fixes #28: allow anonymous well-known intrinsic objects

[michaelficarra]

Fixes #28. I'm assuming here that "named" is the opposite of "anonymous" currently used in the spec. If we think that's unclear, I can change it to "non-anonymous", but that sounds like a double negative to me.

/cc @anba for review

[ljharb]

This seems like an ambiguity; I would expect that an anonymous function with a name of the empty string (whether in owned property or not) would appear as an anonymous function.

Is there a better way to phrase this that doesn’t hinge on IdentifierName?

[michaelficarra]

@ljharb Since IdentifierName is optional, I think that works out just fine.

[anba]

I'm assuming here that "named" is the opposite of "anonymous" currently used in the spec. If we think that's unclear, I can change it to "non-anonymous", but that sounds like a double negative to me.

https://tc39.github.io/ecma262/#sec-ecmascript-standard-built-in-objects uses "is not identified as an anonymous function", so "Additionally, if _func_ is a <emu-xref href="#sec-well-known-intrinsic-objects">Well-known Intrinsic Object</emu-xref> and is not identified as an anonymous function, [...]" should work out.

Also revert 994e506 to make %ThrowTypeError% anonymous again?

I would expect that an anonymous function with a name of the empty string (whether in owned property or not) would appear as an anonymous function.

Based on https://tc39.github.io/ecma262/#sec-ecmascript-standard-built-in-objects and https://tc39.github.io/ecma262/#sec-function-instances-name, I'd say a function is anonymous if and only if the function does not have an own name property.

[michaelficarra]

Also revert 994e506 to make %ThrowTypeError% anonymous again?

Why?

[ljharb]

It's really nice for debugging to have everything named; what's the benefit to making it anonymous?

[anba]

Because we have multiple anonymous built-in functions and only giving %ThrowTypeError% a name because it is also a well-known object seems questionable. (This is with my spec hat on, for example when only considering the implementation in SpiderMonkey, I'd say remove all the anonymous built-in functions.)

[ljharb]

I'd prefer any built-in function that's available at realm creation time be named; are there others besides ThrowTypeError?

[anba]

It's more likely a user gets a reference to, let's say a Proxy revocation function, than a reference to %ThrowTypeError%. So when we speak about the debugging experience, adding a name to Proxy revocation functions seems more useful than adding a name to %ThrowTypeError%, right?

[michaelficarra]

@anba Regardless, it is better to have it be named than not named.

edit: Though now that it's not strictly necessary as part of this proposal, I will remove it. But I still think we should name them when it would be helpful to do so.

[anba]

Regardless, it is better to have it be named than not named.

In that case adding the name should happen through a normal PR to https://github.com/tc39/ecma262 instead of this proposal. :-)