Re: [OAUTH-WG] auth-param syntax, was: OK to post OAuth Bearer draft 15?

[Julian Reschke <[e-mail]>]

On 2011-12-31 00:19, Mike Jones wrote:
> I did already back the statement that this is the working group consensus with the e-mails attached in this note sent to you on December 12, 2011:
>    - http://www.ietf.org/mail-archive/web/oauth/current/msg08042.html

I replied in 
<http://www.ietf.org/mail-archive/web/oauth/current/msg08043.html>:

"I'm not disagreeing with the decision not to allow "\" in the value. 
What I'm disagreeing with is writing the ABNF in a way that will make it 
likely for implementers to special-case OAuth parameters when they 
should not."

So you're citing a consensus for a related but different question. I 
recommend to read the mailing thread to the end.

> As for your assertion that the specs are in conflict, yes, the Bearer spec includes a different decision than a RECOMMENDED clause in the HTTPbis spec (which was added after the Bearer text was already in place).  However, it is not violating any MUST clauses in the HTTPbis spec.  Given that no MUSTS are violated, I don't see it mandatory for this tension to be resolved in favor of one spec or the other in order for both to be approved as RFCs.  I look forward to seeing that happen soon in both cases (and for the OAuth core spec as well).

As a matter of fact, the HTTPbis P7 text on considerations for new 
schemes doesn't use any BCP14 keywords at all. That's on purpose, 
because we think they should be used with care, and in particular that 
they should only be used to discuss the protocol, not the style of other 
specifications.

So it's really not relevant; what's essential is the intent of the spec 
text, and I believe that is VERY clear:

    o  The parsing of challenges and credentials is defined by this
       specification, and cannot be modified by new authentication
       schemes.  When the auth-param syntax is used, all parameters ought
       to support both token and quoted-string syntax, and syntactical
       constraints ought to be defined on the field value after parsing
       (i.e., quoted-string processing).  This is necessary so that
       recipients can use a generic parser that applies to all
       authentication schemes.

(Note the "cannot").

So again, if you disagree with this statement, please argue your case in 
the HTTPbis WG.

If you *do* agree, but somehow feel that the bearer spec can't do this, 
the bearer spec should document the reason (just like when an 
implementation fails to implement a SHOULD).

As to the question of timing (when certain paragraphs were added): yes, 
HTTPbis P7 changed based on feedback and review of the OAuth bearer spec 
(triggered by James Manger). That's a feature.

If it hadn't, for instance, the bearer spec wouldn't conform to the base 
grammar *at all*. See 
<http://trac.tools.ietf.org/wg/httpbis/trac/ticket/195>.

Best regards, Julian