fetch the descendants of and link a module script only handles [[ParseError]] case

[allstarschh]

What is the issue with the HTML Standard?

In fetch the descendants and link a module script, it will call LoadRequestedModules to fetch the descendants, which will in turn call HostLoadImportedModules.
And if there's any exception thrown during HostLoadImportedModules, it will call FinishLoadingImportedModule.
But only the ParseError case, which is handled in step 14.3.3 of HostLoadImportedModules, will be passed to the loadState.[[ParseError]]

If loadState is not undefined and loadState.[[ParseError]] is null, set loadState.[[ParseError]] to parseError.

Then the loadPromise from LoadRequestedModules will be rejected, in step 7.1 of fetch the descendants and link a module script

If state.[[ParseError]] is not null, set moduleScript's error to rethrow to state.[[ParseError]] and run onComplete given moduleScript.

But what about the other errors thrown during HostLoadImportedModules? For example, when the module type allowed returns false, or when the resolutionError is thrown in Step 9. These doesn't set the [[ParseError]] state, hence when the loadPromise is rejected in fetch the descendants and link a module script, it will run step 7.2.

Otherwise, run onComplete given null.

This will lead to the onerror handler of the script will be called,

The change is made from #8253, but contradicts the original behavior.
Originally, resolution errors and errors for module types were treated as parse errors.
https://web.archive.org/web/20221130124954/https://html.spec.whatwg.org/#validate-requested-module-specifiers

The wpt, [/import-maps/bare-specifiers.sub.html], and [/html/semantics/scripting-1/the-script-element/import-attributes/invalid-type-attribute-error.html] still expect the legacy behavior

CCing @nicolo-ribaudo

[nicolo-ribaudo]

This change was indeed not intentional. Is it correct that fetch errors should call onComplete given null, while everything else should be propagated as a script with an error to rethrow?

[domenic]

Yes, I think that was the intended behavior. If you are able to fix that back to how it was previously, that would be much appreciated!

[nicolo-ribaudo]

I'll work on this, but it will take a few weeks