microlinkhq / async-memoize-one

98%

push

github

fix(types): return what the memoized function actually is (#26)

* fix(types): return what the memoized function actually is

`memoizeOne` returns a new async arrow function: properties on `fn` are
not copied, `fn` is called without `this`, and the result is always a
Promise. The old `(fn: T) => T` signature claimed otherwise.

The declaration now returns `(...args) => Promise<Awaited<Result>>`,
requires `this: void`, and types `isEqual` arguments as `fn`'s
parameters. Generic functions keep their type parameters, and a second
overload keeps `memoizeOne<FnType>(...)` compiling.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* test(types): lock comparator arguments and nested unwrapping

Widening isEqual's arguments, the explicit overload's parameters, or
replacing Awaited with a single unwrap left the type test green.

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(types): lock the comparator's previous arguments

Widening only oldArgs back to any[] left the type test green.

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(types): lock required parameters and boolean comparators

Making the wrapper's parameters optional, or widening the comparator
return to any, left the type test green.

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(types): lock options, arity, and empty returns

Banning false, an omitted option, a null comparator, extra arguments,
or Promise<void> left the type test green. The promise constraint no
longer uses any.

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(types): lock option values and comparator returns

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(types): lock the promise constraint and other returns

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(types): lock explicit undefined on cachePromiseRejection

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(types): lock exactOptionalPropertyTypes

Co-authored-by: Cursor <cursoragent@cursor.com>

* test(typ... (continued)

8 of 9 branches covered (88.89%)

Branch coverage included in aggregate %.

32 of 32 relevant lines covered (100.0%)

4.53 hits per line