SetMutableBinding and ImportBindings

[jmdyck]

This concerns:

• immutable indirect bindings (IIB) created by CreateImportBinding for module Environment Records, and
• the SetMutableBinding method for declarative Environment Records (which includes module Environment Records).

Currently, SetMutableBinding doesn't explicitly say how it handles an IIB. Mostly, it doesn't need to be explicit:

• an IIB is initialized when created, so the 'has not yet been initialized' step doesn't apply, and
• an IIB is immutable, so the 'is a mutable binding' step doesn't apply.

So it's clear that SetMutableBinding won't change the IIB, the only question is whether to throw a *TypeError* or not. Which depends on:

• whether an IIB is a strict binding, or
• whether the passed-in _S_ is *true*.

Now it definitely isn't stated that an IIB is a strict binding, but it might be intended. And I think the passed-in _S_ is always *true* for any case involving an IIB (i.e., for a module Environment Record), but I'm not completely sure.

So: does SetMutableBinding always throw a *TypeError* when the binding in question is an IIB?

[jmdyck]

(The answer may affect how I finally resolve PR #2288.)

[michaelficarra]

I don't think there's anything preventing us from saying it's strict, so that seems the preferable option. In which case, a TypeError should be thrown.

This should be testable, right? Can we see what implementations do here?

[jmdyck]

This should be testable, right? Can we see what implementations do here?

If the passed-in _S_ is *true*, a TypeError should be thrown regardless of the strictness of an IIB. So the most interesting test would cause the passed-in _S_ to be *false*, because then the results would reveal whether implementations treat an IIB as a strict binding. However, I don't know if it's possible to construct such a test.

[michaelficarra]

Ah, since direct evals in strict mode code are also strict, I don't know how to exercise this either.