Re: [scim] Uniqueness Violation Response Code

[Phillip Hunt <[e-mail]>]

It seems to me that http protocol issues should not be in schema. 

However it is a MAY and thus 7644 overrides. 

I don’t think this can be handled as errata but rather a good item for a bis. 

Phil

> On Mar 12, 2020, at 12:29 PM, Shelley <[e-mail]> wrote:
> 
> 
> RFC 7643 Section 7 [1] indicates the following:
> 
>> uniqueness  A single keyword value that specifies how the service
>>    provider enforces uniqueness of attribute values.  A server MAY
>>    reject an invalid value based on uniqueness by returning HTTP
>>    response code 400 (Bad Request).
> 
> 
> This contradicts the requirements in RFC 7644 Section 3.3 [2] which indicates that service providers "MUST return HTTP status code 409 (Conflict)" when the creation of a requested resource would conflict with an existing resource, i.e. by violating an attribute's uniqueness. Therefore, servers MAY NOT reject invalid values based on uniqueness with a 400, as RFC 7643 Section 7 currently indicates.
> 
> I assume 409 is the preferred response here, and so I was initially going to report an errata that proposed 409 instead of 400 in RFC 7643 Section 7; however, to avoid dependencies on specific HTTP response codes from RFC 7644, I would propose the section could be updated as follows instead, such that the HTTP response code is omitted entirely:
> 
>> uniqueness  A single keyword value that specifies how the service
>>    provider enforces uniqueness of attribute values.  A server MAY
>>    reject an invalid value based on uniqueness.
> 
> 
> Any feedback is welcome. I'm happy to report the proposed errata if there is consensus; otherwise, I at least wanted to bring awareness to the contradiction between specs here.
> 
> [1] https://tools.ietf.org/html/rfc7643#section-7 
> [2] https://tools.ietf.org/html/rfc7644#section-3.3
> _______________________________________________
> scim mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/scim