Unary minus on 'any' should infer type to 'number | bigint', not just 'number'

🔎 Search Terms

"unary minus", "negate bigint"

🕗 Version & Regression Information

• This is the behavior in every version I tried, and I reviewed the FAQ for entries about bigint

⏯ Playground Link

https://www.typescriptlang.org/play/?target=99&ts=5.7.2#code/MYewdgzgLgBARjAvDAtACgIxgJQG4BQA9ITKTAHoD8+okscATEqnAcWRdTeNDGM+gx4iJMlW50+TZCjBtRpcRN4BDAFwwVYAJ7NWIjuNq8AHgJXzDXY7FoA3Zmjsat27EgB8qO5bHUbMABOzPZocMLsfsq24HbSME4aYACuALZwAKbBAD7wAJYA5nlgUO6IXig+BlEBgfH2DGERCpz4QA

💻 Code

const b = -(1n);
//    ^?
const b2 = -b;
//    ^?

const n = -(1);
//    ^?
const n2 = -n;
//    ^?

const a: any = b;
//    ^?
const x = -a;
//    ^?

const conv = (v: any) => -v;
//    ^?
const r = conv(b);
//    ^?
const conv2 = (v: number | bigint) => -v;
//    ^?
const r2 = conv2(b);
//    ^?

🙁 Actual behavior

Applying an unary minus on an any variable infers the type to number (as can be seen in the "hat-question mark" comments or the .D.TS tab for x, conv, and r). The inferred type for the constants x and r is thus wrong (they actually contain a BigInt value).

Unary minus on a number correctly infers number, and (more importantly) unary minus on a bigint correctly infers bigint. Same for unary minus on a number | bigint which infers number | bigint.

🙂 Expected behavior

Applying an unary minus on an any variable should infer the type to number | bigint.

ECMAScript specifies (https://tc39.es/ecma262/multipage/ecmascript-language-expressions.html#sec-unary-minus-operator) unary minus as doing a ToNumeric (which returns either a Number of a BigInt) and then applying the appropriate unaryMinus abstract operation depending on the type of the value. The result is thus either a Number or a BigInt.

(the unary plus operator however converts to Number values only)

Additional information about the issue

No response

Inizia dal TypeScript Playground collegato e riproduci i tipi dedotti per il meno unario su any, number, bigint e number | bigint. Traccia la gestione da parte del compilatore dell'inferenza dei tipi del meno unario e i relativi test; il lavoro è completo quando il meno unario su any viene dedotto come number | bigint, mentre il comportamento esistente per number e bigint rimane corretto.