Discuss
  • Ignoring arguments

Ignoring arguments

# Cyril Auburtin (8 years ago)

Similarly to:

var [,x,,y] = [1,2,3,4,5,6]; array destructuring, or [a,,b] array creation,

I think it could be interesting to let a field empty in function arguments

[1,2,3,4].map( (,i) => i )
Array.from({length:10}, (,i) => i )
function test(a,,b) { }
var x = test(a,,b)

by altering the current syntax that doesn't allow it

Most often people use _ as a way to mark ignored fields, but when (more rarely) there are more than 1 ignored field, you need another identifier. A standard way would be interesting rather

# Renki Ivanko (8 years ago)

You can already do it like this:

((...[,,foo]) => foo)(1, 2, 3) // -> 3

R. untu.ms

# Bob Myers (8 years ago)

I'm quite sure this syntax is invalid.

# Renki Ivanko (8 years ago)

It works in V8 at least, so I assumed it's correct.

# Caitlin Potter (8 years ago)

You might think so, but you'd be wrong. Perfectly valid formal parameters

# Caitlin Potter (8 years ago)

It is.

tc39.github.io/ecma262/#prod-FunctionRestParameter

tc39.github.io/ecma262/#prod

# Bob Myers (8 years ago)

Eliding array elements is not "similar" to eliding function formal parameters. The latter is extremely rare, hardly readable, confusing, bug-prone, and unnecessary because there is already a "standard way" which is to use any old parameter name you want:

function foo(UNUSED1, UNUSED2, x)

Most linters will not complain, or there are ways to shut them up if they do.

If you want to throw away an argument, just throw it away.

function skipFirstParam(fn) { return ((first, ...args) => fn(...args)); }

[1,2,3,4].map(skipFirstParam(i => i));

Or use Renki's solution.

Bob

On Sun, May 29, 2016 at 9:23 PM, Cyril Auburtin <cyril.auburtin at gmail.com>

wrote:

# Logan Smyth (8 years ago)
function fn(...[a, b, c, ...rest]){}

is mostly no different than

function fn(...args){
    let [a, b, c, ...rest] = args;
}

This wasn't officially allowed in ES2015, where only an Identifier was allowed as a rest param target in arguments, but the behavior has since been expanded to cover this case in ES2016: tc39/ecma262/commit/d322357e6be95bc4bd3e03f5944a736aac55fa50

Not that I think this example is particularly readable :)

# Renki Ivanko (8 years ago)

You could stop with "rare"; having to make up unused names is an obvious smell in comparison.

foo(UNUSED1, UNUSED2, x)

foo(_, __, x)

foo(,, x)

foo(...[,, x])

The latter is shorter and more explicit and would not be any more confusing if it became common.

# Cyril Auburtin (8 years ago)

Since functions arguments is an array under the hood, they could 'more behave the same'

Both function arguments and arrays accept spreading: [1, 2, ...args] and fn(1,2, ...args)

a function definition like (,i) => {}, would be the equivalent of var [,i] = arguments

an invocation fn(,,i) would be the equivalent of [,,i]

It's possible with (...[,i]) => {}, (_,i)=>{} like Renki said, but slightly less simply

Are there possible issues with that 'extension' of function syntax?

# Renki Ivanko (8 years ago)

One more similarity is that both function parameters and destructuring allow default values: (foo = 1) vs [foo = 1].

# Cyril Auburtin (8 years ago)

Just bumping this up

Calling foo(1) where foo is defined with 3 arguments, lets the 2 others undefined, this behavior is already a bit magic and similar to the behavior of an array, so I still think foo(a,,b,,,c) should be like foo(...[a,,b,,,c])

Other example:

var m=new Map([[1], [2,], [3,7]]) // Map {1 => undefined, 2 => undefined, 3 => 7} // here commas fantasies are allowed in arrays
m.set(3) // Map {1 => undefined, 2 => undefined, 3 => undefined} // setting implicitely value as undefined
m.set(3, ) // not allowed, which should be m.set(...[3,]) or m.set(3, undefined)

and again, it would help for callbacks too, something( ( , , thirdArg) => {} )

I saw this jeffmo.github.io/es-trailing-function-commas, it seems like a sub-case

# Isiah Meadows (8 years ago)

For what it's worth, elided array elements aren't available in strict mode (they're a syntax error), and it's not likely anyone is going to want to add a new sloppy mode only feature. Plus, it's not easy to tell at a glance if it's a typo or intentional. Math.min(x,,y) is a likely error, but a syntax error is much more informative than NaN or a very odd undefined. I don't see the point of not making it explicit.

# Claude Pache (8 years ago)

Le 10 août 2016 à 11:01, Isiah Meadows <isiahmeadows at gmail.com> a écrit :

For what it's worth, elided array elements aren't available in strict mode

Are you sure?

# Isiah Meadows (8 years ago)

Okay. I'll admit I'm wrong here. I misremembered. I knew they were more of a legacy thing, I just didn't know they were available in strict mode.

# Alan Johnson (8 years ago)

What if you could use a . as a wildcard? I don’t think it would conflict with existing syntax, and it would avoid binding a usable identifier. It would be more obvious than nothing between the commas.

# Bob Myers (8 years ago)

Confused by this thread.

What if you could use a . as a wildcard?

You can already use _ or anything you want, as was already pointed out in early iterations of this idea.

it would avoid binding a usable identifier.

What is the problem with that?

What is the problem that this proposal is trying to solve? Any chance we could move on?

Bob

# Cyril Auburtin (8 years ago)

Confused by this thread.

It's quite simple though, _ would be a useless allocated variable, you can't have more than 1, ...

What is the problem that this proposal is trying to solve?

ignoring/skipping argument in functions declarations (indeed the invocations could be let apart)

Math.min(x,,y)

yes this kind of thing would give NaN, that's why you can write it with more spaces to avoid a typo mistake Math.min(x, , y), but indeed you could still have something like Math.min(x, /*y (no longer used)*/ , z). Like said above, I realize the function invocations are sensible.

So, it could be only for function declarations

div.addEventListener('click', () => { } ) says to skip all arguments, and your callback scope is free of any additional vars

arr.forEach( x => { } ) or arr.forEach( (x) => { } ) says to only consider first argument, others are skipped, and not in scope

Array.from({length: 19}, (, i) => i ) would be similar for the second argument, similarly to array destructuring. It's not just for saving one character, that's really not the matter, it's for standardizing this way, because some people use _, some don't, .... (Yes . would even be better than a variable name like _, something similar to Haskell's 'holes')

# Kris Siegel (8 years ago)

ignoring/skipping argument in functions declarations (indeed the invocations could be let apart)

I am of the opinion that this is not a problem that needs to be solved at the language standardization level. The commas (or really any separator) for "skipping" a parameter looks like a mistake and is unintuitive (it's certainly not obvious to me what its intent is had I seen it without the context of this thread).

This seems like a problem solved with better coding practices. I typically make required parameters first and then any optional ones in a separate, single parameter called options. Callbacks can be done with chaining / promises / I fmake them required / etc. Skipping callbacks usually isn't a good idea. We could talk all day about who's coding practice is better than who's but ultimately this problem should be solved by developing / adopting good coding practices.

(Honestly I'm not a fan of the skipping in arrays either but considering it's an array I feel like it's at least a magnitude less confusing than doing it with function parameters)

# Isiah Meadows (8 years ago)

If you're talking about ignored function parameters in a declaration or expression, much like _ in several functional languages (like Haskell and Scala), I could see value in that. I'd prefer an explicit token for that, though, something like this:

// ImmutableJS: convert map to list, only
// keeping values with integer keys
map
.filter((*, key) => typeof key === "number" && key % 1 === 0)
.toList()
# J Decker (8 years ago)

On Thu, Aug 11, 2016 at 1:28 PM, Isiah Meadows <isiahmeadows at gmail.com>

wrote:

If you're talking about ignored function parameters in a declaration or expression, much like _ in several functional languages (like Haskell and Scala), I could see value in that. I'd prefer an explicit token for that, though, something like this:

'undefined' is what, too many characters to type?

( ( ignore_me, ignore_me2, asdf ) => { /* ... */ } ) (undefined, undefined, {whatever:1} )

// ImmutableJS: convert map to list, only

# Michał Wadas (8 years ago)

If you want to have new syntax for this feature, keyword void can be easily reused:

((void, void, void, param)=>param)(1,2,3,4) // 4

But I don't think it's worth it.

# Cyril Auburtin (8 years ago)

( ( ignore_me, ignore_me2, asdf ) => { /* ... */ } ) (undefined, undefined, {whatever:1} )

Let's forget about functions invocations, a bad idea, so that's more the ignore_me, ignore_me2 that are quite polluting/unpractical

I'm not either a fan of skipping in arrays, but again function arguments is an array, that same syntax would represent literally holes. There's also like for invocations the risks of a typo, but the coder has to be really distracted to not see it..

Thanks for that discussion, it'll very likely never make it to the spec, and sorry my last message had typos, corrected on esdiscuss.org topic/ignoring-arguments, mailing-lists :/

# J Decker (8 years ago)

On Thu, Aug 11, 2016 at 5:55 PM, Cyril Auburtin <cyril.auburtin at gmail.com>

wrote:

( ( ignore_me, ignore_me2, asdf ) => { /* ... */ } ) (undefined, undefined, {whatever:1} )

Let's forget about functions invocations, a bad idea, so that's more the ignore_me, ignore_me2 that are quite polluting/unpractical

I'm not either a fan of skipping in arrays, but again function arguments is an array, that same syntax represent literally holes. there's also like for invocations the risks of a typo, but the coder has to be really distracted to not see it..

Thanks for that discussion, it'll very likely never make it to the spec, and sorry my last message had typos, corrected on esdiscuss.org/topic/ignoring-arguments, mailing-lists :/

but the point of that really is... why can't you just use the token 'undefined' instead of trying to use whitespace?

# Cyril Auburtin (8 years ago)

If we are talking about function definitions, It doesn't do exactly what a hole is supposed to do

[1,2,3,4].map((undefined, i)=>i) // undefined act as a variable
[1,2,3,4].map((undefined, undefined, i)=>i) // so no

2016-08-12 4:04 GMT+02:00 J Decker <d3ck0r at gmail.com>:

# Cyril Auburtin (8 years ago)

off-topic: I received this nice mail by Bob, this person was the only one to not understand the point of this discussion, why does someone act like that?

ps: He's also wrong in that function(_, _, arg) is allowed, it's not in strict mode, not for arrow functions either

Thanks for that discussion, it'll very likely never make it to the spec,

Most people would read the above as signing off/out of the discussion. Would that it had been so.

There is an English expression that comes to mind here: BEATING A DEAD HORSE.

No one cares about the proposal. It is somewhere between irrelevant, and actively bad--primarily by harming readability. You act as if writing out an unused parameter is somehow a wanton act of waste that is eating into the world's limited supply of available variables.

Minor point, but it is syntactically valid to write function(_, _, arg), so you don't have to "use up another variable binding" with the second unused parameter.

In case you didn't notice, the proposal is not getting any traction and will never get any traction. Move on.

, Bob

# Bob Myers (8 years ago)

Can the moderator please delete this post, since it was private mail never intended for publication. Also, please look at blocking options. Bob