Open Raynos opened 4 years ago
I think the thing you're encountering has more to do with the fact that we don't look at the contextual types during inference for any invocation expression except for ordinary calls. For example, this works:
class MyServer {
constructor () {
/** @type {Map<string, string[]>} */
this.functions = foo()
}
}
/**
* @template T
* @return {Map<T, T[]>}
*/
function foo() {
throw ""
}
Though it really depends on the problem we're trying to solve. In this case, I think this would be fixed by broadening our inference process, but it's definitely the case that you can't explicitly pass type arguments in JS files.
I think this would be fixed by broadening our inference process
I think that would be a good solution for new Map()
and also new Array()
( Although for new Array()
you can use []
and the inference does work ).
I understand that inference is not perfect in TypeScript and that being explicit is always an option when inference does not work.
you can't explicitly pass type arguments in JS files.
Unfortunately the explicit syntax does not exist for JS files when it comes to type arguments.
I can't think of anything other then
class MyServer {
constructor () {
/** @type {Map<string, string[]>} */
this.functions = new Map/** @generic {<string, string[]>} */()
}
}
This a little strange, but I would prefer strange and possible over impossible.
TypeScript Version: 3.7.x-dev.201xxxxx
Search Terms: JSDoc Map generics assignment any unsafe
Code
Expected behavior:
Expected to be able to say what the Map contains when creating a new map. In typescript you can do
new Map<string, string[]>()
but theres no explicit generic parameter function invocation syntax for JSDocActual behavior:
The type
Map<any, any>
exists. This fails the eslint rule ofno-unsafe-assignment
and this fails theno-any
rule of tslint etc.Note that doing something similar for arrays works because there's a special case inference for
[]
and it also works forSet
because the constructor is defined differently and it has generic inference in the initialization statement.Playground Link: https://www.typescriptlang.org/play?useJavaScript=true#code/MYGwhgzhAECyCeBlApgJwG5ugbwFDWmAHsA7CAF1QFdhyjVoAKASh3wOgHoAqb6AAXLwADshywwwgDwVUASxIBzADTRZCxQG0AugD4AvtG6d2BcgAs5EAHQAzKiVpzSMALzQSyAO5xJLdvq4gbjEZOTQALZIaJgM7p4+CCgYaP5RybF2Dk4u1orI5IwA5LZEREXMuCHgUHDRKQx4BKGyNHQMLGwcXLwCQqI4AIKoqGDwMpQaBkYm3RZWWY7kzmTQ7joBQVUt4ekxWPHedRmplXsNizlkeQXFpeXMQA
Related Issues: Yes, in eslint https://github.com/typescript-eslint/typescript-eslint/issues/2109