update to Unicode 15.1

[JLHwung]

No description provided.

[JLHwung]

The Unicode 15.1 has marked U+30FB and U+FF65 as Other_ID_Continue

https://www.unicode.org/Public/15.1.0/ucd/PropList.txt

so they are now valid ES6+ identifiers again, though the browsers are not catching up yet.

@evanw Do you want to en-quote them in literal keys for browser compat? Or do you think we can move forward and just remove this test

esbuild/internal/js_parser/js_parser_test.go

Lines 722 to 734 in 02dae18

	func TestUnicodeIdentifierNames(t *testing.T) {
	// There are two code points that are valid in identifiers in ES5 but not in ES6+:
	//
	// U+30FB KATAKANA MIDDLE DOT
	// U+FF65 HALFWIDTH KATAKANA MIDDLE DOT
	//
	expectPrinted(t, "x = {x・: 0}", "x = { \"x・\": 0 };\n")
	expectPrinted(t, "x = {x･: 0}", "x = { \"x･\": 0 };\n")
	expectPrinted(t, "x = {xπ: 0}", "x = { xπ: 0 };\n")
	expectPrinted(t, "x = y.x・", "x = y[\"x・\"];\n")
	expectPrinted(t, "x = y.x･", "x = y[\"x･\"];\n")
	expectPrinted(t, "x = y.xπ", "x = y.xπ;\n")
	}

Babel updated to Unicode 15.1 three weeks ago and no one ever complain that the katagana middle dot is not quoted anymore.

[evanw]

I think they still have to be quoted. According to this, Unicode 4.1 through Unicode 15 omitted these characters from ID_Continue by accident. Unicode 15.1 adds them back but they are still syntax errors in older browsers. For example, Firefox 118 throws Uncaught SyntaxError: illegal character U+30FB when trying to run the unquoted code.

[JLHwung]

Thanks for the background update, it is very helpful.

On the other hand, U+30FB will be valid ID_Continue again, I think we should start accepting these two in IdentifierName, but quote them in literal object key for browser compatibility. Otherwise once browsers update to Unicode 15.1, it will seems like a bug when browser accepts U+30FB in ID_Continue but esbuild throws on them.

Ah we are already taking this approach, now it looks good to me.