fix: record field names incorrectly emitted by genType

[illusionalsagacity]

I managed to find the source of this with Claude's help, though I don't have enough experience with ocaml to know for sure if this is an idiomatic fix.

resolves #8086

[pkg-pr-new]

Open in StackBlitz

rescript

npm i https://pkg.pr.new/rescript-lang/rescript@8087

@rescript/darwin-arm64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/darwin-arm64@8087

@rescript/darwin-x64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/darwin-x64@8087

@rescript/linux-arm64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/linux-arm64@8087

@rescript/linux-x64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/linux-x64@8087

@rescript/runtime

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/runtime@8087

@rescript/win32-x64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/win32-x64@8087

commit: 8586107

[cristianoc]

Is it possible to avoid quoting in the first place instead of un-quoting later?

[cknitt]

@cristianoc Codex says:

Avoiding the un-quoting here would mean normalizing escaped identifiers earlier (likely when the parser/typedtree builds Ident.t for escaped names). That’s a broader change and could affect other compiler stages that might be relying on the escaped form. The current fix is a localized workaround in genType that keeps the normalization close to the emission logic.

[cristianoc]

@cristianoc Codex says:

Avoiding the un-quoting here would mean normalizing escaped identifiers earlier (likely when the parser/typedtree builds Ident.t for escaped names). That’s a broader change and could affect other compiler stages that might be relying on the escaped form. The current fix is a localized workaround in genType that keeps the normalization close to the emission logic.

That's a generic laziness claim rather then: you need to change X, Y, Z. These are 23 places, not worth it.

[cristianoc]

that said, fighting laziness requires resources: such as knowing when to shrink the context, make more things available, know when to push back etc

So feel free to use your judgment on whether it's worth the effort and dig deeper. Or just merge as is.

[illusionalsagacity]

After cutting Opus loose on attempting to unquote, it reverted the unquoting change in

rescript/compiler/syntax/src/res_scanner.ml

Lines 340 to 348 in ff9d6d3

	let name = Ext_ident.unwrap_uppercase_exotic ident in
	if name = String.empty then (
	let end_pos = position scanner in
	scanner.err ~start_pos ~end_pos
	(Diagnostics.message "A quoted identifier can't be empty string.");
	Token.Lident ident)
	else if Ext_ident.is_uident name then Token.Lident ident
	(* Exotic ident with uppercase letter should be encoded to avoid confusing in OCaml parsetree *)
	else Token.Lident name

There's a test failure in exotic_labels_test.cmj with an assertion failure in compiler/ml/env.ml. This is the key issue - the parsetree is getting confused. Let me look at this test file and the env.ml assertion.

The error is:
Fatal error: exception File "compiler/ml/env.ml", line 838, characters 63-69: Assertion failed

This is exactly what the original comment was warning about - "Exotic ident with uppercase letter should be encoded to avoid confusing in OCaml parsetree".Let me look at the exotic_labels_test.res file and the env.ml file to understand what's happening.

Investigating the original pull request reveals critical insights about identifier management. The OCaml type checker fundamentally requires quoted uppercase identifiers for correct internal storage and lookup. Attempting to remove quoting at the scanner level would catastrophically break type checking mechanisms. The localized fix in the genType module represents the most appropriate solution.

it's quoted here with the comment introduced in this pr

[illusionalsagacity]

this fixes type \"UppercaseName" being emitted literally, breaking the typescript syntax