Basic support for JSDoc @param {function}

[angelozerr]

This PR provides a basic support for JSDoc {function} :

/**
 * @param {function} fn
**/
function test(fn) {
}

It should be improved to :

• support args of function :

/**
 * @param {function(number)} fn
**/
function test(fn) {
}

• support this of function :

/**
 * @param {function(this:number)} fn
**/
function test(fn) {
  this. // here Ctrl+Spaces shows methods of number.
}

In other words it should be cool to have the same features than JetBrains. See http://blog.jetbrains.com/webide/2012/10/validating-javascript-code-with-jsdoc-types-annotations/

[marijnh]

Is this specified somewhere? I can't find anything about it on in the JSDoc docs.

[angelozerr]

@marijnh you are right, it seems that JSDoc doens't support that. @callback is used for that. See :

• Ability to document callback arguments jsdoc/jsdoc#260
• http://usejsdoc.org/tags-callback.html

So it should be cool if doc_comment could support @callback.

Perhaps it could be interesting to gives the capability to extends doc_comment to support a custom syntax?

[marijnh]

Wow, that's a strange community. They seem entirely unable to deal with recursive syntax, or to grasp the fact that typedefs could already play the part of defining function types, or to notice that callback isn't a proper name for all function types.

So, um, when something is finally decided on, I guess we could support it. I'd also be okay with supporting the much saner Google Closure syntax. But I'd want an actually useful implementation that also parses parameter and return types -- just saying that something is a function can already be done by using Function as its type, I'm not sure this patch adds anything.

[angelozerr]

Yes you are right we can use @param {Function} instead of using @param {function}. My patch give sthe capability to useFunctionorfunction`.

But my orignal problem is for my tern-lint. I have a validator which validates if a token is a function. To do that I do type.getFunctionType(). In the case when type is a Function.prototype, it returns null. I don't know if it's a bug?

[marijnh]

I didn't know Function.prototype was a function (the only function, apparently, whose prototype is Object.prototype). Attached patch fixes this. Does that solve your problem?

[paulvi]

👍 sign up

[paulvi]

please comment why just closing in #561 and #438

[marijnh]

There was no further response for five months, so I assumed the discussion was over.