Normative: Add import.meta[Symbol.toStringTag]

[ExE-Boss]

See tc39/proposal-import-meta#18 and #1970 (comment).

---

Preview (#sec-meta-properties-runtime-semantics-evaluation)

[ljharb]

I failed to mention this in this week's meeting, so it will probably have to wait until the next meeting to be discussed.

[bmeck]

I'm not convinced we should do this.

• Hosts can supply toStringTag already
• Hosts can replace the prototype of import.meta objects (none do ?)
• Unlike other objects created by the spec this is generated by host data and not something we have looked at adding properties to at TC39. I'd even go so far as to state TC39 should never populate these objects as even things designed to have some host interoperability like import.meta.url are not consistent across platforms for various known reasons.
• There are many of these objects unlike the other things we add @@toStringTag, it seems odd to add many own properties where other namespace objects only add it to a central location (because they are a centralized location).

[ljharb]

The fourth point is i think addressed by https://tc39.es/ecma262/#sec-@@tostringtag - Module Namespace objects have multiples and have own Symbol.toStringTag properties, and since they're own properties, the second point isn't relevant.

The third point in particular remains, and is compelling, but I still think it's worth a quick discussion in plenary. I'll add it to the agenda for the meeting after next (timezones are prohibitive for me for the next meeting).

[ljharb]

The committee decided not to provide consensus for this at today's plenary.

However, in the future if there are more compelling use cases, or if debuggability (the lack of a tag) turns out to be a bigger issue, the committee is welcome to revisit it.