987110 - Make codegen fail if a dictionary shadows something on Object.prototype

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Boris Zbarsky [:bzbarsky]]

Because that won't work with the Xrays-to-Object bholley wants to set up.

The full list, if I read GlobalObject::initFunctionAndObjectClasses is:

1) "constructor"
2) Everything in object_methods (toSource, toString, toLocaleString, valueOf,
   watch, unwatch, hasOwnProperty, isPrototypeOf, propertyIsEnumerable,
   __defineGetter__, __definesetter__, __lookupGetter__, __lookupSetter__)
3) __proto__

Object.getOwnPropertyNames seems to return that list, minus __proto__ (because it's JSPROP_SHARED or something?).

Jason, do you know how likely it is that new things will get added here?

Comment 1

[Bobby Holley (:bholley)]

(In reply to Boris Zbarsky [:bz] from comment #0)
> Because that won't work with the Xrays-to-Object bholley wants to set up.

For reference, this is bug 987111.

Comment 2

[Boris Zbarsky [:bzbarsky]]

Attachment: Disallow dictionary members that would shadow Object.prototype properties.

Comment 3

[Boris Zbarsky [:bzbarsky]]

Attachment: Disallow dictionary members that would shadow Object.prototype properties.

Comment 4

[Bobby Holley (:bholley)]

Comment on attachment 8395730 [details] [diff] [review]
Disallow dictionary members that would shadow Object.prototype properties.

Review of attachment 8395730 [details] [diff] [review]:
-----------------------------------------------------------------

rs=me

::: dom/bindings/test/mochitest.ini
@@ +35,5 @@
>  [test_lenientThis.html]
>  [test_lookupGetter.html]
>  [test_namedNoIndexed.html]
>  [test_named_getter_enumerability.html]
> +[test_Object.prototype_props.html]

can we call this test_Object_prototype_props.html? This is the first time I've ever seen a . in a filename like this, so it seems inconsistent, and likely to break some fringe piece of automation.

Comment 5

[Boris Zbarsky [:bzbarsky]]

I considered that, but we have a ton of test names like this already, in fact:

find . -name "test_[^.]*.[^.]*.[^.]*" | grep -v .hg/store | wc -l
     225

Comment 6

[Boris Zbarsky [:bzbarsky]]

Er, I guess that includes the imptests .foo.json files, but even if I exclude those:

find . -name "test_[^.]*.[^.]*.[^.]*" | grep -v .hg/store | grep -v json | wc -l
     160

and all of these look legit.

Comment 7

[Bobby Holley (:bholley)]

Shrug. It just seems much more conventional to use an underscore, and less surprising.

Comment 8

[Boris Zbarsky [:bzbarsky]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/db3cd3239562 but I kept the naming.  It makes sense to do it if it doesn't break anything, and makes it clearer that Object.prototype is not the same thing as "Object prototype" here.

Comment 9

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/db3cd3239562