Expressing that a property is non-deletable

Tom Van Cutsem tomvc.be at gmail.com
Wed Dec 19 06:02:30 PST 2012


2012/12/19 David Bruant <bruant.d at gmail.com>

> Le 18/12/2012 20:09, Mark S. Miller a écrit :
>
>> I see no reason why this needs to be a reflected property. As to
>> whether it is an exotic internal property or just prose, that is a
>> specification expository issue for which I defer to Allen. But the
>> spec only needs such extra state for the exotic global object. There's
>> nothing general about it.
>>
> I fully agree on the fact that if there was a "deletable" attribute it
> would be "local" to the global object and wouldn't need to be an attribute
> reflected on all objects.
> In ES5, we've been thinking of property attributes as absolute thing that
> apply to every single objects. Maybe it's time to give each sort of object
> a little bit of freedom to explain how their properties work, to allow the
> reflection methods to describe their internal contract more accurately than
> they currently can with configurable/writable/**enumerable/get/set
> In this case, maybe there is value in having WindowProxy objects telling
> us whether or not a property is deletable or not (because the configurable
> boolean which is used for that in other objects doesn't have the same
> semantics here).
>
> Maybe numerical attributes of NodeList could provide a "live" or
> "syncWithDOMTree" attribute clearly showing that they are not your regular
> numerical properties.
>
> I'm putting maybe-s because I don't know if there is value in doing that,
> but the point I am trying to make is that this can be the way out to enable
> all objects to respect invariants while allowing different objects to
> express their internal contract.


Yes. This is exactly what I meant with using custom attributes to document
exotic behavior.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.mozilla.org/pipermail/es-discuss/attachments/20121219/8055282b/attachment.html>


More information about the es-discuss mailing list