> -----Original Message-----
> From: Justin Richer [mailto:[email protected]]
> Sent: Tuesday, March 15, 2011 2:38 PM

> Having a standard set of oauth errors on the protected
> resource is a good idea. That said:

Not practical. Given the wide range of token types and schemes, each will have 
to define its own way of handling errors. For example, MAC uses HTTP codes 
because the error codes were a direct map and defining extra values was simply 
wasteful. Bearer takes a different approach. No value in forcing a unified 
scheme approach here.
 
> Like I said in a previous email, I'm happy with it being a uri-based 
> extension,
> like grant_type.

Great.

> I still think there's value in structuring what the short codes are

It's just a pre-defined, non-extensible set. Anything else is a URI, just like 
grant type. Exactly the same.

> I disagree that the errors are all about the client being able to 
> automatically
> do something about them -- they're just as useful to push up the stack to
> developers.

For that we have two human readable error fields.

EHL


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

Reply via email to