satyr / coco Goto Github PK

Reasons

• Scope mixups:

  a = 1
  say do     (a = a) -> a  # 1
  say do []= (a = a) -> a  # undefined
  

• do being a simple sugar for () is consistent and easier to grok.

Syntax

let (a = b) -> # (function(a){})(b)
let (a) ->     # (function(a){})(a)
let (@a) ->    # (function(a){})(this.a)
let ~>         # (function(){}).call(this)

λ = ->

return that if get()

↓
var that;
if (that = get()) {
return that;
}

function f1() ->
f2 = function () ->

↓
var f2;
function f1(){}
f2 = (function(){
function f2(){}
return f2;
}());

a[4,2]

↓
[a[4], a[2]];

f ...

↓
f.apply(this, arguments);

a: 1
a: 2

↓

Before

({
  a: 1,
  a: 2
});

After

SyntaxError: duplicated property name "a"

switch task
case prepare
  do sing
  fallthrough  # prevents auto-break
default
  do dance

↓
switch (task) {
case prepare:
sing();
default:
dance();
}

• They are kind of verbose, and don't read well when combined with other symbols (such as @:: import ...).
• Can act as replacements after harmony defines import for real.

<<< for import, <<<< for import all?

Should be able to optimize if the right operand is a bare array/object.
C:: import d: e, (f): g
↓
var _ref;
_ref = C.prototype, _ref.d = e, _ref[f] = g;

edit: On second thought, array case is pretty pointless.

As you can now do the same without them using groov-y dots.

ok = if cond
  expr()
  true
else
  false

↓

Before

var ok;
ok = (function(){
  if (cond) {
    expr();
    return true;
  } else {
    return false;
  }
}());

After

var ok;
ok = cond ? (expr(), true) : false;

Currently, do (x = y) -> ... compiles to

(function(x){
  x == null && (x = y);
  ...
}());

which is really just an inefficient way of writing

(function(x){
  ...
}(y));

Changing do to rewrite in this way would also allow you to write do (x = x), which currently is nonsensical, to capture a particular value of x (useful in loops).

I would also suggest that do (x) be made a shorthand for do (x = x).

cf. https://gist.github.com/556213

a.b   ||= {}
a.b.c ||= []
a.b.c.push e

↓
a%[email protected] e # ((a.b ||= {})c ||= [])push e

Useful, or ugly?

a.b&.c

↓

__bind((_ref = a.b).c, _ref)

ref. jashkenas/coffeescript#856

\word
\+

↓
'word';
'+';

class C
  (@attr) =>

↓

var C;
C = (function(){
  function C(_arg){
    var _this = new _ctor;
    _this.attr = _arg;
    return _this;
  }
  C.name = "C";
  function _ctor(){}
  _ctor.prototype = C.prototype;
  return C;
}());

(a = 0, b ?= 1, c ||= 2) -> # where "a" is a sugar for "b"

↓
(a = 0, b ? 1, c || 2) -> # "a" stays for compatibility
The destructuring defaults would then be less confusing:
[x = 0, y or= 1] = z
↓
[x ? 0, y or 1] = z

Any thoughts of integrating this in coco?

jashkenas/coffeescript#835

Install instructions state:
git clone git:github.com/satyr/coco.git && cd coco && bin/coke i

But should be:
git clone git://github.com/satyr/coco.git && cd coco && bin/coke i

The property syntax is already special in allowing an indented expression as RHS:
$ coco -bpe '
> k:
> 1
> '
({
k: 1
});
But not block:
$ coco -bpe '
> k:
> 1
> 2
> '
Error: Parse error on line 3: Unexpected 'TERMINATOR'
Allowing the latter as an implicit array literal seems reasonable.

code &&+= '\n'

↓
code && (code += '\n');

Would it a terrible idea to allow other languages within coco?

Here's a simple example of what I am thinking:

paragraph = 
    :markdown
    Hello, 
    ======
    Today is a *fine* day.

$('#welcome-message').inject paragraph

The :markdown is just an example of how one might signify the beginning of a new language block. I guess coco would need to have a 'hook' for authors to plugin/reference other language compilers. I imagine it wouldn't be too hard to manage that part. The issue may be more the usefulness of such a feature, and enough use-cases to make it worthwhile.

obj[\x, \y] = a
obj{x, y}   = o

↓
obj['x'] = a[0], obj['y'] = a[1];
obj.x = o.x, obj.y = o.y;

Currently the import or <<< feature is handy but a version that allows for only undefined values to be imported would help I think.

Below is a common use-case:

human = (args)->
    defaults =
        height: 6
        hair: 'brown'

    args <<< defaults
    console.log args, defaults


human do
    country: 'canada'
    job: 'mail man'
    height: 3

In this example, defaults.height overrides args.height.

Maybe the syntax could be something like:

<<<?
<<<<?

I am not sold on those at all, just potentials.
Maybe we could change <<< to behave non-destructively by default. Then we could provide bang syntax to force overrides, like such:

<<<!
<<<<!

Expressions like []() should be syntax errors at compile time.

ref. jashkenas/coffeescript#1069

The current workaround (L: → break L;) is too awkward.

ref. jashkenas/coffeescript#640

Apparently there is a problem with package.json. According to isaacs the bin hash should be filenames not module ids.

~ $ npm install coco --loglevel verbose npm info it worked if it ends with ok
npm verb cli [ 'install',
npm verb cli 'coco',
npm verb cli '--loglevel',
npm verb cli 'verbose' ]
npm info using [email protected]
npm info using [email protected]
npm verb config file /Users/stuarts/.npmrc
npm verb config file /Users/stuarts/local/etc/npmrc
npm verb install [ 'coco' ]
npm verb install [ 'coco' ]
npm verb install pkglist [ 'coco' ]
npm verb must install [ 'coco' ]
npm verb install coco
npm verb must install? [ [ 'coco' ], true ]
npm verb GET coco
npm verb etag "2V412V9E96NYDOLEOXV5L3HMB"
npm verb response http://registry.npmjs.org/coco npm verb etag coco from cache
npm verb Supported versions [ '[email protected]',
npm verb Supported versions '[email protected]',
npm verb Supported versions '[email protected]',
npm verb Supported versions '[email protected]',
npm verb Supported versions '[email protected]',
npm verb Supported versions '[email protected]',
npm verb Supported versions '[email protected]',
npm verb Supported versions '[email protected]',
npm verb Supported versions '[email protected]',
npm verb Supported versions '[email protected]' ]
npm verb coco tag latest
npm verb install [email protected]
npm verb must install? [ [ 'coco', '0.3.2' ],
npm verb must install? true ]
npm verb [email protected] exact
npm verb caching /Users/stuarts/local/lib/node/.npm/.cache/coco/0.3.2/package.json
npm verb install list [ [ 'coco', '0.3.2' ] ]
npm verb mkdir done: /Users/stuarts/local/lib/node/.npm/coco/0.3.2
npm verb unpackTar uid, gid [ undefined, undefined ]
npm verb success gzip "--decompress" "--stdout" "/Users/stuarts/local/lib/node/.npm/.cache/coco/0.3.2/package.tgz"
npm verb success tar "-mvxpf" "-" "-C" "/Users/stuarts/local/lib/node/.npm/coco/0.3.2"
npm verb unpacked, building coco,0.3.2
npm verb caching /Users/stuarts/local/lib/node/.npm/coco/0.3.2/package/package.json
npm info preinstall [email protected]
npm verb deps being resolved [ { name: 'coco',
npm verb deps being resolved description: 'Unfancy CoffeeScript',
npm verb deps being resolved keywords:
npm verb deps being resolved [ 'language',
npm verb deps being resolved 'compiler',
npm verb deps being resolved 'coffeescript',
npm verb deps being resolved 'javascript' ],
npm verb deps being resolved author:
npm verb deps being resolved { name: 'satyr',
npm verb deps being resolved email: '[email protected]',
npm verb deps being resolved url: 'http://satyr.github.com' },
npm verb deps being resolved version: '0.3.2',
npm verb deps being resolved homepage: 'http://satyr.github.com/coco/',
npm verb deps being resolved repository:
npm verb deps being resolved { type: 'git',
npm verb deps being resolved url: 'git://github.com/satyr/coco.git' },
npm verb deps being resolved licenses:
npm verb deps being resolved [ { type: 'MIT',
npm verb deps being resolved url: 'https://github.com/satyr/coco/raw/master/LICENSE' } ],
npm verb deps being resolved engines: { node: '>=0.2.6' },
npm verb deps being resolved directories: { lib: './lib', bin: './bin' },
npm verb deps being resolved main: './lib/coco',
npm verb deps being resolved bin:
npm verb deps being resolved { coco: './lib/command',
npm verb deps being resolved coke: './lib/coke' },
npm verb deps being resolved id: '[email protected]',
npm verb deps being resolved engineSupported: true,
npm verb deps being resolved npmVersion: '0.3.15',
npm verb deps being resolved nodeVersion: 'v0.4.2',
npm verb deps being resolved files: [ '' ],
npm verb deps being resolved defaultsLoaded: true,
npm verb deps being resolved dist:
npm verb deps being resolved { shasum: 'd5aff68b204e5f5f84f7db0298b2af6d2da3370e',
npm verb deps being resolved tarball: 'http://registry.npmjs.org/coco/-/coco-0.3.2.tgz' },
npm verb deps being resolved _bundledDeps: [] },
npm verb deps being resolved [] ]
npm verb linkBins [email protected]
npm verb linkBin coco ./lib/command
npm verb linkBin coke ./lib/coke
npm verb linkMans [email protected]
npm ERR! failed to link bins
npm ERR! install failed Error: ENOENT, No such file or directory '/Users/stuarts/local/bin/[email protected]'
npm info install failed rollback
npm info uninstall [ '[email protected]' ]
npm verb removing [ [ 'coco', '0.3.2' ] ]
npm verb from cache /Users/stuarts/local/lib/node/.npm/coco/0.3.2/package/package.json
npm info preuninstall [email protected]
npm info uninstall [email protected]
npm info auto-deactive not symlink
npm verb uninstall [email protected] remove links
npm verb uninstall [email protected] remove bins
npm verb uninstall [email protected] remove public modules
npm info postuninstall [email protected]
npm verb uninstall [email protected] remove package dir
npm info uninstall [email protected] complete
npm info install failed rolled back
npm ERR! Error: ENOENT, No such file or directory '/Users/stuarts/local/bin/[email protected]'
npm ERR! Report this entire log at http://github.com/isaacs/npm/issues
npm ERR! or email it to [email protected]
npm ERR! Just tweeting a tiny part of the error will not be helpful.
npm ERR! System Darwin 10.6.0
npm ERR! argv { remain: [ 'coco' ],
npm ERR! argv cooked:
npm ERR! argv [ 'install',
npm ERR! argv 'coco',
npm ERR! argv '--loglevel',
npm ERR! argv 'verbose' ],
npm ERR! argv original:
npm ERR! argv [ 'install',
npm ERR! argv 'coco',
npm ERR! argv '--loglevel',
npm ERR! argv 'verbose' ] }
npm verb exit [ 2, true ]
npm not ok

Allow arbitrary trailing alphabets, for specifying units etc.

setTimeout cb, 512ms

day = 24hours * 60seconds

return if get()

↓

var _ref;
if (_ref = get()) return _ref;

ref. https://github.com/fab13n/parsec-coffee-script/blob/master/src/Lexer.coffee#L200

/// \n #{? if multi then 'g' else '' }///

↓
RegExp('\n', multi ? 'g' : '')

When installing coco manually using readme instructions, the following occurs:

node.js:63
    throw e;
    ^s:275:15)
    at require (node.js:411:14)
    at Object.<anonymous> (/Users/KUHRT/coco/bin/coco:20:5)
    at Module._compile (node.js:462:23)
    at Module._loadScriptSync (node.js:469:10)
    at Module.loadSync (node.js:338:12)
    at loadModule (node.js:283:14)
    at require (node.js:411:14)
    at Object.<anonymous> (/Users/KUHRT/coco/bin/coke:11:15)
    at Module._compile (node.js:462:23)

I can overcome this by manually copying the optpase file from the lib directory to the bin directory. I also npm'd optparse but haven't tried loading the module from there (after I realized the problem code was using a local path). I see that you've symlinked the coke file to bin so I am not sure why this is happening.

Am I doing something wrong?

with f()
  @p = q

↓
(function(){
return this.p = q;
}).call(f());

With syntax like:

obj = {
  prop: function -> 42
  prop: function (item) -> alert item
}

↓

var obj;
obj = {
  get prop(){
    return 42;
  },
  set prop(item){
    return alert(item);
  }
};

So that code like
### (c) 2010 me ###
"use strict"
I = -> it
works as expected.

So that this:
$ coco -e 'console.log arguments is (switch case 1 then arguments)'
false
results in true.

Before

console.log(arguments === (function(){
  switch (false) {
  case !1:
    return arguments;
  }
}.apply(this, arguments)));

After

console.log(arguments === (function(_args){
  switch (false) {
  case !1:
    return _args;
  }
}(arguments)));

So that hacks like

`EXIT:`
for i from 3 to 5
  for j from 7 to 9
    `break EXIT` if Math.random() < .2
    say i * j

works as expected.

Introduce 0b101010 and 0o52, then disallow 052 per ES5.

a <- f
x
(b, c) <- g ..., d
y

↑

f (a) ->
  x
  g (b, c) ->
    y
  , d

• The function body will be all the following lines of that Body (as in grammar.co).
• ... can be used as a placeholder for the callback function, as normally it is useless in the presence of another argument.

ref. jashkenas/coffeescript#1032

Since the return value of JS delete is hardly useful and rarely used.
v = delete o.p
delete o.k
↓
var _ref;
v = (_ref = o.k, delete o.k, ref);
delete o.k;

[1 to 3]
[1 to 5 by 2]

↓

[1, 2, 3]
[1, 3, 5]

Should be able to expand within lexer.

Now that you can super ... instead of super (which compiles to the same thing), make it a literal for greater flexibility (such as a.map super.bind this).