Replies: 4 comments
|
I think we previously discussed that standard error responses for OAuth2 do not have to be reformated to Error Info, but indeed we do not have it documented in the guidelines. Possible solution is layered approach:
Could it be easily distinguished with diffrent implementation architectures and in compliance tests? |
|
From Telefónica's perspective, we advocate for maintaining and explicitly documenting the current CAMARA requirement, which states that 401 and 403 errors must include the JSON error body specified by CAMARA. Our position is based on the following key arguments:
It is worth noting that Telefónica’s implementation currently complies with both requirements: we provide the CAMARA-defined JSON body and adhere to RFC 6750 (including the e.g.
Proposal: CC @PedroDiez |
|
Hi @jpengar , Thanks again for the response and for continuing to follow this topic. From my perspective, this is not only about technical possibility. It is mainly about whether this is desirable in deployments where token handling is separated from application logic. Making a JSON body mandatory for token-validation failures means introducing API-specific behavior into the gateway/security layer. Even if that is technically possible, it is not always wanted, and in some environments it can also be limited by security or operational governance. I can see why keeping this consistent for CAMARA consumers is important, especially if certification and existing integrations already depend on it. My point is just that there is also a broader interoperability and implementation tradeoff when protected-resource token handling moves away from the RFC 6750 model that many gateways and integrations already rely on. The main point for me is keeping a clear separation between token handling and application/API behavior. For protected-resource calls using Bearer access tokens, RFC 6750 already provides a well-known model for authentication failures through the HTTP status code and the WWW-Authenticate challenge. Keeping the JSON body optional, or at least not assuming it for token handling, makes that separation easier to maintain. I also think the impact is broader than interoperability between API providers alone. It affects integrators as well, because it influences whether CAMARA APIs can continue to follow the same protected-resource patterns and tooling that are already common across other OAuth-protected APIs. The more specific CAMARA becomes in this area, the more additional handling is needed around something that is otherwise already standardized. My preferred direction would still be the layered approach suggested by @rartych . If the group prefers to keep the current requirement, I think it would still help to document this more explicitly as an implementation risk, because this kind of boundary can be easy to overlook, especially when token handling and API implementation sit in different components or teams. |
|
I think it is relevant to have other views within the community, this topic has several important implications and would be needed as much consensus as possible |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Description
Context
During the ICM meeting on April 8th, 2026, we discussed the overlapping between RFC 6750 and CAMARA Commonalities regarding authentication error handling (401/403).
Current situation
WWW-Authenticateheader in 401/403 responses) in the protected resource request.→ API providers MUST support the CAMARA-defined JSON error body.
WWW-Authenticate: Bearerheader.Identified challenge
At the same time:
Relaxing the requirement could introduce ambiguity for API consumers, who would not know whether to expect:
ICM conclusion
No change to current requirements was agreed.
The issue is acknowledged as a valid implementation challenge.
Further discussion is needed on whether:
Ask to Commonalities
Since error response formats are defined in Commonalities, this topic requires joint coordination between ICM and Commonalities.
Specifically, we need to clarify:
Status
Pending further discussion between ICM and Commonalities.
All reactions