Nits

[iVanlIsh]

No description provided.

[johnathan79717]

I think we should throw a TypeError here if we already know the value isn't valid.

I'm still not sure that we really need public in the IPAddressSpace type since we don't have a valid case to use it. I guess that's why I called it RequestTargetAddressSpace in the first place so that we only need to include the valid values that should be used.

Suggested change

	<dd>Set |request|'s [=target IP address space=] to [=IP address
	space/public=]
	<dd>[=Throw=] a {{TypeError}}.

[iVanlIsh]

I think we need this because we need to define if a request's IP address space is less public then the initiator's. In this case, we need to define initiator's IP address space first, which can be public.

[iVanlIsh]

The problem is actually that we shouldn't put this general definition in the Fetch API section. I think it will make more sense to put it somewhere more generic.

[johnathan79717]

We have already defined the public address space in https://wicg.github.io/private-network-access/#ip-address-space-public. I meant we didn't need it in the IDL, i.e. the JS API. My understanding is that we should define what we will use in the Javascript in the IDL.

[johnathan79717]

If you insist keeping "public" in the IDL, I still think throwing a TypeError here is better than a network error later in the algorithm.

[iVanlIsh]

I meant we should share the same enum other than create a new one for fetch API only.

I don't think this is a TypeError on one hand, and on the other, "public" is forbidden to bypass mixed content check sounds like a better logic than a fetch request cannot be "public" targetAddressSpace.

[iVanlIsh]

The problem is actually that we shouldn't put this general definition in the Fetch API section. I think it will make more sense to put it somewhere more generic.

[johnathan79717]

Thanks for making the changes.