Closed emlun closed 4 months ago
From the WG @ 7/10: I don't have an objection to making this change, but it will mean that PublicKeyCredentialRequestOptionsJSON
will define a combination of DOMString
and USVString
strings. We're exploring whether or not we need to update all DOMString
in ...JSON
dictionaries to USVString
Alternative strategy: define a PublicKeyCredentialRpEntityJSON
that specifies DOMString
so all the ...JSON
dictionaries specify a consistent type for strings.
2024-07-10 WG call: We'll revert the change to PublicKeyCredentialRequestOptionsJSON
and add PublicKeyCredentialRpEntityJSON
as described above.
I made the changes discussed on last week's WG call, but doing that made me question if this was actually the correct resolution. I now believe it is not, and I propose abandoning this PR in favour of the alternative resolution in #2098. See #2098 for expanded discussion and rationale.
rp.id
inPublicKeyCredentialCreationOptions
andrpId
inPublicKeyCredentialRequestOptions
represent the same thing, but with different types. The WG agreed on the 2024-05-15 call that both should beUSVString
. Strictly speaking this is a breaking change (changing a type bound in input (contravariant) position to be more restrictive), but in practice this shouldn't be able to break any applications since then those credentials wouldn't have worked inget()
anyway.Fixes #2066.
Preview | Diff