Bad diagnostic for any P!

[AnthonyLatsis]

Description

No response

Reproduction

protocol P {}

// error: using '!' is not allowed here; perhaps '?' was intended? [implicitly_unwrapped_optional_in_illegal_position]
// error: optional 'any' type must be written '(any P2)?' [incorrect_optional_any]
let _: any P!

Expected behavior

A single error, optional 'any' type must be written '(any P)!', as in the case of let _: any P?.

Environment

Swift version 6.0-dev (LLVM 48dc0c6fa01ce9e, Swift 19fbe5b)
Target: x86_64-apple-macosx14.0

Additional information

This https://github.com/apple/swift/labels/good%20first%20issue will suit an experienced Swift user.

[CrazyFanFan]

@AnthonyLatsis Would modifying PrintOptions::forDiagnosticArguments() be the correct solution to this problem? If so, I believe I could attempt to resolve it.

[AnthonyLatsis]

No, this is a problem with the diagnosis of invalid code, not the format used to print diagnostic messages.

[fabricerenard12]

Hi AnthonyLatsis, if no one is already addressing this issue, I would love to start working on it. Could you please assign it to me?

[iMostfa]

Hi @AnthonyLatsis i'm interest in this issue,

i've made a PoC solution for this issue here, i'm not sure if it's the right way to fix it though, 71d3c4f

i wanted to avoid touching TypeResolver::resolveImplicitlyUnwrappedOptionalType
and instead try to fix it from TypeResolver::resolveExistentialType

but i found that it's not possible as already too late we get a diagnostics emitted from: TypeResolver::resolveImplicitlyUnwrappedOptionalType before we emit the another (correct diagnostics in this case) from resolveExistentialType:

i ran tests in sema and it looks like my fix doesn't cause any regressions, what do you think ?

this's the current state after applying my fix:

and tests in sema are ✅

would apperciate if i get this one assigned to me, and a hint if my approach needs to be changed

[AnthonyLatsis]

Hi, Mostfa! We prefer doing code review through pull requests, so you are very much encouraged to open one to apple:main if you are ready to discuss a solution.

[iMostfa]

Thanks @AnthonyLatsis For the quick response

PR Opened #74461