Re: [OAUTH-WG] I-D Action: draft-ietf-oauth-token-exchange-07.txt

[Brian Campbell <[e-mail]>]

Following up on this, I'd like to propose a different and less invasive
change to the "actor_token" text. The new wording is below and not much
different than the text in the current draft. Barring any solid objections
to this in the next week or so, I'll publish -08 at which point I believe
the document will be ready for WGLC.

actor_token

OPTIONAL.  A security token that represents the identity of the acting
party. Typically this will be the party that is authorized to use the
requested security token and act on behalf of the subject.



On Thu, May 11, 2017 at 9:58 AM, Brian Campbell <[e-mail]>
wrote:

> The token exchange framework facilitates deployments like this one
> https://help.salesforce.com/articleView?id=remoteaccess_oaut
> h_asset_token_flow.htm or https://developer.box.com/docs
> /getting-started-with-new-box-view, for example, and I don't think pure
> plug and play interoperability is a realistic goal. The framework promotes
> interoperability in the form of common patterns and parameters that can be
> supported in libraries, products, and services.
>
> There's not one "other case" I have in mind but rather just broadening the
> text somewhat to more straightforwardly accommodate other cases.  One
> potential example is where the actor_token represents an authorizing party
> (again maybe needed for policy or auditing) to the token exchange event
> itself rather than the party that's having access rights assigned to it
> (implicitly with impersonation or explicitly with delegation).
>
>
>
> On Tue, May 9, 2017 at 9:55 AM, Denis <[e-mail]> wrote:
>
>> Brian,
>>
>> Even if Token Exchange is a framework, the goal is to be finally able to
>> interoperate.
>>
>> Whether we have one or two parameters, would you be able to provide a
>> precise semantics for the "other case" you have in mind ?
>>
>> Denis
>>
>> Yes, I omitted your comments in that post because I'd previously replied
>> to you in a separate message where I said that the "actor_token is a
>> security token so that's not an issue that needs to be addressed."
>> https://www.ietf.org/mail-archive/web/oauth/current/msg17247.html
>>
>> The other point you've just made about having very precise semantics for
>> a field is a fair one. However, I wanted to avoid introducing yet another
>> field (or really two fields b/c of the associated *_type for each inbound
>> token field), for what felt like a minor semantic variation that could be
>> easily accommodated by the existing framework, to the draft that already
>> has a lot of options and parameters on the request. And Token Exchange
>> really is a framework. I think that, to some extent, the framework is a bit
>> of a Rorschach test for deployers and implementers to utilize to solve
>> their specific issues and needs. I expect that will be the case regardless.
>> And I am proposing to somewhat genericize the text around one request
>> parameter to be more reflective of that.
>>
>> I would like to hear from others in the WG though.
>>
>> On Tue, May 9, 2017 at 3:06 AM, Denis <[e-mail]> wrote:
>>
>>> Brian,
>>>
>>> You omitted to include my comments in this post. So here it is again:
>>>
>>> ===========================================================
>>>
>>> The current text is:
>>>
>>> actor_token OPTIONAL. A security token that represents the identity of
>>> the party that is authorized to use the requested security token and act on
>>> behalf of the subject.
>>>
>>> This sentence is indeed wrong since an actor-token is not a security
>>> token.
>>>
>>> So your proposed change does not solve this issue: actor_token
>>> OPTIONAL.  A security token that represents the identity of the acting
>>> party.
>>>
>>> The current text states:
>>>
>>> Typically, in the request, the subject_token represents the identity of
>>> the party on behalf of whom
>>> the token is being requested while the actor_token represents the
>>> identity of the party to whom the access
>>> rights of the issued token are being delegated.
>>>
>>> Logically, the definition should be along the following lines:
>>>
>>>  actor_token OPTIONAL. Indicates the identity of the party to whom the
>>> access rights of the issued token are being delegated.
>>>
>>> If there is no delegation, then this field (which is optional) will not
>>> be used.
>>>
>>> ===========================================================
>>>
>>> I read your argumentation, but I maintain my comment. Each field should
>>> have a precise semantics.
>>>
>>> If you want to have another semantics, you should propose to define
>>> another field with its precise meaning.
>>>
>>> Denis
>>>
>>> Let me throw out a bit more context about this. The "actor_token" might,
>>> in a delegation scenario, represent the identity of the party to whom the
>>> access rights of the issued token are being delegated. That's the typical
>>> delegation scenario that is discussed in the draft. However, the
>>> "actor_token" might also be utilized/needed by the AS in an impersonation
>>> scenario for policy or auditing reasons even when the resulting issued
>>> token doesn't contain info about the delegation or actor. Similarly, the
>>> actor might not be strictly doing the impersonation but rather just be a
>>> party (again maybe needed for policy or auditing) to the token exchange
>>> event itself.  When I wrote the "actor_token" text in section 2.1 some ~18
>>> months ago I had the delegation scenario at the front of my mind and
>>> (clearly) intended to accommodate it. However, I didn't intend to limit it
>>> to only that and, looking at the text again, I think what is there now is
>>> too prescriptive and narrow. Thus my proposing to generalize the text
>>> somewhat.
>>>
>>>
>>>
>>>
>>> On Mon, May 8, 2017 at 10:29 AM, Brian Campbell <
>>> [e-mail]> wrote:
>>>
>>>> I do have one minor issue I'd like to raise that relates to some
>>>> conversations I've been a party to recently about implementations and
>>>> applications of token exchange.
>>>>
>>>> I think that the current text in §2.1 for the "actor_token" is overly
>>>> specific towards the delegation scenario. I'd propose the language be
>>>> generalized somewhat to allow more versatility in applications/deployments
>>>> of the token exchange framework. Here's that text:
>>>>
>>>>    actor_token
>>>>       OPTIONAL.  A security token that represents the identity of the
>>>>       acting party.
>>>>
>>>>
>>>>
>>>>
>>>> On Mon, May 8, 2017 at 8:01 AM, Rifaat Shekh-Yusef <
>>>> [e-mail]> wrote:
>>>>
>>>>> Hi All,
>>>>>
>>>>> The last email from Brian addresses the multiple audiences/resources
>>>>> issue with an error code, and we did not see any objection to this approach
>>>>> so far.
>>>>>
>>>>>
>>>>> *Authors,*
>>>>>
>>>>> Are there any other open issues with this draft?
>>>>> Do you believe it is ready for WGLC?
>>>>>
>>>>> Thanks,
>>>>>  Rifaat & Hannes
>>>>>
>>>>>
>>>>>
>>>>> On Fri, Mar 31, 2017 at 11:03 AM, Brian Campbell <
>>>>> [e-mail]> wrote:
>>>>>
>>>>>> As mentioned during the Chicago meeting the "invalid_target" error
>>>>>> code that was added in -07 was intended to give the AS a standard way to
>>>>>> reject request with multiple audiences/resources that it doesn't understand
>>>>>> or is unwilling or unable to process based on policy or whatever criteria .
>>>>>> It was intended as a compromise, of sorts, to allow for the multiple
>>>>>> resources/audiences in the request but provide an easy out for the AS of
>>>>>> saying it can't be supported based on whatever implementation or security
>>>>>> or policy it has.
>>>>>>
>>>>>> On Tue, Mar 28, 2017 at 1:32 AM, Nat Sakimura <[e-mail]>
>>>>>> wrote:
>>>>>>
>>>>>>> There are cases where tokens are supposed to be consumed at multiple
>>>>>>> places and the `aud` needed to capture them. That's why `aud` is a
>>>>>>> multi-valued field.
>>>>>>>
>>>>>>> On Mon, Mar 27, 2017 at 11:35 AM Torsten Lodderstedt <
>>>>>>> [e-mail]> wrote:
>>>>>>>
>>>>>>>> May I ask you to explain this reason?
>>>>>>>>
>>>>>>>> Am 27.03.2017 um 08:48 schrieb Mike Jones <
>>>>>>>> [e-mail]>:
>>>>>>>>
>>>>>>>> For the same reason that the “aud” claim is multi-valued in JWTs,
>>>>>>>> the audience needs to stay multi-valued in Token Exchange.  Ditto for
>>>>>>>> resources.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                                                        Thanks,
>>>>>>>>
>>>>>>>>                                                        -- Mike
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> *From:* OAuth [mailto:[e-mail]
>>>>>>>> <[e-mail]>] *On Behalf Of *Brian Campbell
>>>>>>>> *Sent:* Monday, March 27, 2017 8:45 AM
>>>>>>>> *To:* Torsten Lodderstedt <[e-mail]>
>>>>>>>> *Cc:* oauth <[e-mail]>
>>>>>>>> *Subject:* Re: [OAUTH-WG] I-D Action:
>>>>>>>> draft-ietf-oauth-token-exchange-07.txt
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Thanks for the review and question, Torsten.
>>>>>>>>
>>>>>>>> The desire to support multiple audience/resource values in the
>>>>>>>> request came up during a review and discussion among the authors of the
>>>>>>>> document when preparing the -03 draft. As I recall, it was said that both
>>>>>>>> Salesforce and Microsoft had use-cases for it. I incorporated support for
>>>>>>>> it into the draft acting in the role of editor.
>>>>>>>>
>>>>>>>> From an individual perspective, I tend to agree with you that
>>>>>>>> allowing for multiple audiences/resources adds a lot of complexity that's
>>>>>>>> like not needed in many (or most) cases. And I would personally be open to
>>>>>>>> making audience and resource mutual exclusive and single valued. A question
>>>>>>>> for the WG I suppose.
>>>>>>>>
>>>>>>>> The "invalid_target" error code that was added in -07 was intended
>>>>>>>> to give the AS a standard way to deal with the complexity and reject
>>>>>>>> request with multiple audiences/resources that it doesn't understand or is
>>>>>>>> unwilling or unable to process. It was intended as a compromise, of sorts,
>>>>>>>> to allow for the multiples but provide an easy out of saying it can't be
>>>>>>>> supported based on whatever implementation or policy of the AS.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On Sun, Mar 26, 2017 at 9:00 AM, Torsten Lodderstedt <
>>>>>>>> [e-mail]> wrote:
>>>>>>>>
>>>>>>>> Hi Brian,
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> thanks for the clarification around resource, audience and scope.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Here are my comments on the draft:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> In section 2.1 it states: „Multiple "resource" parameters may be
>>>>>>>> used to indicate
>>>>>>>>
>>>>>>>>       that the issued token is intended to be used at the multiple
>>>>>>>>
>>>>>>>>       resources listed.“
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Can you please explain the rational in more detail? I don’t
>>>>>>>> understand why there is a need to ask for access tokens, which are good for
>>>>>>>> multiple resources at once. This is a request type more or less exclusively
>>>>>>>> used in server to server scenarios, right? So the only reason I can think
>>>>>>>> of is call reduction.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On the other side, this feature increases the AS's complexity, e.g.
>>>>>>>> its policy may prohibit to issue tokens for multiple resources in general
>>>>>>>> or the particular set the client is asking for. How shall the AS handles
>>>>>>>> such cases?
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> And it is getting even more complicated given there could also be
>>>>>>>> multiple audience values and the client could mix them:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> "Multiple "audience" parameters
>>>>>>>>
>>>>>>>>       may be used to indicate that the issued token is intended to
>>>>>>>> be
>>>>>>>>
>>>>>>>>       used at the multiple audiences listed.  The "audience" and
>>>>>>>>
>>>>>>>>       "resource" parameters may be used together to indicate
>>>>>>>> multiple
>>>>>>>>
>>>>>>>>       target services with a mix of logical names and physical
>>>>>>>>
>>>>>>>>       locations.“
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> And in the end the client may add some scope values to the „meal“,
>>>>>>>> which brings us to
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> „Effectively, the requested access rights of the
>>>>>>>>
>>>>>>>>    token are the cartesian product of all the scopes at all the
>>>>>>>> target
>>>>>>>>
>>>>>>>>    services."
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> I personally would suggest to drop support for multiple audience
>>>>>>>> and resource parameters and make audience and resource mutual exclusive. I
>>>>>>>> think this is sufficient and much easier to implement.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> kind regards,
>>>>>>>>
>>>>>>>> Torsten.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Am 11.01.2017 um 20:04 schrieb Brian Campbell <
>>>>>>>> [e-mail]>:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Draft -07 of "OAuth 2.0 Token Exchange" has been published. The
>>>>>>>> primary change in -07 is the addition of a description of the relationship
>>>>>>>> between audience/resource/scope, which was a request or comment that came
>>>>>>>> up during the f2f meeting in Seoul.
>>>>>>>>
>>>>>>>> Excerpted from the Document History:
>>>>>>>>
>>>>>>>>    -07
>>>>>>>>
>>>>>>>>    o  Fixed typo (desecration -> discretion).
>>>>>>>>    o  Added an explanation of the relationship between scope,
>>>>>>>> audience
>>>>>>>>       and resource in the request and added an "invalid_target"
>>>>>>>> error
>>>>>>>>       code enabling the AS to tell the client that the requested
>>>>>>>>       audiences/resources were too broad.
>>>>>>>>
>>>>>>>> ---------- Forwarded message ----------
>>>>>>>> From: <[e-mail]>
>>>>>>>> Date: Wed, Jan 11, 2017 at 12:00 PM
>>>>>>>> Subject: [OAUTH-WG] I-D Action: draft-ietf-oauth-token-exchang
>>>>>>>> e-07.txt
>>>>>>>> To: [e-mail]
>>>>>>>> Cc: [e-mail]
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> A New Internet-Draft is available from the on-line Internet-Drafts
>>>>>>>> directories.
>>>>>>>> This draft is a work item of the Web Authorization Protocol of the
>>>>>>>> IETF.
>>>>>>>>
>>>>>>>>         Title           : OAuth 2.0 Token Exchange
>>>>>>>>         Authors         : Michael B. Jones
>>>>>>>>                           Anthony Nadalin
>>>>>>>>                           Brian Campbell
>>>>>>>>                           John Bradley
>>>>>>>>                           Chuck Mortimore
>>>>>>>>         Filename        : draft-ietf-oauth-token-exchange-07.txt
>>>>>>>>         Pages           : 31
>>>>>>>>         Date            : 2017-01-11
>>>>>>>>
>>>>>>>> Abstract:
>>>>>>>>    This specification defines a protocol for an HTTP- and JSON-
>>>>>>>> based
>>>>>>>>    Security Token Service (STS) by defining how to request and
>>>>>>>> obtain
>>>>>>>>    security tokens from OAuth 2.0 authorization servers, including
>>>>>>>>    security tokens employing impersonation and delegation.
>>>>>>>>
>>>>>>>>
>>>>>>>> The IETF datatracker status page for this draft is:
>>>>>>>> https://datatracker.ietf.org/doc/draft-ietf-oauth-token-exchange/
>>>>>>>>
>>>>>>>> There's also a htmlized version available at:
>>>>>>>> https://tools.ietf.org/html/draft-ietf-oauth-token-exchange-07
>>>>>>>>
>>>>>>>> A diff from the previous version is available at:
>>>>>>>> https://www.ietf.org/rfcdiff?url2=draft-ietf-oauth-token-exc
>>>>>>>> hange-07
>>>>>>>>
>>>>>>>>
>>>>>>>> Please note that it may take a couple of minutes from the time of
>>>>>>>> submission
>>>>>>>> until the htmlized version and diff are available at tools.ietf.org
>>>>>>>> .
>>>>>>>>
>>>>>>>> Internet-Drafts are also available by anonymous FTP at:
>>>>>>>> ftp://ftp.ietf.org/internet-drafts/
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> OAuth mailing list
>>>>>>>> [e-mail]
>>>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> OAuth mailing list
>>>>>>>> [e-mail]
>>>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> OAuth mailing list
>>>>>>>> [e-mail]
>>>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>>>>
>>>>>>> ...
>>
>> [Message clipped]
>
>
>