Fix parsing invalid function parameters

[kimdv]

Fixes #2254

[kimdv]

@swift-ci please test

[ahoppen]

Thanks. Looks good to me. I think just a few tests need to be updated.

[kimdv]

Thanks. Looks good to me. I think just a few tests need to be updated.

Not sure what is right.
We have the test case testInitAccessorsWithDefaultValues.

It will change init(initialValue: <#type#>) {} -> init(<#identifier#>: initialValue) {}.

I think this is more wrong.

Would it make sense to assume that if it starts with a lower case, it's an identifier and if it starts with an uppercase it's a type?

[ahoppen]

Would it make sense to assume that if it starts with a lower case, it's an identifier and if it starts with an uppercase it's a type?

I think that’s very reasonable behavior for error recovery. I have wanted to do this in some case in the past but I think that change never made to to a PR.

[kimdv]

Would it make sense to assume that if it starts with a lower case, it's an identifier and if it starts with an uppercase it's a type?

I think that’s very reasonable behavior for error recovery. I have wanted to do this in some case in the past but I think that change never made to to a PR.

I was digging a bit into it.
Where do you think is a propper place, would it be canParseType()?

[ahoppen]

I would try shoving the uppercase check into the line that currently has.

 let canParseType = withLookahead { $0.canParseType() && $0.at(.comma, .rightParen) }

[kimdv]

@swift-ci please test

[ahoppen]

Thanks, Kim. This looks good to me.

[ahoppen]

The macOS formatting failure is unrelated. We’re currently working on a fix, should be fixed tomorrow-ish, could you re-trigger CI then?

[kimdv]

The macOS formatting failure is unrelated. We’re currently working on a fix, should be fixed tomorrow-ish, could you re-trigger CI then?

Yes no problem. Will rebase and re-trigger.

[ahoppen]

For reference, this is the PR that fixes the format issue: swiftlang/swift#79380

[kimdv]

@swift-ci please test