With the core drafts finally settling in, I think it's time for the WG
to look into a widely-used usecase of OAuth 1.0 that's not currently
addressed directly by OAuth2: classical 2-legged OAuth 1.0, or tokenless
signed http fetch. 

People use this method today to replace developer key credentials, but
also to ensure validity of parameters in the request. Tokens (especially
signed tokens) can help in the first case, but not so much in the
second, unless I'm missing a key feature of JWT. 

I'm personally fine with profiling the 1.0 signature method for use with
generic HTTP messages, apart from OAuth and tokens entirely. I'd be
happier with a system that had an explicit signature base string (such
as what Brian Eaton proposed ages ago) instead of the blind magic that
OAuth 1.0 uses.

I personally wouldn't know where to start with normative language for
something like this, as I am not a signatures/crypto guy. But I'd be
happy to help edit such a document and work on profiles and use cases
for this. 

My question for the WG is then: are signed HTTP fetch messages something
we want to take on, and who would be willing/able to edit such a
document?

 -- Justin

_______________________________________________
OAuth mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/oauth

Reply via email to