[Openid-specs-ab] Review Comments on Discovery

Mike Jones Michael.Jones at microsoft.com
Thu Nov 7 23:43:06 UTC 2013


I'm OK with most of these proposed resolutions.  I'll note that the bitbucket version already addresses the comment about the issuer values being equal.

I disagree with changing the term "Authorization Endpoint" to "Authorization and Authentication Endpoint".  This isn't the OAuth terminology and would cause more confusion than clarity.  And it would change the text in *lots* of places.  We've already defined "Authentication Request", which makes sense, since it has a different definition than "Authorization Request".  But I *really* don't want us to have two terms for exactly the same thing.

The "portion" and "component" language I believe is rooted in the underlying RFCs (or at least is supposed to be).  Please look into how these terms are used in those documents before suggesting changes.

Support for request_uri is not MTI for non-open deployments, so the request_uri_parameter_supported parameter is needed.

Nat, per our hallway conversation - if you want to send us proposed text changes for these and you have the time to do so - preferably as Word diffs, that would be helpful.  (Specific proposed wording for proposed issue resolutions is always welcome from anyone! :))  That would also make it easy for people to review the exact proposed changes.

                                                            -- Mike

From: Nat Sakimura [mailto:sakimura at gmail.com]
Sent: Thursday, November 07, 2013 10:32 AM
To: Mike Jones
Cc: Torsten Lodderstedt; Openid-specs Ab
Subject: Re: [Openid-specs-ab] Review Comments on Discovery

Mike:

OK. As long as you can do it reasonably quickly, I am fine. Much better to have a single person editing at this stage.

For the WG:

If you do not agree to the proposed DoC, you need to respond quickly now, within a day or two.

2013/11/7 Mike Jones <Michael.Jones at microsoft.com<mailto:Michael.Jones at microsoft.com>>
Please do not edit any of the sources.  The working group has not had time to review any of this.  I know that I haven't.

From: Nat Sakimura [mailto:sakimura at gmail.com<mailto:sakimura at gmail.com>]
Sent: Thursday, November 07, 2013 9:42 AM
To: Torsten Lodderstedt
Cc: Openid-specs Ab; Mike Jones
Subject: Re: [Openid-specs-ab] Review Comments on Discovery

Mike -

If Torsten agrees to the Proposed DoC, shall I apply the edit to the XML since you have boat road of edits to toher documents including JOSE and OAuth WG stuff?

Torsten,

Is the PDoC agreeable?

2013/11/7 Nat Sakimura <sakimura at gmail.com<mailto:sakimura at gmail.com>>
Proposed DoC (Disposition of Comment)

2013/11/6 Torsten Lodderstedt <torsten at lodderstedt.net<mailto:torsten at lodderstedt.net>>
Hi Mike,

here are my comments (some of them have already been discussed during the meeting on Sunday). Note: For your convenience, I also attached the word file containing my comments.

regards,
Torsten.

-----

Sections 2 & 4 describe a discovery process composed of three (mostly optional) steps (issuer discovery, create config URI, retrieve config data). I think it would be helpful to describe the whole process first, before diving into the details. This would support the reader to get guidance and allow to point out optional and mandatory parts of this spec.

section 2 heading - [already discussed] Isn't this section about the discovery of the OP's issuer URI? This should be spelled out. I got screwed since I thought the URI discovered here is directly used to retrieve config data. I may not be the only one.

2.1.1.

1. step - Why is XRI mention given all input must be a URI? I thought XRIs are a super-set of URIs.

Reject.
XRI defines particular shortcut syntax. We are special casing it here to ease the transition from OpenID 2.0.


2. step - Seems the URI must always contain an authority component? I think this should be stated explicitly.

Accept in principle.
Exact text needed.


2.1.2

This is stuff is hard to read and understand. Some example strings would be helpful for the reader. This would also lay a bridge to the next section.
Accept in principle.
Exact text needed.

"A string of any other type" - What ist he other type you refer to? Any other than a URI?

Accept in principle.
Perehaps rephrase it as
"A string not starting with =, @, !"


"2. If the userinfo component is present " - What about the host?

Accept in principle.
Add note explaining that for userinfo component is to be present, host component has to be present.

As a separate note, perhaps we should align the language "portion" and "component" as well as the order of the appearance in this paragraph.


3.

"issuer" - Is this value required to be the same as the issuer URI (possibly) discovered by the process described in section 2? [already discussed, yes]

Noted.


"authorization_endpoint
OPTIONAL. URL of the OP's Authentication and Authorization Endpoint" - This endpoint is also the OAuth 2.0 authz endpoint, this should be mentioned (as for the following parameter).

Accept in principle.
Add NOTE:.


"userinfo_endpoint"
"This URL MUST use the https scheme and MAY contain port, path, and query parameter components. " - I think this particular normative text should be given in the core, only. Otherwise we end up with normative text regarding a certain concept spread/copied over the OIDC document suite, which makes maintaining consistence more difficult.

Accept in principle.
Drop the second sentence. If the same does not appear in the Core, add them to the Core.


"scopes_supported
... The server MUST support the openid scope value. " - See above, this is not about discovery. I think it should at most be a note, but I would prefer a reference to the respective section in core.

Accept in principle.
Make it a NOTE as this is rather important to bring it up.


"response_types_supported
... Dynamic OpenID Providers MUST support the code, id_token, and the token id_token response type values." - Same here, could refer to the MTI section in core.

Accept.


grant_types_supported
... Dynamic OpenID Providers MUST support the authorization_code and implicit grant type values and MAY support the urn:ietf:params:oauth:grant-type:jwt-bearer grant type defined in JSON Web Token (JWT) Profile for OAuth 2.0 Client Authentication and Authorization Grants [OAuth.JWT]." - Same here. Moreover, I'm surprised this text mentions the JWT Bearer token profile. This is not even mentioned in core. What is the rationale?

Accept in principle.
Just refer the core.


"subject_types_supported
... Valid types include pairwise and public." - Are there other types envisioned?

Noted.
We are not precluding the possibilities.


"... The algorithm RS256 MUST be included. The value none MAY be supported, but MUST only be used with the Authorization Code Flow." - Normative text on non-discovery aspects again. I would suggest to make it a note or a reference instead.

Accept in principle.
Just reference Core as a normative text.
Add NOTE that RS256 is required by the Core avoiding normative language.


"request_object_encryption_alg_values_supported
... Authorization Server ..." - I would suggest to call it "OP" consistently.

Accept.
Replace Authorization Server with OP.


"token_endpoint_auth_methods_supported
OPTIONAL. JSON array containing a list of authentication methods supported by this Token Endpoint. The options are client_secret_post, client_secret_basic, client_secret_jwt, and private_key_jwt, ... " - Same here about normative text

Accept in principle.


"request_uri_parameter_supported
OPTIONAL. Boolean value specifying whether the OP supports use of the request_uri parameter, with true indicating support. If omitted, the default value is true. " - I'm wondering why the default is different for request object and reference. I would expect false for both of them.

Discuss.
The hisotrical reason is that request_uri  used to be in MTI.
Need to check if it is true still.


4.

"This step is OPTIONAL. The OpenID Provider endpoints and configuration information MAY be obtained out-of-band." - I'm wondering why this is optional as I consider this the main part of discovery. Or does it refer to derivation of the config URI based on the issuer URI? Then I would suggest to change section heading to something like "Deriving OpenID Provider Config URI". This might also require to push 4.1. to the top level.

Reject.
All the information obtained in this section may be obtained oob and not using discovery at all. However, it would be good to explain a little more about it.


".. Providers supporting discovery MUST support receiving WebFinger requests via TLS" - Is this statement misplaced. As it refers to WebFinger, it might belong to section 2?

 Accept.
Move to section 2.


_______________________________________________
Openid-specs-ab mailing list
Openid-specs-ab at lists.openid.net<mailto:Openid-specs-ab at lists.openid.net>
http://lists.openid.net/mailman/listinfo/openid-specs-ab



--
Nat Sakimura (=nat)
Chairman, OpenID Foundation
http://nat.sakimura.org/
@_nat_en



--
Nat Sakimura (=nat)
Chairman, OpenID Foundation
http://nat.sakimura.org/
@_nat_en



--
Nat Sakimura (=nat)
Chairman, OpenID Foundation
http://nat.sakimura.org/
@_nat_en
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-ab/attachments/20131107/098702e0/attachment-0001.html>


More information about the Openid-specs-ab mailing list