Rule Enhancement Proposal: func-names with never

[mysticatea]

EDIT: this proposal was updated in #6059 (comment)

---

From requireAnonymousFunctions and requireMatchingFunctionName.

This proposal is to add some options.

{
    "func-names": [
        "error",
        "always" or "never",
        {
            "sameAsVariable": false
        }
}

• "always" (default) - This is the current behavior.
• "never" - This will disallow a use of function expression names.
• sameAsVariable (default is false) - If this is true, the function expression name must be the same as the variable name which is the destination of the assignment. For example: let foo = function foo() {};. But if the destination is module.exports, this rule will ignore it.

[platinumazure]

Should sameAsVariable: true also check for assignment to a MemberExpression at all? (e.g., foo.bar = function baz () { })

Should we allow a "tolerate" value for the first option, so that a user can decide that s/he does not always want function names, but when a function name is specified, it must match the assigned variable name?

[mysticatea]

Should sameAsVariable: true also check for assignment to a MemberExpression at all? (e.g., foo.bar = function baz () { })

Yes.

Should we allow a "tolerate" value for the first option, so that a user can decide that s/he does not always want function names, but when a function name is specified, it must match the assigned variable name?

Ah, interesting.
Maybe requireMatchingFunctionName should not merge with func-names rule.

[platinumazure]

Should sameAsVariable: true also check for assignment to a MemberExpression at all? (e.g., foo.bar = function baz () { })

Yes.

So if sameAsVariable: true, does this mean users cannot assign functions to object members at all?

Ah, interesting.
Maybe requireMatchingFunctionName should not merge with func-names rule.

I think I agree-- that could be a separate rule, which wouldn't do much if function expressions don't have names, but it would execute safely and not confuse the options schema of this proposed rule.

[mysticatea]

From requireAnonymousFunctions.

This proposal is to add "never" option to func-names rule.

{
    "func-names": [
        "error",
        "always" or "never"
    ]
}

• "always" (default) - This is the current behavior.
• "never" - This will disallow a use of function expression names.

Valid:

/*eslint func-names: ["error", "never"]*/

function foo() {
}
var foo = function() {
};

Invalid:

/*eslint func-names: ["error", "never"]*/

var foo = function foo() {
};

[Mouvedia]

related: #938

[alberto]

@eslint/eslint-team opinions?

I can champion this one

[mikesherov]

👍

[Mouvedia]

👍

Here's another reason—if you care about cross-browser issues—to make this happen: https://kangax.github.io/nfe/#safari-bug

Would it be asking for too much to have a only if not part of an assignment expression option for never?

[nzakas]

Looks like we are good to move forward here.

@Mouvedia this issue is for creating an option that is compatible with JSCS functionality. If you want more than that, please open a new issue.

[Mouvedia]

@nzakas after 3 refusals, I don't think Ill submit a new one…
I learned my lesson :)