Object id, hash, etc?

Michael McGlothlin mike.mcglothlin at gmail.com
Wed Sep 9 13:14:20 UTC 2015


I think using WeakMaps is a good idea. Using a Symbol will keep from cluttering the namespace with names others might want to use (id,label,etc) and seems more consistent than some external means of accessing the id. When I tried defining the value before I ran into issues with the prototype chain which is why I was using get and a second 'hidden' property. Using WeakMap seems like it'd fix the issue.

Can't say I've used Object.assign much - will have to experiment with it.

Thanks,
Michael McGlothlin
Sent from my iPhone

> On Sep 9, 2015, at 7:34 AM, Andrea Giammarchi <andrea.giammarchi at gmail.com> wrote:
> 
> Well, to start with, you don't hide symbols, these are all exposed through `Object.getOwnPropertySymbols` or Reflect.ownKeys so for your use case looks like using just a non enumerable property would do as well.
> And btw, even if Symbols don't show up in for/in and Object.keys, these are by default also enumerable so Object.assign will copy them around (reason I've used defineProperty with the value instead of just setting the symbol, so it's non enumerable by default).
> 
> As summary: do you want to be sure that object has a unique "lable" or "id" associated with it? Use WeakMaps, otherwise be prepared to possible clones around if some library uses Object.assign thinking that's a good way to copy properties around.
> 
> I still suggest the Labeler approach that Mark previously indicated.
> 
> Regards
> 
> 
> 
> 
>> On Wed, Sep 9, 2015 at 1:15 PM, Michael McGlothlin <mike.mcglothlin at gmail.com> wrote:
>> Using frozen objects and such seems a lot more complex to me than just using a property. Using a symbol to hide the property from cluttering seems to go along with how other things are being done. And it's my understanding that other than hiding it that using a Symbol for a name is nothing special - does it have some negative factor that should make using it less attractive?
>> 
>> 📱 Michael McGlothlin
>> 
>>> On Sep 9, 2015, at 5:36 AM, Andrea Giammarchi <andrea.giammarchi at gmail.com> wrote:
>>> 
>>> That's similar to the Labeler proposed by Mark, except it needs to search for the thing twice (wm.has + wm.get instead of just wm.get which would be  just fine as check if you count from 1 instead of 0 ;-) )
>>> 
>>> Although I need to understand if you have a wm why wouldn't you just use that instead of a Symbol ... so the Labeler seems again a more appropriate pattern for this problem.
>>> 
>>> Just my thoughts
>>> 
>>>> On Wed, Sep 9, 2015 at 10:51 AM, Michał Wadas <michalwadas at gmail.com> wrote:
>>>> Following solution will work:
>>>> 
>>>> (function(){
>>>> const wm = new WeakMap();
>>>> Symbol.identity = Symbol( 'Symbol.identity' );
>>>> let i = 0;
>>>> Object.defineProperty(
>>>>   Object.prototype,
>>>>   Symbol.identity,
>>>>   {
>>>>     get: function () {
>>>>       if (wm.has(this)) return wm.get(this);
>>>>       wm.set(this, Symbol(i++));
>>>>       return wm.get(this);
>>>>     }
>>>>   }
>>>> );
>>>> })()
>>>> 
>>>> 
>>>> 2015-09-08 23:29 GMT+02:00 Andrea Giammarchi <andrea.giammarchi at gmail.com>:
>>>>> as side note, that's just a lazy assignment that doesn't need two symbols and a constant get invoke ...
>>>>> 
>>>>> ```js
>>>>> Symbol.identity = Symbol( 'Symbol.identity' );
>>>>> var OBJECT_ID = 0;
>>>>> Object.defineProperty(
>>>>>   Object.prototype,
>>>>>   Symbol.identity,
>>>>>   {
>>>>>     get: function () {
>>>>>       // first time this is invoked ... and no more ...
>>>>>       Object.defineProperty(this, Symbol.identity, {value: ++OBJECT_ID});
>>>>>       // from now own, direct property access \m/
>>>>>       return this[Symbol.identity];
>>>>>     }
>>>>>   }
>>>>> );
>>>>> ```
>>>>> 
>>>>> Regards 
>>>>> 
>>>>>> On Tue, Sep 8, 2015 at 9:57 PM, Michael McGlothlin <mike.mcglothlin at gmail.com> wrote:
>>>>>> I try to keep it pretty simple. It's not fancy but the times you want fast and dirty information like this are the same times you don't want to have to define it manually.
>>>>>> 
>>>>>>  Symbol.identity = Symbol( 'Symbol.identity' );
>>>>>>  const identity = Symbol( 'identity' );
>>>>>>  var OBJECT_ID = 0;
>>>>>>  Object.defineProperty( Object.prototype, Symbol.identity, {
>>>>>>   get: () => {
>>>>>>    if ( !Object.hasOwnProperty.call( this, identity ) ) {
>>>>>>     this[ identity ] = ++OBJECT_ID;
>>>>>>    }
>>>>>>    return this[ identity ];
>>>>>>   }
>>>>>>  } );
>>>>>> 
>>>>>>> On Tue, Sep 8, 2015 at 1:44 PM, Mark S. Miller <erights at google.com> wrote:
>>>>>>> See Labeler at
>>>>>>> http://wiki.ecmascript.org/doku.php?id=harmony:weak_maps#unique_labeler
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>>> On Tue, Sep 8, 2015 at 11:16 AM, joe <joeedh at gmail.com> wrote:
>>>>>>>> Didn't send to list, something is wrong with my reply all. Sorry about that. Stupid mobile gmail.
>>>>>>>> 
>>>>>>>> ---------- Forwarded message ----------
>>>>>>>> From: "joe" <joeedh at gmail.com>
>>>>>>>> Date: Sep 8, 2015 11:15 AM
>>>>>>>> Subject: Re: Object id, hash, etc?
>>>>>>>> To: "Garrett Smith" <dhtmlkitchen at gmail.com>
>>>>>>>> Cc: 
>>>>>>>> 
>>>>>>>> I agree with this request. This is the logical complement to valueof, I think. And yes, for most ID use cases this isn't a good fit, but we're not talking about the general case, just the cases where a python style id() function *is* appropriate.
>>>>>>>> 
>>>>>>>> Joe
>>>>>>>> 
>>>>>>>>> On Sep 8, 2015 9:08 AM, "Garrett Smith" <dhtmlkitchen at gmail.com> wrote:
>>>>>>>>> On 9/8/15, Michael McGlothlin <mike.mcglothlin at gmail.com> wrote:
>>>>>>>>> > Is there a reason not to provide an object id and hash value as other
>>>>>>>>> > languages often do? I find myself defining an identity symbol on objects
>>>>>>>>> > with a value from a simple global counter. It makes it easier to debug if I
>>>>>>>>> > can just look at a log and see what objects and functions were active.
>>>>>>>>> >
>>>>>>>>> > Likewise a hash value would simplify a lot of comparisons and make it easier
>>>>>>>>> > to debug.
>>>>>>>>> >
>>>>>>>>> > I especially find the ID useful when working with bound functions as they
>>>>>>>>> > can be difficult to tell apart. I like to change toString() to provide the
>>>>>>>>> > ID followed by the code (flattened to one line).
>>>>>>>>> >
>>>>>>>>> >
>>>>>>>>> 
>>>>>>>>> NFE's are safe to use where IE8 support isn't needed*.
>>>>>>>>> (function aa(){}).name; // "aa"
>>>>>>>>> 
>>>>>>>>> As for using Object IDs and Object Pooling, lexically-scoped values
>>>>>>>>> have benefits over global ID generators (as your counter). The values
>>>>>>>>> can be passed in as a parameter to the Factory (useful with Private
>>>>>>>>> Proxy). There other benefits to this pattern regarding memory
>>>>>>>>> management and application design. Contact me if you'd like to learn
>>>>>>>>> more.
>>>>>>>>> 
>>>>>>>>> * https://kangax.github.io/nfe/
>>>>>>>>> --
>>>>>>>>> Garrett
>>>>>>>>> @xkit
>>>>>>>>> ChordCycles.wordpress.com
>>>>>>>>> garretts.github.io
>>>>>>>>> personx.tumblr.com
>>>>>>>>> _______________________________________________
>>>>>>>>> es-discuss mailing list
>>>>>>>>> es-discuss at mozilla.org
>>>>>>>>> https://mail.mozilla.org/listinfo/es-discuss
>>>>>>>> 
>>>>>>>> _______________________________________________
>>>>>>>> es-discuss mailing list
>>>>>>>> es-discuss at mozilla.org
>>>>>>>> https://mail.mozilla.org/listinfo/es-discuss
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> -- 
>>>>>>>     Cheers,
>>>>>>>     --MarkM
>>>>>>> 
>>>>>>> _______________________________________________
>>>>>>> es-discuss mailing list
>>>>>>> es-discuss at mozilla.org
>>>>>>> https://mail.mozilla.org/listinfo/es-discuss
>>>>>> 
>>>>>> 
>>>>>> _______________________________________________
>>>>>> es-discuss mailing list
>>>>>> es-discuss at mozilla.org
>>>>>> https://mail.mozilla.org/listinfo/es-discuss
>>>>> 
>>>>> 
>>>>> _______________________________________________
>>>>> es-discuss mailing list
>>>>> es-discuss at mozilla.org
>>>>> https://mail.mozilla.org/listinfo/es-discuss
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.mozilla.org/pipermail/es-discuss/attachments/20150909/675d2e33/attachment.html>


More information about the es-discuss mailing list