"How would this proposal work with caching?" is misleading

[linusg]

The determination of whether the type attribute is part of the module cache key is left up to hosts (as it is for all import attributes).

That's not true. From https://tc39.es/ecma262/#sec-HostLoadImportedModule:

If this operation is called multiple times with two (referrer, moduleRequest) pairs such that:

• the first referrer is the same as the second referrer;
• ModuleRequestsEqual(the first moduleRequest, the second moduleRequest) is true;

and it performs FinishLoadingImportedModule(referrer, moduleRequest, payload, result) where result is a normal completion, then it must perform FinishLoadingImportedModule(referrer, moduleRequest, payload, result) with the same result each time.

The import attributes are part of the module request so they are mandatory for the cache key.

[nicolo-ribaudo]

When the import attributes proposal was still "import assertions", we used to require that import assertions were not part of the cache key: import "foo" and import "foo" assert { attr: "value" } were required to return the same module, because the attribute was not part of the cache key and thus not a distinguishing factor.

Since the update to "import attributes", that's not true anymore. Now we allow hosts to look at the attributes when determining the cache key, so they are free whatever they want: both guaranteeing that import "foo" and import "foo" with { attr: "value" } return the same module, or choosing to return different modules, are valid behaviors.

We only restrict hosts to not use more information for building the cache key other than the referrer, the specifier, and the attributes.

[linusg]

You're correct, but it wasn't clear to me that the FAQ section is addressing imports of the same specifier with and without attributes. I was thinking about this in terms of importing the same specifier with the same { type: "bytes" } attributes.

Since the proposal is advertised as a direct replacement for existing file reading APIs one might think that it's up to the host whether dynamic import() will reload the file from disk - which is not the case due to the imposed caching requirements.

[styfle]

@linusg Want to make a PR to clarify the wording here?