no-misused-generics: false negative

[OliverJAsh]

// Errors as expected
type MyFunction = <T>(param: T) => number;

// Expected error, got none
const myFunction = <T>(param: T) => 1;

[cartant]

If you have TypeScript configured to parse JSX, the RHS of that variable declaration will be parsed as a JsxElement and will be ignored by the rule. Try:

const myFunction = <T,>(param: T) => 1;
//                   ^

[OliverJAsh]

Unfortunately that doesn't seem to help. Reduced test case: https://github.com/OliverJAsh/eslint-no-misused-generics-test

[cartant]

Not going to fix this 'cause I don't have the time to figure out why it's doing what it's doing:

eslint-plugin-etc/tests/rules/no-misused-generics.ts

Lines 105 to 117 in 4152f18

	// This test should fail, but it doesn't because the implementation tests
	// the signature with:
	//
	// tsutils.isFunctionWithBody(signature)
	//
	// And if that returns true, the implementation assumes that the type
	// parameter is used.
	//
	// fromFixture(stripIndent`
	// // https://github.com/cartant/eslint-plugin-etc/issues/30
	// const func = <T,>(param: T) => null;
	// ~ [canReplace { "name": "T", "replacement": "unknown" }]
	// `),

I didn't write this rule; I just ported it. And, TBH, I'm beginning to think that it's more trouble than it's worth. It's too simplistic and won't work with valid, but context-sensitive generic usage. I'm inclined to delete it and give the package a major version bump.

[cartant]

Thinking about it some more, my guess is that the original author made the (correct) decision that it's too difficult to determine whether what's returned relates to the parameter with the type parameter. E.g. this would be okay:

const myFunction = <T>(param: T) => param;

[OliverJAsh]

You're totally right. I actually filed the same issue back when we were using Wotan 🤦. Sorry! fimbullinter/wotan#690