paulmillr / es6-shim Goto Github PK

should be Infinity

I think that worrying about index.html for tests is a problem. It forces, in particular, relying on a global "expect" to be set (that test_helpers sets for command-line). It makes testling tests fail, and I can only assume jepso tests as well.

I think that it would be better to eliminate the HTML entirely, or at least, use browserify to output all the JS for the html file (and store it in the repo if necessary). That way, the tests could be tweaked to only work in a node environment, but the html file would still be usable if someone wanted it (although I'm not sure why they would).

Thoughts?

According to the EcmaScript6 draft specification (as well as examples on various random websites), Map.prototype.keys is a function that takes no arguments and returns an array; when using this "shim" keys is instead a field of a map that directly stores the array. The result means that coding against this shim will not actually be compatible with an official implementation of Map.

I'm interested in es6, but I don't know how to use es6-shim.

Is there any example(simple one is enough)?

The current implementation is O(N) for most Map/Set operations.

=> should be >=, I believe. (This is not the fat-arrow operator already, right?)

SameValueZero should be used as comparator according to latest spec, but is it worth to follow?

Object.defineProperty isn't shimmable in ES3, which makes the entire ES6 shim only usable in an ES5 environment.

Any shimmable parts of ES6 that do not rely on unshimmable parts of ES5 should use ES3-compatible syntax, to maximize its browser compatibility. If there are shimmable parts of ES6 that require a true ES5 environment, ideally those would be in a separate component/file.

Chrome has probably O(log(n)) access time etc.

Browsing the code of test/index.html

saw these links -
script src="../node_modules/mocha/mocha.js"
script src="../node_modules/chai/chai.js"
script src="../node_modules/es5-shim/es5-shim.min.js"
script src="../es6-shim.js"

but Node_modules are not there in main source folder ? Not sure if this is a problem ?

MDN examples etc

seems,
denormalized number are not supported on iPad, so
and Number.MIN_VALUE is buggy ( === 0)

because of this, many tests from #85 fails on iPad

according to ES6 (not ES5!) spec, http://people.mozilla.org/~jorendorff/es6-draft.html#sec-15.7.3.3

In the IEEE-764 double precision binary representation, the smallest possible value is a denormalized number. If an implementation does not support denormalized values, the value of Number.MIN_VALUE must be the smallest non-zero positive value that can actually be represented by the implementation.

so, can you shim Number.MIN_VALUE ?
seems, not, because it is not configurable...

but other value should be used for tests:

var MIN_VALUE = Number.MIN_VALUE === 0 ? 2.2250738585072014e-308 : Number.MIN_VALUE; // smallest normalized value

Math.log10(1e15) is 14.999999999999996 in Chrome 19

This should be adapted & rewritten for es6-shim.

Also it doesn't seem to test weak map gc properly. I don't know how to do this automatically.

Missing :- Assert: codePoints is a well-formed rest parameter object.
http://people.mozilla.org/~jorendorff/es6-draft.html#sec-15.5.3.3

according to spec
http://people.mozilla.org/~jorendorff/es6-draft.html

?

Would be really nice to have lodash-like custom builds, I guess. Some people need only 1-3 functions from here.

They seem to be missing.

• Object.assign
• Object.mixin
• Array.from mapfn arg
• Array#find
• Array#findIndex
• String#normalize
• String#raw
• String.fromCodePoint tests
• String#codePointAt tests
• Array#keys, Array#values, Array#entries

The following assertions for Map fail in Firefox:

expect(map.keys).to.be.an.instanceof(Function);
expect(map.values).to.be.an.instanceof(Function);

map.keys/map.values appear to be undefined

The following assertion for Set fails in Firefox:

expect(set.size).to.equal(2);

set.size appears to be a function rather than a number.

P.S. this is the cause of the red cross in #32

According to testling, es6-shim isn't working on IE 9. The es5-shim works there, so in theory we should work on IE 9, too.

https://mail.mozilla.org/pipermail/es-discuss/2012-May/022838.html

From the - http://people.mozilla.org/~jorendorff/es6-draft.html#sec-15.5.4.21

NOTE 2 The repeat function is intentionally generic; it does not require that its this value be a String object. Therefore, it can be transferred to other kinds of objects for use as a method.

So if i do

var k = 2;
k.repeat(3) , is this should result in 222 ?

Shim doesn't work on my Internet Explorer 8 - 8.0.6001.18702IS on Windows XP.
The error message translated into English would be "Object doesn't have such property or method".
Looks like Object.keys is undefined here.

>>Object.keys
undefined

I'm using the latest version, directly from master.

Let me know if you need more information. I'll be happy to provide it.

test cases:

Math.expm1(-38) should be -1,
current result is -183115516321770.7

Math.expm1(709) should be near 8.218407461554972e+307,
current result is 8.484265749165578e+76

current:

    startsWith: function(substring) {
      return this.indexOf(substring) === 0;
    },

    endsWith: function(substring) {
      var substr = String(substring);
      var index = this.lastIndexOf(substr)
      return index >= 0 && index === this.length - substr.length;
    },

suggestion:

    startsWith: function(substring) {
      return this.lastIndexOf(substring, 0) === 0;
    },

    endsWith: function(substring) {
      substring = String(substring);
      var d = this.length - substring.length;
      return d >= 0 && this.indexOf(substring, d) === d;
    },

• String.prototype.reverse
• Array.prototype.contains

https://mail.mozilla.org/pipermail/es-discuss/2012-March/021919.html

The Node.js implementation in harmony mode of Map and Set offers no way to iterate over the items (at current stage, 0.8.15) so if I accidentally use it with the es6-shim enabled it will break if I run the same code in --harmony mode.

This difference just made my life more difficult, I think it would help other devs too if it was made sure to not occur.

http://wiki.ecmascript.org/doku.php?id=harmony:more_math_functions

Proposed are: Math.log10, log2, log1p, expm1, cosh, sinh, tanh, acosh, asinh, atanh, hypot, trunc; perhaps gamma and erf; and perhaps randomInt(n).

Despite loading the ES6 shim, FF's Map prototype is not aligned with the shim's Map prototype--at the very least, it's missing forEach.

es6 shim's defineProperties method looks correct to me, but I have not had time to properly debug it or even determine if that is the actual problem.

Is it possible to provide a minified version as well, that way I can just bower to install the minified version.

The implementation given is actually a casual version of Array.from and the tests you wrote are also incorrectly testing Array.of as though it were Array.from

Example of correct behaviour:

Array.of( 1, 2, 3 );
//  [ 1, 2, 3 ];

There are tests that assert:

Array.of( arguments ) === [];

... which is incorrect. That would be correct for Array.from( arguments ); Array.of( arguments ) should actually return [ arguments ];

require.js is looking for simple calls to define statically, it can't assume any method call to define on an arbitrary object is calling its define.

just FYI, thanks for the shim!

Missing :- substitutions is a well-formed rest parameter object.

http://people.mozilla.org/~jorendorff/es6-draft.html#sec-15.5.3.4

NOTE Each key is visited only once with the value that is current at the time of the visit. If the value associated with a
key is modified after it has been visited, it is not re-visited. Keys that are deleted after the call to forEach begins and
before being visited are not visited. New keys added, after the call to forEach begins are visited.

Would you take a patch to add imperfectly shimmable features (a la WeakMap) in an es6-sham.js module? This is how Kris Kowal's refactored es5-shim and it makes a lot of sense. I understand why you don't want it in es6-shim proper but WeakMap is very shimmable regardless.