update precedent section in README

[michaelficarra]

1. it's not a brand check like isError or isArray, it's just a check
2. link to isError in 262, it's no longer a proposal
3. add reference to Array.isArray as well
4. mention the associated magic behaviour for each of the precedent-setting methods

[ljharb]

Of course it's a brand check, it looks for an internal slot.

[michaelficarra]

Brands are indelible and innate. All Dates are created with a [[DateValue]] slot which cannot be removed. Thus it's a brand. The [[Prototype]] slot having a particular value is not a brand because its value can change over time, "removing" the "brand".

[bakkot]

This proposal (at least as currently specified) does not look at the [[Prototype]] slot. There's different operations which do that subsequent to checking IsPromise, but this proposal would just be the IsPromise part, and you can then check the [[Prototype]] slot yourself if you want to match those operations.

[michaelficarra]

Ah I see, this is just checking [[PromiseState]]. That does appear to be a brand.

[mhofman]

Yup this is specifically about exposing the brand check that is used as part of different checks through the spec. The goal is to be able to know what behavior a value will have when used by these spec operations (e.g. await), so that userland can decide to handle other values differently.

[mhofman]

I was originally reticent to mention the Array.isArray case since the IsArray AO has special behaviors for proxies which wouldn't exists for promises, but I think that difference is not meaningful for most.

[michaelficarra]

I think it's actually a good thing to mention isArray for that reason so that we can explicitly call out to committee that we do not want to pierce proxies like isArray.