Move detach check after argument coercion in resize

[syg]

Closes #116

[syg]

cc @phoddie

[ljharb]

I’d rather duplicate the steps to be consistent with validating the receiver before arguments, but this is fine too.

[bakkot]

Alternatively we could stop coercing things entirely, thus eliminating this whole class of issue in this and future APIs. (On the agenda for July.)

[littledan]

Is there a test in test262 (even in staging) which exposes this corner of the semantics? By #116 I take it that, initially, there wasn't a test enforcing the opposite order, at least.

[syg]

Is there a test in test262 (even in staging) which exposes this corner of the semantics? By #116 I take it that, initially, there wasn't a test enforcing the opposite order, at least.

I don't think there was, no.

[littledan]

OK, I'm fine with a fix like this landing, but IMO there should be tests covering things that we bother making a PR about (even if it's an edge case, though IMO it's OK if the tests are in staging) before the proposal reaches Stage 4.

[syg]

OK, I'm fine with a fix like this landing, but IMO there should be tests covering things that we bother making a PR about (even if it's an edge case, though IMO it's OK if the tests are in staging) before the proposal reaches Stage 4.

Getting the tests migrated out of staging/ is a Stage 4 requirement anyhow and I'm currently working on that PR. I'll add a test for this in that PR. If I can't finish before the meeting, then I'll wait to ask for Stage 4 next time.