On Aug 20, 2009, at 4:17 PM, Bill Burke wrote:
>
>
> Paul Sandoz wrote:
>> On Aug 19, 2009, at 7:33 PM, Bill Burke wrote:
>>> I'm doing an EJB example in my book and came across the scenario
>>> of EJBException. Maybe the spec should define a default
>>> ExceptionMapper for it? Here's a simple one I did:
>>>
>>> @Provider
>>> public class EJBExceptionMapper implements
>>> ExceptionMapper<EJBException>
>>> {
>>> @Context
>>> private Providers providers;
>>>
>>> public Response toResponse(EJBException exception)
>>> {
>>> if (exception.getCausedByException() == null)
>>> {
>>> return Response.serverError().build();
>>> }
>>> Class cause = exception.getCausedByException().getClass();
>>> ExceptionMapper mapper = providers.getExceptionMapper(cause);
>>> if (mapper == null)
>>> {
>>> return Response.serverError().build();
>>> }
>>> else
>>> {
>>> return mapper.toResponse(exception.getCausedByException());
>>> }
>>> }
>>> }
>> Note that the above solution assumes that by default there is an
>> exception mapper for WebApplicationException. Only instances
>> WebApplicationException that have no entity should be mapped:
>
> I am not following you. Don't see where WEbApplicationException
> comes into play here.
>
>> https://jsr311.dev.java.net/nonav/releases/1.0/spec/
>> spec3.html#x3-280003.3.4 Thus one cannot always assume that
>> providers.getExceptionMapper for a cause class that is assignable
>> to WebApplication will return a non-null value.
>
> Ah, you're saying that if the cause is a WebApplicationException I
> need to return the Response of the WAE if it exists.
>
Almost: if the cause is a WebApplicationException whose Response has
an entity then the Response must be returned. Otherwise, a mapper can
be obtained and if that mapper is null then the Response must be
returned.
I think this probably highlights some improvements we could make to
the specification of providers.getExceptionMapper.
Paul.