r? @dingxiangfei2009

rustbot has assigned @dingxiangfei2009.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

Yea, it's def a better fix than what the other PRs attempted. I tried finding a way that rejects this earlier, but it doesn't seem obvious to me how to do it robustly.

Your fix feels like it carries a risk because we are ignoring a buggy situation, but maybe that can be alleviated by asserting that an error has already been reported

Your fix feels like it carries a risk because we are ignoring a buggy situation, but maybe that can be alleviated by asserting that an error has already been reported

I see, in that case, would using span_delayed_bug() be appropriate?

edit: like

BodyOwnerKind::Closure if self.is_const_fn(def_id) => {
    if let Some(ctx) = self.hir_body_const_context(self.local_parent(local_def_id)) {
        return Some(ctx);
    }
    self.dcx().span_delayed_bug(
        self.def_span(def_id),
        "msg",
    );
    return Some(ConstContext::ConstFn);
}

That works, but in this case we can even just be an assert that errors were reported. We have like 2 of them in rustc, and I'm on mobile, so I can't find the function rn, but there is a function for getting an ErrorReported if any errors were emitted

there is a function for getting an ErrorReported if any errors were emitted

Maybe you mean DiagCtxtHandle::has_errors(). I’ve just pushed a version that uses it.

📌 Commit 9be8fca has been approved by oli-obk

It is now in the queue for this repository.