It would be great if the spec leads could join this discussion.
Zitat von Bill Burke <bburke_at_redhat.com>:
> On 3/1/2016 12:00 PM, markus_at_headcrashing.eu wrote:
>> Bill,
>>
>> no I did not miss that example. But -as many specs do- I assume
>> that examples are non-normative but solely illustrative.
>>
>
> You've never dealt with the TCK :)
>
>> Actually neither the spec nor the JavaDocs explicitly say that root
>> resources MUST have @Path. In fact, Jersey deploys a root resource
>> WITHOUT having @Path. That's the actual reason why I am asking:
>> What DOES Jersey do in that case (it does NOT complain at least),
>> and how MUST a JAX-RS implementation deal with @Path-free root
>> resources?
>>
>
> Jersey is just messed up then. How is the annotation deployer
> supposed to know the difference between a root resource and a sub
> resource locator? You are supposed to be able to have your WAR
> automatically scanned without having to explicitly declare your
> services in your application.
>
>> In fact I don't understand why we cannot simply add a short line to
>> the spec instead of discussion: "An un-annotated root resource
>> class MUST be treated as if @Path("") would be existing at the
>> class". That would be unambiguously make clear how Jersey, RestEasy
>> and CXF MUST deal with that case.
>>
>
> I think this is incorrect.
>
> --
> Bill Burke
> JBoss, a division of Red Hat
> http://bill.burkecentral.com