Implement stage 3 proposal globalThis

[rhuanjl]

This PR implements the stage 3 proposal globalThis.

1. Actual implementation is 2 lines of code
2. I've added a default enabled flag so it can be flagged off if ever necessary
3. I've added a few tests which I think should be sufficient
4. The large size of the diff is only due to it needing a byte code regen

cc: @ljharb @pleath

closes: #1658

[ljharb]

Thanks, LGTM!

[fatcerberus]

Hooray! 🎉

Going off on a tangent, I find the name of globalThis to be awkward; I understand why it was chosen (global, the best option, being dead in the water thanks to web compat), but a camelcased (pseudo)keyword is weird and it’s also somewhat ambiguous—as in strict mode (and modules), this is actually undefined when used in the global scope.

[ljharb]

See https://github.com/tc39/proposal-global/blob/master/NAMING.md for the constraints

[fatcerberus]

Thanks, yeah, I understand the challenges around the proposal (I was very sad when global was ruled out), it’s mostly the ambiguity that gets to me in the end - since this is largely provided specifically so that strict-mode code can access the global easily (per the linked document), “global this” is a bit misleading - since the actual “global this” in strict mode is undefined (I think...)

[ljharb]

It'd be great to get this merged :-)

[boingoing]

LGTM. Also want to agree the name globalThis is an odd choice.

[boingoing]

Hi @rhuanjl, would you mind rebasing this one? I will help merge it.

[rhuanjl]

I've noticed that there is an agenda item for the TC39 meeting due to be held this week to discuss changing the name of globalThis, should we hold off on getting this merged until after that discussion?

https://github.com/tc39/agendas/pull/511/files

@ljharb

[ljharb]

@rhuanjl Given that we already discussed it in November, and decided not to change the name, I find it highly unlikely that anything will change - in addition, Chrome and Safari are already both shipping under the current name.

It's up to yall of course, but I'd suggest merging as-is, and in the unlikely event of a name change, it can be updated.

[rhuanjl]

@ljharb I can see (from it being crossed out on the agenda) that the name change discussion has taken place - can you tell us if the name is being changed/is likely to be?

[ljharb]

@rhuanjl there was not consensus for any change, no.

[rhuanjl]

@boingoing as per @ljharb's comment it sounds like the name won't be changing. I've rebased on latest master, ok to merge?

[boingoing]

@rhuanjl Yup, I'll run the change through our internal test suites and merge it when it's green.

[boingoing]

@rhuanjl Took the liberty of rebasing and adding a new bytecode GUID into your branch. Hope you don't mind. Landed via 5d83d3e

Thanks very much for the submission. Always nice to see your contributions.