Odd idea

[Brendan Eich]

On May 19, 2008, at 5:41 PM, Mark S. Miller wrote:

> On Mon, May 19, 2008 at 4:42 PM, Brendan Eich <brendan at mozilla.org>  
> wrote:
>> On May 19, 2008, at 4:22 PM, Mark S. Miller wrote:
>>> But ES4 is also sideways in this sense. There's a bunch of stuff in
>>> Mozilla's JS1.8 that didn't make it into ES4.
>>
>> Namely? As noted, some pieces are prototypes that will be adjusted  
>> to match
>> the ES4 type-based counterpart (the iteration protocol hook,  
>> e.g.). What
>> bunch of stuff is in 1.8 that did not make it into the latest ES4  
>> drafts?
>
> Ok, I looked, and it's a lot less than I expected. "watch" is an
> example. (Unless I didn't notice its inclusion).

Object.prototype.watch and unwatch have been in SpiderMonkey for ~12  
years (memory fades -- in rev 1.1 on cvs.mozilla.org, I recall  
implementing in Netscape's private CVS years before mozilla.org was  
founded).


>>> Also, there's a tremendous amount of stuff in ES4 that was never  
>>> in a
>>> JavaScript.
>>
>> Except under the hood, off limits to programmers, reserved for the  
>> built-ins
>> and the DOM.
>
> Huh? Classes, Type declarations, Namespaces!,

Classes under the hood (interfaces too) in built-ins and the DOM and  
browser object models. You cannot bootstrap JS in JS without  
something like classes.

Type declarations mean several things, but let's pick just structural  
types: under the hood in SpiderMonkey at least.

Namespaces: fair enough, although the built-ins still get special  
treatment. What else are the [[Get]], etc. "internal property names"  
in ES1-3, but non-default namespace prefixes disguised as semantic  
brackets. These names are mangled in a way that cannot be spelled in  
the language, yet they are property names.


> perhaps Packages and/or
> Units, if these are still on the table.

They were cut -- please try to keep up, we ES4-ers are spending time  
keeping up with 3.1 :-/.


> Namespaces is a huge addition to the complexity of the language, and
> the one I'm least happy about.

Actually, I agree with you that namespaces add more complexity than  
classes alone (if that's what you mean). We've been working on them  
for a long time (Waldemar can tell you about the first go-round).  
They're too useful to lose in favor of privileged built-in names and  
__UGLY__ conventions for the lusers.


>> and repeat that we wouldn't fold any 3.1 into a distinct *JS*  
>> version number.
>
> Ok then, I'm happy to stop arguing about this. I just thought that
> this odd idea might be seen as helpful. If not, forget it. We have
> enough substantive issues to argue about ;).

Will do, although a transcendental version number will come in handy  
some day, I'm sure ;-).

/be