Hi Sergey,

please see inline.

On 25/05/2017 17:04, Sergey Beryozkin wrote:

Hi Pavel


On 25/05/17 15:54, Pavel Bucek wrote:

Hi Sergey, others,

you suggest that the exception mapper would be part of MVC spec (not the implementation, so not in javax.* package but more like org.impl.something.*?). I think that checking package name is fragile concept and should be avoided if possible, not to mention that this setup would be not ideal for MVC implementations.

I thought that the MVC spec was shipping its own default mapper, that is way I was suggesting checking javax.*.
if it is not the case then I agree it would not work well...

I believe we shouldn't introduce new annotations whenever there is a need for a marker (which, in this case, can be easily defined without it); we already do have plenty of them..

Also, note that @Priority is not use to control what is excluded. It is used accordingly to what it is supposed to be used - to sort providers. JAX-RS implementation then needs to pick single one in most occasions and it will naturally pick more important one.

Leading to the exclusion in the end. With Filters it is obviously not the case, all of them are run.

yes, exactly.

I was trying to offer another point of view on that issue, seemed like you are suggesting that the proposal uses @Priority in a way that contradicts its purpose.

Since I believe we have working proposal (backed up by the implementation),

IMHO that should not be used to affect the discussion given that the proposal is still under the discussion ?

I'm sorry if that statement felt like something too offensive.

All I wanted to say there was "hey, this proposal does work, we know it, since we do have backing implementation. with tests. :)"

I could use the argument that says "we need to have RI implementation for anything we do propose here, because if we don't, the feature cannot be part of the spec", since we have to have everything ready for PFD, but .. it is not the case here.

let's stick to following:

- clarify what is built-in and user-defined provider using Santiagos definition:

if it is registered using the JAX-RS API or discovered via class scanning (i.e. annotated with @Provider), then it is user-defined regardless of their physical location (library, container, etc.). Built-in ones should not be made available using those mechanisms.

- define that user-defined providers will be sorted by the @Priority with respect to existing, already defined algorithms. That means it will influence which Provider is defined only when other conditions are already applied and more than one provider is equal in terms of the defined algorithm (i.e. exception type distance for ExceptionMappers).

Described change fits well in Java EE environment, since using @Provider is quite common among other specifications thus is already natural for majority of users.

Please let us know whether you have any strong objections or suggestions how to improve proposed rules.

Looks like we have no other option but to use @Priority even though it will mean @Priority is used differently when applied to exception mappers (or param converters ?). It feels wrong but I guess we don't have other options (the new annotation will do well but it is just too late for it to be introduced)

I do see it as "contextual" priority and the context is defined by the provider definition. I don't have any examples from Java EE in mind right now, but sorting something (by the priority) and then picking "the best" provider seems to be natural. Also, if a provider is a filter and filter contract defines that all filters have to be executed, then using @Priority for sorting them doesn't feel wrong. I do agree that there is a difference, but the difference is in providers declaration, not in the semantics of @Priority.

Of course, I can be wrong ;) Please let me know what do you think.

Best regards,
Pavel

Sergey

Thanks and regards,
Pavel

On 25/05/2017 11:30, Sergey Beryozkin wrote:

I guess that what I also suggested when we talked with Christian about @DefaultProvider - the only problem from what I understood MVC will've been already released by the time JAX-RS 2.1 gets out.

To be honest, I'm still thinking, checking the origin (package) of the mapper can work. If for ex MVC ships its ExceptionMapper it is still a built-in mapper in context of MVC, and it can be identified to be a built-in one because it will be in a javax. namespace. Such providers should get a lower priority OOB (compared to the otherwise equal competing mappers), without having to depend on @Priority.

@Priority is there to sort the providers, making sure they execute in the right order. Using it as a mechanism to exclude does not seem quite right...Though it can work...

Cheers, Sergey