Open alextok opened 8 months ago
The case for keeping it in the header is if there is a use case in which the user agent would need a valid access token to access the /securesession/startsession endpoint. Is that a possibility? The proposed standard just mentions allowing "...the server to link registration with some preceding sign in flow." If that's the whole story then I agree we don't need it in the header.
I removed it from the header here: https://github.com/WICG/dbsc/commit/cffa9fbc0515382a45880c6fb140c1612cf3073f
Keeping this issue open in case there is a case for keeping it in the header like @mattjm mentions. I think it would be best for the server to expect it in the JWT as it is signed by the key.
In startsession we deliver authorization artifact in two different ways as Authorization header and as part of JWT body.
https://github.com/WICG/dbsc?tab=readme-ov-file#start-session
I think we need to have one way of doing this. I prever JWT body, as it cryto bound to keys.