experimental commit: derive names for function expressions when we can · michaelficarra/CoffeeScriptRedux@45c5d56

devinrhode2 · 2012-12-31T16:46:58Z

Exactly what sort of js will this output?

vendethiel · 2012-12-31T17:01:55Z

Click on the "Diff suppressed. Click to show." ;).

devinrhode2 · 2012-12-31T17:41:40Z

This will indeed cloud things up with every function a having a reflective a$ function being defined (and a !== $a !!) in oldIE

@kangax has the ultimate know how when it comes to this. Could you provide a recommendation?

devinrhode2 · 2012-12-31T17:43:38Z

To prevent clouding the namespace, prevent memory leaks, etc, this style should probably be used:

var someFn = (function() {
  function someFn() { 
    //vanilla function declaration!
  }
  return someFn;
})();

vendethiel · 2012-12-31T17:44:55Z

See the different comments of @michaelficarra

devinrhode2 · 2012-12-31T18:00:40Z

I suppose var f = function f$(){}; isnt' that bad, I have some js that uses this style same style var a = function aFn(){};

But the style I suggest 2 comments up would probably less error prone, I'm not 100% sure.

One factor is that var f = function f$(){} opens the door for people to recurse in strange scenarios with f$

In many scenarios, a regular function declaration is a clean cross browser way to define a function with a name (jQuery has 55 function declarations) But of course function declarations only work in certain scenarios

michaelficarra · 2012-12-31T18:06:22Z

@devinrhode2: f$ is a fresh name, and is therefore never referenced elsewhere.

devinrhode2 · 2012-12-31T18:13:01Z

True - sounds totally kosher to me, but some experimentation could be done with how IE performs using either style.

…

-Devin http://zerply.com/DevinRhode2 http://zerply.com/devinrhode2

On Mon, Dec 31, 2012 at 12:06 PM, Michael Ficarra ***@***.***: @devinrhode2 https://github.com/devinrhode2: f$ is a fresh name, and is therefore never referenced elsewhere. — Reply to this email directly or view it on GitHubhttps://github.com/michaelficarra/CoffeeScriptRedux/commit/45c5d56bd77394550105fa0047e1a9c099631313#commitcomment-2369701.

devinrhode2 · 2012-12-31T18:21:18Z

However, oldIE treats these as function declarations which brings function hoisting into play.. Not sure if this would mess anything up

On Monday, December 31, 2012, Devin Rhode wrote: True - sounds totally kosher to me, but some experimentation could be done with how IE performs using either style. -Devin http://zerply.com/DevinRhode2 http://zerply.com/devinrhode2 On Mon, Dec 31, 2012 at 12:06 PM, Michael Ficarra < ***@***.*** <javascript:_e({}, 'cvml', ***@***.***');>> wrote: > @devinrhode2 https://github.com/devinrhode2: f$ is a fresh name, and > is therefore never referenced elsewhere. > > — > Reply to this email directly or view it on GitHubhttps://github.com/michaelficarra/CoffeeScriptRedux/commit/45c5d56bd77394550105fa0047e1a9c099631313#commitcomment-2369701.

##

…

-Devin, http://Zerply.com/DevinRhode2 Sent from Gmail Mobile

devinrhode2 · 2012-12-31T18:22:34Z

Consider If thing Fm -> 99 Else Fm -> 98

On Monday, December 31, 2012, Devin Rhode wrote: However, oldIE treats these as function declarations which brings function hoisting into play.. Not sure if this would mess anything up On Monday, December 31, 2012, Devin Rhode wrote: > True - sounds totally kosher to me, but some experimentation could be > done with how IE performs using either style. > > -Devin http://zerply.com/DevinRhode2 > http://zerply.com/devinrhode2 > > On Mon, Dec 31, 2012 at 12:06 PM, Michael Ficarra < > ***@***.***> wrote: > > > @devinrhode2 https://github.com/devinrhode2: f$ is a fresh name, and > > is therefore never referenced elsewhere. > > > > — > > Reply to this email directly or view it on GitHubhttps://github.com/michaelficarra/CoffeeScriptRedux/commit/45c5d56bd77394550105fa0047e1a9c099631313#commitcomment-2369701. ## -Devin, http://Zerply.com/DevinRhode2 Sent from Gmail Mobile

##

…

-Devin, http://Zerply.com/DevinRhode2 Sent from Gmail Mobile

kangax · 2013-01-02T13:33:38Z

Why not just use var inspect = function inspect() { ... }? If assignment is already happening in local scope, the JScript NFE identifier leakage is not a concern.

michaelficarra · 2013-01-02T13:45:54Z

@kangax:

$ node
> var inspect = function inspect(self){ return inspect === self; }
undefined
> inspect(inspect)
true
>

That would be false in IE6, no? If not, that's actually an option. We just have to watch out for this case:

obj.default = -> "what should this function's name be?"

We could always make the name safe by prepending an underscore or something in those cases, though. Fresh names (as in this implementation) also solve that problem.

There's also this problem:

obj = {}
name = "not a function"
obj.name = -> # "name" in here would now refer to the NFE, not the `name` var

So in that case, the function name does need to be fresh.

vendethiel · 2013-01-02T13:58:45Z

I remember microsoft's old ajax framework. They used this style : (iirc)
a.b = function a$b (){}

michaelficarra · 2013-01-02T14:02:23Z

@Nami-Doc: Compile this for me: a[some arbitrary complex expression].b = ->. Also, it still needs to be fresh so it doesn't conflict with the outside user-defined var a$b.

epidemian · 2013-01-02T14:05:19Z

I may be completely mistaken, but i was under the impression that the problem with IE was that the name of the named function expression was hoisted into the enclosing scope, therefore making this code behave differently across browsers:

console.log(foo); // undefined on most browsers; not undefined on IE, maybe, IDK 
var foo = function foo() {};

epidemian · 2013-01-02T14:08:45Z

Using fresh names as this patch does is the way to go IMO. Some kinds of assignable expressions, like foo.bar = -> ... could be special cased to genSym 'foo$bar' as @Nami-Doc suggested, which i think might be reasonable; but meaningful names for things like foo["do#{lolwut}"] = -> ... are probably impossible to find 😅

michaelficarra · 2013-01-02T14:09:52Z

@epidemian: That is true. I remembered it was exposed, I forgot it was also hoisted. What a terrible series of bugs they introduced... Alright, that's completely off the table. All names must be fresh (as they are in this commit). Still, as @Nami-Doc mentioned, #131 is closed because of the FUD around this bug. I don't want it to negatively influence the merge into jashkenas/coffee-script.

edit: You beat me to it.

vendethiel · 2013-01-02T14:30:49Z

Actually, you van still try.
foo["do#{bar}"] = -> giving something likefoo$do$$bar$$ is a solution but probably too complex. Anyway, having these specialcased (unnamed) doesn't seem that bad to me.

devinrhode2 · 2013-01-02T17:36:44Z

For these unpredictable runtime scenarios, foo[key] we could just add in 3 random numbers in place of key

On Wednesday, January 2, 2013, Nami-Doc wrote: Actually, you van still try. foo["do#{bar}"] = -> giving something likefoo$do$$bar$$ is a reasonable solution but probably too complex. Anyway, having these specialcased (unnamed) doesn't seem that bad to me. — Reply to this email directly or view it on GitHubhttps://github.com/michaelficarra/CoffeeScriptRedux/commit/45c5d56bd77394550105fa0047e1a9c099631313#commitcomment-2373344.

##

…

-Devin, http://Zerply.com/DevinRhode2 Sent from Gmail Mobile

vendethiel · 2013-01-02T17:40:36Z

genSym takes care of conflicts

kangax · 2013-01-02T22:49:15Z

@michaelficarra no, afaik the inspect snippet should return true since foo from variable declaration+assignment is shadowing foo from IE's incorrect function declaration. But because of this incorrect parsing of NFE as function declaration it does get hoisted, like @epidemian said. Dunno if this would be a problem here.

devinrhode2 · 2013-01-02T23:33:26Z

I've done some js where I always did var a = function a(){}; consistently everywhere and I had to change it to var a = function aFn(){}; to get it working in ie (only thoroughly tested and verified ie9) I would expect errors from the pattern var a = function a(){};

On Wednesday, January 2, 2013, Juriy Zaytsev wrote: @michaelficarra https://github.com/michaelficarra no, afaik the inspect snippet should return true since foo from variable declaration+assignment is shadowing foo from IE's incorrect function declaration. But because of this incorrect parsing of NFE as function declaration it does get hoisted, like @epidemian https://github.com/epidemian said. Dunno if this would be a problem here. — Reply to this email directly or view it on GitHubhttps://github.com/michaelficarra/CoffeeScriptRedux/commit/45c5d56bd77394550105fa0047e1a9c099631313#commitcomment-2375710.

##

…

-Devin http://zerply.com/DevinRhode2 http://zerply.com/devinrhode2

Constellation · 2013-01-03T05:20:49Z

Because of JScript bug, IE6 leaks function expression name definition to upper scope.

console.log(typeof leak);  // => "function"
var i = function leak() {
};

michaelficarra · 2013-01-03T05:59:22Z

@Constellation: Yes, we know. That's why they're fresh names. Then it doesn't matter that it leaks into the containing scope, since it's guaranteed to never be referenced.

Constellation · 2013-01-03T06:22:44Z

Then it doesn't matter that it leaks into the containing scope, since it's guaranteed to never be referenced.

I see. Thanks!

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Commit

There are no files selected for viewing

25 comments on commit `45c5d56`

devinrhode2 commented on `45c5d56` Dec 31, 2012

vendethiel commented on `45c5d56` Dec 31, 2012

devinrhode2 commented on `45c5d56` Dec 31, 2012

devinrhode2 commented on `45c5d56` Dec 31, 2012

vendethiel commented on `45c5d56` Dec 31, 2012

devinrhode2 commented on `45c5d56` Dec 31, 2012

michaelficarra commented on `45c5d56` Dec 31, 2012

devinrhode2 commented on `45c5d56` Dec 31, 2012 via email

devinrhode2 commented on `45c5d56` Dec 31, 2012 via email

devinrhode2 commented on `45c5d56` Dec 31, 2012 via email

kangax commented on `45c5d56` Jan 2, 2013

michaelficarra commented on `45c5d56` Jan 2, 2013

vendethiel commented on `45c5d56` Jan 2, 2013

michaelficarra commented on `45c5d56` Jan 2, 2013

epidemian commented on `45c5d56` Jan 2, 2013

epidemian commented on `45c5d56` Jan 2, 2013

michaelficarra commented on `45c5d56` Jan 2, 2013

vendethiel commented on `45c5d56` Jan 2, 2013

devinrhode2 commented on `45c5d56` Jan 2, 2013 via email

vendethiel commented on `45c5d56` Jan 2, 2013

kangax commented on `45c5d56` Jan 2, 2013

devinrhode2 commented on `45c5d56` Jan 2, 2013 via email

Constellation commented on `45c5d56` Jan 3, 2013

michaelficarra commented on `45c5d56` Jan 3, 2013

Constellation commented on `45c5d56` Jan 3, 2013

Commit

There are no files selected for viewing

25 comments on commit 45c5d56

devinrhode2 commented on 45c5d56 Dec 31, 2012

Choose a reason for hiding this comment

vendethiel commented on 45c5d56 Dec 31, 2012

Choose a reason for hiding this comment

devinrhode2 commented on 45c5d56 Dec 31, 2012

Choose a reason for hiding this comment

devinrhode2 commented on 45c5d56 Dec 31, 2012

Choose a reason for hiding this comment

vendethiel commented on 45c5d56 Dec 31, 2012

Choose a reason for hiding this comment

devinrhode2 commented on 45c5d56 Dec 31, 2012

Choose a reason for hiding this comment

michaelficarra commented on 45c5d56 Dec 31, 2012

Choose a reason for hiding this comment

devinrhode2 commented on 45c5d56 Dec 31, 2012 via email

Choose a reason for hiding this comment

devinrhode2 commented on 45c5d56 Dec 31, 2012 via email

Choose a reason for hiding this comment

devinrhode2 commented on 45c5d56 Dec 31, 2012 via email

Choose a reason for hiding this comment

kangax commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

michaelficarra commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

vendethiel commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

michaelficarra commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

epidemian commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

epidemian commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

michaelficarra commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

vendethiel commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

devinrhode2 commented on 45c5d56 Jan 2, 2013 via email

Choose a reason for hiding this comment

vendethiel commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

kangax commented on 45c5d56 Jan 2, 2013

Choose a reason for hiding this comment

devinrhode2 commented on 45c5d56 Jan 2, 2013 via email

Choose a reason for hiding this comment

Constellation commented on 45c5d56 Jan 3, 2013

Choose a reason for hiding this comment

michaelficarra commented on 45c5d56 Jan 3, 2013

Choose a reason for hiding this comment

Constellation commented on 45c5d56 Jan 3, 2013

Choose a reason for hiding this comment

25 comments on commit `45c5d56`

devinrhode2 commented on `45c5d56` Dec 31, 2012

vendethiel commented on `45c5d56` Dec 31, 2012

devinrhode2 commented on `45c5d56` Dec 31, 2012

devinrhode2 commented on `45c5d56` Dec 31, 2012

vendethiel commented on `45c5d56` Dec 31, 2012

devinrhode2 commented on `45c5d56` Dec 31, 2012

michaelficarra commented on `45c5d56` Dec 31, 2012

devinrhode2 commented on `45c5d56` Dec 31, 2012 via email

devinrhode2 commented on `45c5d56` Dec 31, 2012 via email

devinrhode2 commented on `45c5d56` Dec 31, 2012 via email

kangax commented on `45c5d56` Jan 2, 2013

michaelficarra commented on `45c5d56` Jan 2, 2013

vendethiel commented on `45c5d56` Jan 2, 2013

michaelficarra commented on `45c5d56` Jan 2, 2013

epidemian commented on `45c5d56` Jan 2, 2013

epidemian commented on `45c5d56` Jan 2, 2013

michaelficarra commented on `45c5d56` Jan 2, 2013

vendethiel commented on `45c5d56` Jan 2, 2013

devinrhode2 commented on `45c5d56` Jan 2, 2013 via email

vendethiel commented on `45c5d56` Jan 2, 2013

kangax commented on `45c5d56` Jan 2, 2013

devinrhode2 commented on `45c5d56` Jan 2, 2013 via email

Constellation commented on `45c5d56` Jan 3, 2013

michaelficarra commented on `45c5d56` Jan 3, 2013

Constellation commented on `45c5d56` Jan 3, 2013