fix(ast_macros): raise compile error on invalid generate_derive input.

[rzvxa]

It checks 2 things. 1) The input is a supported derive 2) The given identifier is the same as the fully qualified target trait.

The latter makes sure that the trait for derive is included in the scope.

Part of oxc-project/backlog#124

Here's an expanded example of how we assert traits:

const _:() = {
    {
        trait AssertionTrait: ::oxc_allocator::CloneIn<'static> {}
        impl<T: CloneIn<'static>> AssertionTrait for T {}
    };
};

It makes sure CloneIn is the same as ::oxc_allocator::CloneIn and more importantly requires the user to include the trait if they wish to use it with generate_derive.

It also provides LSP jump to definition.

[rzvxa]

• fix(ast_macros): raise compile error on invalid generate_derive input. #4766 👈
• fix(ast, ast_codegen): use generate_derive instead of visitable for generating span derives. #4747
• refactor(ast_codegen): declutter the main file. #4744
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

Join @rzvxa and the rest of your teammates on Graphite

[codspeed-hq]

CodSpeed Performance Report

Merging #4766 will not alter performance

_{Comparing 08-08-fix_ast_macros_raise_compile_error_on_invalid_generate_derive_input (f5eeebd) with main (7ea058d)}

Summary

✅ 29 untouched benchmarks

[overlookmotel]

This is absolutely ingenious!

You can fool it with:

use oxc_span::GetSpan as GetSpanReal;
trait GetSpan: GetSpanReal {}

#[ast]
#[generate_derive(GetSpan)]
struct Foo { /* blah */ }

But I can't see any way to catch that one, and anyone who does such a bizarre thing probably deserves everything they get!

I've pushed 2 more commits.

1. Adding comments. I've removed the stuff about "all computation is cached... without any complex operation" because I don't think we can reasonably claim there's no complexity any more.
2. A nit - just removing an unnecessary ; from generated code.

If you're happy with those changes, please go ahead and merge.

[rzvxa]

You can fool it with:

use oxc_span::GetSpan as GetSpanReal;
trait GetSpan: GetSpanReal {}

#[ast]
#[generate_derive(GetSpan)]
struct Foo { /* blah */ }

But I can't see any way to catch that one, and anyone who does such a bizarre thing probably deserves everything they get!

Crap, You are right; And I can't figure out how exactly we can prevent it. If I can think of one I'll follow it up with another PR.

[graphite-app]

Merge activity

• Aug 8, 3:05 PM EDT: The merge label 'merge' was detected. This PR will be added to the Graphite merge queue once it meets the requirements.
• Aug 8, 3:06 PM EDT: rzvxa added this pull request to the Graphite merge queue.
• Aug 8, 3:09 PM EDT: rzvxa merged this pull request with the Graphite merge queue.

[overlookmotel]

Crap, You are right; And I can't figure out how exactly we can prevent it. If I can think of one I'll follow it up with another PR.

Just to say, this is a tiny point. For all practical purposes, this works. And I noticed this morning that "jump to definition" now works too, which is brilliant.

[rzvxa]

Crap, You are right; And I can't figure out how exactly we can prevent it. If I can think of one I'll follow it up with another PR.

Just to say, this is a tiny point. For all practical purposes, this works. And I noticed this morning that "jump to definition" now works too, which is brilliant.

Yeah, the jump to definition is a nice plus.