_id(_) get the HirId from there.

[rzvxa]

You can't get lint attributes from other crates. I would probably just restrict the lint to only work with types from the current crate while you're at it. Upstream crates don't have a definition order from the point of view of the current crate (with the exception of #[repr(C)] structs).

You can use adt_def.did().as_local() to get an Option<LocalDefId> and cx.tcx.local_def_id_to_hir_id(_) get the HirId from there.

Thanks for your quick response, I really appreciate it; Yep if we only lint where adt def is local it should be pretty trivial, It'd be almost identical to how len_zero handles it.