Fix definition of response of POST /_matrix/federation/v1/user/keys/claim
#1559
Conversation
…laim Also fixes the rendering of nested OneOf. Signed-off-by: Kévin Commaille <[email protected]>
Signed-off-by: Kévin Commaille <[email protected]>
| - type: string | ||
| - type: object |
There was a problem hiding this comment.
I'm unconvinced this is correct. Is it true that a string can be returned here? The description doesn't seem to say anything about that being permissible.
@KitsuneRal you seem to have suggested this is correct at #1351 (comment), but I don't know why you said that?
There was a problem hiding this comment.
Good question - might very well be a copy-pasta on my end. Can one-time keys be unsigned - guess not? If not, my suggestion is incorrect and there should be KeyObject here, without alternatives.
There was a problem hiding this comment.
Ok, so it's not a copy-pasta on my end at least. The alternative between string and object has been here since the very beginning, even though undocumented. My comment in #1351 referred to the correct nesting level and I mentioned that string has to be returned only because it has been there before. If it hasn't been correct from the beginning then I'd rather we fix it in a separate PR but I'm fine either way.
There was a problem hiding this comment.
Right, I think I've figured this out!
Before we invented the signed_curve25519 key algorithm for one-time-keys, we just used curve25519 keys. These, not being signed, can be represented just by a string.
The curve25519 algorithm should no longer be used here, but removing them from the spec sounds like a different job.
So, 👍 to this fix.
There was a problem hiding this comment.
Signed-off-by: Kévin Commaille <[email protected]>
As discussed in #1351 (comment) and #1310 (comment).
The rendering of the type now looks like this (to compare with current state in the spec in the response of the endpoint):
Also fixes the rendering of nested
OneOfs, so it has the added bonus of fixing the rendering ofm.room.encrypted.ciphertext:CiphertextInfo:cc @KitsuneRal to make sure I got it right this time.
Preview: https://pr1559--matrix-spec-previews.netlify.app