[Bug]: No ambiguity between formals and nested procs of same name about chapel

[mppf]

The part I am not following is when these code patterns do emit an ambiguity error.

This does it:

record r { }
proc r.test() {
  var foo: int;
  proc r.foo { }
  foo;
}

Right. If it were proc foo then I would think it should be a redeclaration error; otherwise ambiguity. (But I wouldn't view a redeclaration error in this case as particularly bothersome; just conceptually it should be an ambiguity error because proc r.foo is adding something within record r which isn't in the same scope as var foo.

Except I'm realizing now that #25231 unintentionally changed this from an ambiguity error to a redeclaration error. I thought about that and had meant to avoid it, but I also don't know if it matters too much either way.

IMO it'd be reasonable to make an issue about the deficiency and move on to the next thing.

[mppf]

@riftEmber I'm not seeing that this is a dyno bug. On main with chpl, I am not seeing an ambiguity error in this case:

proc test(foo: int) {
  var foo: int; // this does not trigger an ambiguity error
  return foo;
}

var x = test(1);
writeln(x);

I think that we view the formals to be in a scope just outside of the function body. So var foo in this case is closer than the formal foo: int.

The part I am not following is when these code patterns do emit an ambiguity error.

[riftEmber]

Right. If it were proc foo then I would think it should be a redeclaration error; otherwise ambiguity. (But I wouldn't view a redeclaration error in this case as particularly bothersome; just conceptually it should be an ambiguity error because proc r.foo is adding something within record r which isn't in the same scope as var foo.

Okay, that makes sense.

IMO it'd be reasonable to make an issue about the deficiency and move on to the next thing.

Since I'm already warmed up on this I figured it would take about the same amount of time to just fix it as to make an issue, so I did: #25407

[riftEmber]

I'm not seeing that this is a dyno bug. On main with chpl, I am not seeing an ambiguity error in this case
I think that we view the formals to be in a scope just outside of the function body

Oh, whoops. I guess it's fine then.

The part I am not following is when these code patterns do emit an ambiguity error.

This does it:

record r { }
proc r.test() {
  var foo: int;
  proc r.foo { }
  foo;
}

Except I'm realizing now that #25231 unintentionally changed this from an ambiguity error to a redeclaration error. I thought about that and had meant to avoid it, but I also don't know if it matters too much either way.

[riftEmber]

Closing as not a bug