I agree with 3.1.2.4. Invalid Endpoint states:
If an authorization request fails validation due to a missing, invalid, or mismatching redirection URI, the authorization server SHOULD inform the resource owner of the error and MUST NOT automatically redirect the user-agent to the invalid redirection URI. But 4.1.2.1. Error Response states: If an authorization request fails validation due to a missing, invalid, or mismatching redirection URI, the authorization server SHOULD inform the resource owner of the error and MUST NOT automatically redirect the user-agent to the invalid redirection URI. It then defines error message formats with conditions that meet the above paragraph conditions. So which section defines how to handle an invalid Authorization request with missing or invalid elements, other than redirection URI? IBM states that Section 1.7 HTTP Redirection: This specification makes extensive use of HTTP redirections, in which the client or the authorization server directs the resource owner’s user-agent to another destination. While the examples in this specification show the use of the HTTP 302 status code, any other method available via the user-agent to accomplish this redirection is allowed and is considered to be an implementation detail. IBM claims that “any other method available via the user-agent to accomplish this redirection is allowed” in the section allows them to do anything they want. If that is true, why have an OAuth standard? Best regards, Don Donald F. Coffin Founder/CTO REMI Networks 2335 Dunwoody Crossing Suite E Dunwoody, GA 30338-8221 From: Max Gerber <[email protected]> Sent: Friday, August 7, 2026 6:53 PM To: [email protected] Cc: Warren Parad <[email protected]>; oauth <[email protected]>; Jeremy J. Roberts <[email protected]> Subject: Re: [OAUTH-WG] Re: How to properly implement Section 4.1.2.1 Authorization Error Response for OAuth 2.x Two sections define Authorization endpoint errors. 4.1.2.1, which describes errors directed at the client, and § 3.1.2.4 <https://datatracker.ietf.org/doc/html/rfc6749#section-3.1.2.4> , which describes errors directed at the user. Case 4 falls under §3.1.2.4, not § 4.1.2.1 <http://4.1.2.1> : If an authorization request fails validation due to a missing, invalid, or mismatching redirection URI, the authorization server SHOULD inform the resource owner of the error and MUST NOT automatically redirect the user-agent to the invalid redirection URI. So the authorization server _can_ respond however it wants, so long as it informs the user of an error and does not redirect the user-agent back to the client. The 401 HTML response from IBM does this, and so is valid. On Fri, Aug 7, 2026 at 3:40 PM <[email protected] <mailto:[email protected]> > wrote: Hi Warren, Yes. Section 4.1.2.1 does not specifically state that a 401 is how a server should respond. Best regards, Don Donald F. Coffin Founder/CTO REMI Networks 2335 Dunwoody Crossing Suite E Dunwoody, GA 30338-8221 From: Warren Parad <[email protected] <mailto:[email protected]> > Sent: Friday, August 7, 2026 6:22 PM To: [email protected] <mailto:[email protected]> Cc: Emelia S. <[email protected] <mailto:[email protected]> >; oauth <[email protected] <mailto:[email protected]> >; Jeremy J. Roberts <[email protected] <mailto:[email protected]> > Subject: Re: [OAUTH-WG] How to properly implement Section 4.1.2.1 Authorization Error Response for OAuth 2.x It goes both ways, unless forbidden, it might as well be allowed... So which scenario do you see you are running up against, where the RFC explicitly forbids a 401 response? On Sat, Aug 8, 2026, 00:15 <[email protected] <mailto:[email protected]> > wrote: Warren, Thanks for your quick reply. What Section of RFC 6749 allows a 401 status code response for an Authorization Request? I am confused by the response being correct, since all OAuth Authorization Requests are 302/303 Redirect responses. Best regards, Don Donald F. Coffin Founder/CTO REMI Networks 2335 Dunwoody Crossing Suite E <https://www.google.com/maps/search/2335+Dunwoody+Crossing+Suite+E+Dunwoody,+GA+30338-8221?entry=gmail&source=g> Dunwoody, GA 30338-8221 <https://www.google.com/maps/search/2335+Dunwoody+Crossing+Suite+E+Dunwoody,+GA+30338-8221?entry=gmail&source=g> From: Warren Parad <[email protected] <mailto:[email protected]> > Sent: Friday, August 7, 2026 3:32 PM To: [email protected] <mailto:[email protected]> Cc: Emelia S. <[email protected] <mailto:[email protected]> >; [email protected] <mailto:[email protected]> ; Jeremy J. Roberts <[email protected] <mailto:[email protected]> > Subject: Re: [OAUTH-WG] How to properly implement Section 4.1.2.1 Authorization Error Response for OAuth 2.x To be clear, regarding the your questions: Does the IBM Verify Authorization Server properly implement Section 4.1.2.1. Authorization Error Response per the OAuth standard. Yes. Are you stating these are valid OAuth 2.0 Authorization Flow responses when an OpenID Connect server should be acting like an OAuth 2.0 server without OpenID Connect capability? No, these are valid OAuth 2.0 Authorization Flow responses. Is there a hierarchical relationship between OpenID Connect and the IETF RFC 6749 OAuth 2.0 standard? No, there is no relationship. On Fri, Aug 7, 2026 at 9:13 PM <[email protected] <mailto:[email protected]> > wrote: Good afternoon, Emilia and Warren. Thanks for your quick responses. Please clarify which table of results you reference: the first table (expected response) or the second table (IBM Verify responses). If the second table, are you stating these are valid OAuth 2.0 Authorization Flow responses when an OpenID Connect server should be acting like an OAuth 2.0 server without OpenID Connect capability? Can an OpenID Connect server support both OpenID Connect Authentication and perform OAuth 2.0 Authorization? The issue is that we test for compliance with the NAESB ESPI standard, not the OpenID Connect standard, which the ESPI standard does not support. Therefore, the GBA Certification Platform expects the Authorization Server to support the OAuth Authorization Flow requirements. Since I am not familiar with the OpenID Connect specification, is an OpenID Connect solution capable of supporting an OAuth Authorization Flow? Is there a hierarchical relationship between OpenID Connect and the IETF RFC 6749 OAuth 2.0 standard? According to IBM, the OpenID Connect specification supersedes the IETF RFC 6749 OAuth 2.0 standard and should therefore the OAuth 2.0 implementation should support the OpenID Connect Authorization Flow specification. In reviewing the OpenID Connect specification, the Authorization Flow section states that for an OpenID Connect Hybrid Flow, the Authorization Flow should follow the OAuth 2.0 standard. Does this indicate there is no hierarchical relationship between the OpenID Connect and OAuth 2.0 standards? Best regards, Don Donald F. Coffin Founder/CTO REMI Networks 2335 Dunwoody Crossing Suite E <https://www.google.com/maps/search/2335+Dunwoody+Crossing+Suite+E+Dunwoody,+GA+30338-8221?entry=gmail&source=g> Dunwoody, GA 30338-8221 <https://www.google.com/maps/search/2335+Dunwoody+Crossing+Suite+E+Dunwoody,+GA+30338-8221?entry=gmail&source=g> From: Emelia S. <[email protected] <mailto:[email protected]> > Sent: Thursday, August 6, 2026 1:23 PM To: [email protected] <mailto:[email protected]> Cc: [email protected] <mailto:[email protected]> ; Jeremy J. Roberts <[email protected] <mailto:[email protected]> > Subject: Re: [OAUTH-WG] How to properly implement Section 4.1.2.1 Authorization Error Response for OAuth 2.x Hi Donald, Looking at the table of results and test cases, for 1, 3 and 4, these are the correct responses per the results table, as otherwise the authorization server would be an open redirector. For test case 0, the response_type is required, so yes, redirecting back with the error of either invalid_request, or unsupported_response_type if it's present but unsupported, would be correct. For test case 5, redirecting back with invalid_scope would be correct, provided the client and redirect URI match. Do be careful not to mixup response_type (OAuth, OIDC) and response_mode (from OIDC), response_mode can often be inferred from response_type (e.g., code response_type often defaults to query as a response_mode). Hope that helps. Yours, Emelia On 6 Aug 2026, at 18:30, [email protected] <mailto:[email protected]> wrote: I am the Technical Manager of the Green Button Alliance. I am responsible for certifying that electric, natural gas, and water utilities comply with the North American Energy Standards (NAESB) Energy Service Provider Interface (ESPI) standard and need a ruling from the IETF OAuth WG regarding an issue that has arisen for Hydro One Networks, Inc. Hydro One Networks, Inc., the largest electric utility in Ontario, Canada, implemented the NAESB ESPI (a.k.a. Green Button) standard in 2023, which uses the OAuth 2.0 standard for customers to authorize client applications to access their energy data. At the time, they were using an IBM Tivoli Authorization Server, which fully complied with the Green Button Alliance legislatively mandated certification test for Section 4.1.2.1. Authorization Error Response. They recently upgraded to the IBM Verify Authorization Server, which no longer passes the same Section 4.1.2.1. Authorization Error Response tests. IBM has agreed to comply with the IETF OAuth Working Group’s response to my questions. The GBA Certification test platform currently submits the following invalid Authorization Endpoint requests and expects the Authorization Server to comply with Section 4.1.2.1. Authorization Error Response error messages: iTestID Label Test Name Query string Status Code Expected Error Message 0 Case 0: no response_type Malformed Authorization Code Request (No response_type field-value) ?client_id=(client_id)&redirect_uri=(redirect_uri)&scope=(scope)&state=ee148ec2-9898-466e-a013-4dedf1eb3018 302/303 ?error=invalid_request parameter and no or empty location header 1 Case 1: no client_id Malformed Authorization Code Request (No client_id field-value) ?response_type=code&redirect_uri=(redirect_uri)&scope=(scope)&state=ee148ec2-9898-466e-a013-4dedf1eb3018 302/303 ?error=invalid_request parameter and no or empty location header 2 Case 2: invalid response_type Malformed Authorization Code Request (Invalid response_type field-value: ?response_type=badresponsetype& client_id=(client_id)&redirect_uri=(redirect_uri)&scope=(scope)&state=ee148ec2-9898-466e-a013-4dedf1eb3018 302/303 ?error=unsupported_response_type parameter and no or empty location header 3 Case 3: invalid client_id Malformed Authorization Code Request (Invalid client_id field-value) ?response_type=code& client_id=invalid_third_party&redirect_uri=(redirect_uri)&scope=(scope)&state=ee148ec2-9898-466e-a013-4dedf1eb3018 302/303 ?error=unauthorized_client parameter and no or empty location header 4 Case 4: invalid redirect_uri Malformed Authorization Code Request (Invalid redirect_uri field-value) ?response_type=code& client_id=(client_id)&redirect_uri= <https://services.greenbuttondata.org&scope=(scope)&state=ee148ec2-9898-466e-a013-4dedf1eb3018/> https://services.greenbuttondata.org&scope=(scope)&state=ee148ec2-9898-466e-a013-4dedf1eb3018 302/303 ?error=invalid_request parameter and no or empty location header 5 Case 5: invalid scope Malformed Authorization Code Request (Invalid scope field-value) ?response_type=code& client_id=(client_id)&redirect_uri=(redirect_uri)&scope=FB=????&state=ee148ec2-9898-466e-a013-4dedf1eb3018 302/303 ?error=invalid_scope parameter and no or empty location header All scope string values are application variable contents shown as () entries if present. Malformed values are shown as strings. I do not test for missing state= elements, although the NAESB ESPI standard requires them. The following table contains the current response by the IBM Verify Authorization Server for each of the above test cases: iTestID Status Code Response 0 303 <https://cmdcert.greenbuttonalliance.org:8445/ThirdParty/espi/1_1/OAuthCallBack> https://cmdcert.greenbuttonalliance.org:8445/ThirdParty/espi/1_1/OAuthCallBack?error=unsupported_response_type&error_description=... 1 401 The response included a full-screen HTML showing the error “CSIAQ0154E The requested OAuth 2.0 Client does not exist” 2 303 <https://cmdcert.greenbuttonalliance.org:8445/ThirdParty/espi/1_1/OAuthCallBack> https://cmdcert.greenbuttonalliance.org:8445/ThirdParty/espi/1_1/OAuthCallBack?error=unsupported_response_type&error_description... 3 401 The response included a full-screen HTML showing the error “CSIAQ0154E The requested OAuth 2.0 Client does not exist” 4 400 The response included a full-screen HTML showing the error “CSIAQ0167E The redirection URI provided in the request is either invalid or does not match any of the OAuth 2.0 Client’s pre-registered redirect URLs” 5 302 <https://...verify.ibm.com/idaas/mtfim/sps/idaas/login> https://...verify.ibm.com/idaas/mtfim/sps/idaas/login... Note: <https://cmdcert.greenbuttonalliance.org:8445/ThirdParty/espi/1_1/OAuthCallback> https://cmdcert.greenbuttonalliance.org:8445/ThirdParty/espi/1_1/OAuthCallback is the value of the redirect_uri for all but the negative redirect_uri tests The IBM team indicated that case 5 (invalid scope): The OpenID Connect spec allows ignoring unknown scopes vs. returning an invalid_scope error ( <https://openid.net/specs/openid-connect-core-1_0.html#AuthRequest> https://openid.net/specs/openid-connect-core-1_0.html#AuthRequest). The referenced section of the OpenID Connect standard is the Authentication section, which is NOT what case 5 is testing. However, Section 3.1.2.2. Authentication Request Validation of the OpenID Connect specification indicates the Authorization Server MUST validate all the OAuth 2.0 parameters according to the OAuth 2.0 specification. That includes (2) verifying that a scope parameter is present and contains the openid scope value (If no openid scope value is present, the request may still be a valid OAuth 2.0 request but is not an OpenID Connect request). The IBM team indicated that case 4 (invalid redirect): An OpenID Connect certification has precedence over the IETF OAuth 2.0 RFC 6749 and referenced their official OpenID Foundation cert test for invalid redirect URI, which expects an error page, not a redirect ( <https://www.certification.openid.net/log-detail.html?log=Lbzq5uCWSM1Sij3&public=true> https://www.certification.openid.net/log-detail.html?log=Lbzq5uCWSM1Sij3&public=true). The NAESB ESPI standard does not support the OpenID Connect standard, so we contend that the IBM Verify team’s position that their OpenID Connect server does not need to comply with the IETF RFC 6749 and 6750 standards is a “red herring”. For cases 0 and 2, they have demonstrated the ability to support Section 4.1.2.4 Authorization Error Response properly. The business issue and why I am submitting this email is that Hydro One Networks, Inc. is required by Ontario legislation to be certified by the Green Button Alliance as being in compliance with the NAESB ESPI standard. Without that certification, Hydro One Networks, Inc. will face continuing financial penalties from the Ontario Energy Board. Best regards, Don Donald F. Coffin Founder/CTO REMI Networks 2335 Dunwoody Crossing Suite E <https://www.google.com/maps/search/2335+Dunwoody+Crossing+Suite+E+Dunwoody,+GA+30338-8221?entry=gmail&source=g> Dunwoody, GA 30338-8221 <https://www.google.com/maps/search/2335+Dunwoody+Crossing+Suite+E+Dunwoody,+GA+30338-8221?entry=gmail&source=g> _______________________________________________ OAuth mailing list -- <mailto:[email protected]> [email protected] To unsubscribe send an email to <mailto:[email protected]> [email protected] _______________________________________________ OAuth mailing list -- [email protected] <mailto:[email protected]> To unsubscribe send an email to [email protected] <mailto:[email protected]>
_______________________________________________ OAuth mailing list -- [email protected] To unsubscribe send an email to [email protected]
