[SR-10088] Subscript overloads are incorrectly allowed to differ by escaping-ness

[hamishknight]

Previous ID	SR-10088
Radar	None
Original Reporter	@hamishknight
Type	Bug
Status	Resolved
Resolution	Done

Environment

Swift version 5.0-dev (LLVM 94d957ca75, Swift f07a84a)
Target: x86_64-apple-darwin18.2.0

Additional Detail from JIRA

Votes	0
Component/s	Compiler
Labels	Bug, StarterBug, TypeChecker
Assignee	jumhyn (JIRA)
Priority	Medium

md5: d6e10743f8e36163cbd39a4e83fc2569

Issue Description:

For functions, we don't allow overloads to differ by escaping-ness:

struct S {
  func foo(_ fn: () -> Void) {}
  func foo(_ fn: @escaping () -> Void) {} // error: Invalid redeclaration of 'foo'
}

However we do allow this for subscripts:

struct S {
  subscript(_ fn: () -> Void) -> Int { return 0 }
  subscript(_ fn: @escaping () -> Void) -> Int { return 0 } // Okay
}

We should reject such subscript overloads like we do with functions.

[hamishknight]

Marking this as a StarterBug – the issue has to do with the overload types we assign to the declarations, from the function ValueDecl::getOverloadSignatureType() in Decl.cpp. For functions we use mapSignatureFunctionType in order to strip away things like @escaping for the purposes of redeclaration checking. We should ideally also be using this for subscripts.

Note that this is a source breaking change so would need to be guarded by a isSwiftVersionAtLeast check.

[slavapestov]

It's not that source breaking. It's unlikely anyone relies on this.

[swift-ci]

Comment by Frederick Kellison-Linn (JIRA)

Does that mean you don’t think it needs to be version-guarded?

If no one has started work on this yet I’d be happy to take it on.

[slavapestov]

I would always err on the side of simplicity. Run source compat tests first of course. We can always version-guard it later if we ship a public release with the change and people complain.

I think its rare to have subscripts take functions, period; overloading on escaping-ness is even more rare.

[hamishknight]

@slavapestov Fair enough – my main concerns were that subscripts with function parameters can come up with autoclosures, and that changing the overload signature type also affects cross-module shadowing. I do agree that it's very unlikely anyone is relying on overloading on escaping-ness though.

[swift-ci]

Comment by Frederick Kellison-Linn (JIRA)

Hey @xedin—just wanted to give you a heads up that I put a PR up for this issue here: #23457

Don't want you to waste time duplicating work! Sorry about that, should have assigned this bug to myself.

[xedin]

Thanks for letting me know!