Arguments following spread

[barneycarroll]

Thanks for the fantastic work on this. My terse method declarations and bind operations read sensible now. It's a huge boon!

I've spotted the following problems, not sure if it's an issue with themes or the parser - Carthage doesn't suffer (although I find almost everything except for operators and functions impossible to distinguish in Carthage anyway). All these screens were taken in Monokai, but the same problem is visible with Excelsior & Villa-Lobos.

Any arguments following a spread don't play well, see line 125:

[bathos]

Thanks. I tried to nail this quick the other day since I only had a minute, and accidentally did params instead of args. Revising now.

[barneycarroll]

Thanks! Incidentally, this reminded me how irritated I am that a leading rest operation is illegal in the first place. Don't suppose you've got any insight on why this is the case?

[bathos]

I bet there are ES Discuss records about that. Perhaps it would create some hidden complexity. Superficially, permitting a "rightward" mapping is not ambiguous:

function poop(...rest, last) {}
function poop(...rest, penultimate, last) {}

The first thing I can think of is that it does require some new grammar, because this, along with any other parameter declaration with two rests, is nonsensical:

function poop(...wat, ...no) {}

On the other hand, this is actually unambiguous:

function poop(first, ...middle, last) {}

So they could have written the grammar such that any given parameter declaration would contain zero or one rest parameters. I don't think this would have required any hoop jumping, actually -- no cover grammars or anything like that -- so I doubt grammatical complexity was the issue.

So perhaps it has something to do with how function parameters are implemented internally -- maybe it would be somehow disruptive that, if non-terminal rests were permitted, _non-_rest parameters would no longer have specific indices. But I'm just guessing here; for all I know, nobody brought it up? I'm pretty clueless about the internal mechanics for outfitting functions.

[bathos]

Other thoughts: In your own code, you can implement a simple util for supporting right-to-left rest-params. (More complex forms wouldn't be so easy to represent, though):

const reversedRest = fn => {
  const ln = fn.length;

  return (...args) => {
    const rest  = args.slice(0, -ln);
    const named = args.reverse().slice(0, ln);
    return fn(...named, ...rest);
  };
}

const example = reversedRest((tail, ...rest) =>
  `the tail is "${ tail }"; there are ${ rest.length } members prior (${ rest })`
);

console.log(example('a', 'b', 'c', 'd'));

On Babel REPL