Classify nameof<'T> & match … with nameof ident -> … correctly

[brianrourkeboll]

Description

Fixes #10026.

• Record the name resolution of the nameof identifier in type application expressions and patterns to enable correct colorization:
  • Resolve nameof in nameof<'T>.
  • Resolve nameof in match … with nameof ident -> ….

Before

After

Checklist

• Test cases added.
• Release notes entry updated.

[github-actions]

❗ Release notes required

---

✅ Found changes and release notes in following paths:

Change path	Release notes path	Description
src/Compiler	docs/release-notes/.FSharp.Compiler.Service/9.0.300.md

[edgarfgp]

Thanks for this Brian. Will this also cover the cases match ... with | ... -> nameof(..) function | ... -> nameof(...) ?

[brianrourkeboll]

Thanks for this Brian. Will this also cover the cases match ... with | ... -> nameof(..) function | ... -> nameof(...) ?

In the result expression of match clauses? It's already colorized correctly there, no?

This PR covers (1) generic nameof<'T> in expressions and (2) nameof ident in patterns (whether in a match clause or anywhere else).

[edgarfgp]

Thanks for this Brian. Will this also cover the cases match ... with | ... -> nameof(..) function | ... -> nameof(...) ?

In the result expression of match clauses? It's already colorized correctly there, no?

This PR covers (1) generic nameof<'T> in expressions and (2) nameof ident in patterns (whether in a match clause or anywhere else).

In Rider currently looks like this

[brianrourkeboll]

In Rider currently looks like this

I don't have Rider installed on this machine, but what do other "intrinsic functions" (raise, reraise, typeof, typedefof, sizeof) look like in that position?

What does nameof look like in

let f x = nameof x

?

FCS classifies nameof as SemanticClassificationType.IntrinsicFunction (assuming the name resolution has been recorded properly):

fsharp/src/Compiler/Service/SemanticClassification.fs

Line 34 in ad29712

| IntrinsicFunction = 11

fsharp/src/Compiler/Service/SemanticClassification.fs

Lines 157 to 168 in ad29712

	let (|KeywordIntrinsicValue|_|) (vref: ValRef) =
	if
	valRefEq g g.raise_vref vref
	|| valRefEq g g.reraise_vref vref
	|| valRefEq g g.typeof_vref vref
	|| valRefEq g g.typedefof_vref vref
	|| valRefEq g g.sizeof_vref vref
	|| valRefEq g g.nameof_vref vref
	then
	Some()
	else
	None

fsharp/src/Compiler/Service/SemanticClassification.fs

Line 218 in ad29712

| Item.Value KeywordIntrinsicValue, ItemOccurrence.Use, m -> add m SemanticClassificationType.IntrinsicFunction

The VS integration maps that to its own, separate classification, in the same category as keywords:

fsharp/vsintegration/src/FSharp.Editor/Classification/ClassificationDefinitions.fs

Line 57 in ad29712

| SemanticClassificationType.IntrinsicFunction -> ClassificationTypeNames.Keyword

It looks like Rider doesn't use the FCS API here; it must be doing its own classification: https://github.com/search?q=repo%3AJetBrains%2Fresharper-fsharp+SemanticClassificationType&type=code

My guess is that this PR will at least make the nameofs on the left-hand side of the -> have the same color as the nameofs on the right-hand side in Rider, though. (And in VS Code, since Ionide also does not color intrinsic functions as keywords.)

[auduchinok]

@brianrourkeboll Cool, thanks! Could you look at these two things, please? 🙂

[psfinaki]

Nice. Thank you @brianrourkeboll :)