[
https://issues.apache.org/jira/browse/CXF-6307?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14394579#comment-14394579
]
Sergey Beryozkin edited comment on CXF-6307 at 4/3/15 3:38 PM:
---------------------------------------------------------------
Just a short summary based on the analysis/conversations I've done so far.
- MBR can depend on the declarative Consumes support or do the checks
dynamically in its isReadable. Both options are valid/compliant. Either option
can be used independently or both options can be combined.
- MBR can provide a valid implementation of "Boolean/etc for text/plain only"
by using either declarative Consumes or dynamic checks. Spec can not/does not
enforce which approach is used to implement a given requirement
- This test is technically valid
- This test should not be at the CTS level but at a concrete JAX-RS
implementation level (example, at RI level) because it was written with the
assumption that a default MBR uses a declarative approach as opposed to a
dynamic check approach.
Base on the above, but also on the fact that having all of CXF default
providers (many of them with wildcard Consumes as per the spec) will slowdown a
selection process but also be a very sensitive operation in the short term, my
recommendation is to challenge this test.
I'm not opposed in principle to refactoring ProviderFactory and hence I will
keep this issue open and will review how this can be addressed. But given that
we have major releases coming in now, and also because it is very important
that a validity of MBRs doing the dynamic checks is enforced, IMHO the faster
solution to getting behind this test is to challenge it
Cheers, Sergey
was (Author: sergey_beryozkin):
Just a short summary based on the anslysis/conversations I've done so far.
- MBR can depend on the declarative Consumes support or do the checks
dynamically in its isReadable. Both options are valid/compliant. Either option
can be used independently or both options can be combined.
- MBR can provide a valid implementation of "Boolean/etc for text/plain only"
by using either declarative Consumes or dynamic checks. Spec can not/does not
enforce which approach is used to implement a given requirement
- This test is technically valid
- This test should not be at the CTS level but at a concrete JAX-RS
implementation level (example, at RI level) because it was written with the
assumption that a default MBR uses a declarative approach as opposed to a
dynamic check approach.
Base on the above, but also on the fact that having all of CXF default
providers (many of them with wildcard Consumes as per the spec) will slowdown a
selection process but also be a very sensitive operation in the short term, my
recommendation is to challenge this test.
I'm not opposed in principle to refactoring ProviderFactory and hence I will
keep this issue open and will review how this can be addressed. But given that
we have major releases coming in now, and also because it is very important
that a validity of MBRs doing the dynamic checks is enforced, IMHO the faster
solution to getting behind this test is to challenge it
Cheers, Sergey
> Wrong select the message body reader
> ------------------------------------
>
> Key: CXF-6307
> URL: https://issues.apache.org/jira/browse/CXF-6307
> Project: CXF
> Issue Type: Bug
> Components: JAX-RS
> Affects Versions: 3.0.3
> Environment: Windows
> Reporter: Neal Hu
> Fix For: 3.0.5
>
>
> This is a CTS testcase, the resource class is like this:
> @POST
> @Path("boolean")
> public Boolean postBoolean(Boolean bool) {
> if(bool){
> throw new WebApplicationException(Status.NOT_ACCEPTABLE);
> }
> return false;
> }
> The application provided provider is like this:
> public class MyReader implements MessageBodyReader<Boolean> {
> @Override
> public boolean isReadable(Class<?> type, Type type1, Annotation[] antns,
> MediaType mt) {
> return type== Boolean.class;
> }
> @Override
> public Object readFrom(Class<Object > type,
> Type type1,
> Annotation[] antns,
> MediaType mt, MultivaluedMap<String, String> mm,
> InputStream in) throws IOException,
> WebApplicationException {
> return Boolean.valueOf("true");
> }
> }
> The request
> Content-Type:text/plain
> Accept:text/plain
> Method:POST
> Body:false
> According to jsr339 section 4.2.4 Standard Entity Providers
> java.lang.Boolean, java.lang.Character, java.lang.Number Only for text/plain.
> Corresponding primitive types supported via boxing/unboxing conversion.
> An implementation MUST support application-provided entity providers and MUST
> use those in preference to its own pre-packaged providers when either could
> handle the same request. More precisely, step 4 in Section 4.2.1 and step 5
> in Section 4.2.2 MUST prefer application-provided over pre-packaged entity
> providers.
> 4.2.1 Message Body Reader
> 3. Select the set of MessageBodyReader classes that support the media type of
> the request, see Section
> 4.2.3.
> 4. Iterate through the selected MessageBodyReader classes and, utilizing the
> isReadable method of
> each, choose a MessageBodyReader provider that supports the desired Java type.
> It says the entity providers should be sorted by media type firstly then(the
> media type is the same) on Section 4.2.1 and step 5 in Section 4.2.2 MUST
> prefer application-provided over pre-packaged entity providers. So
> org.apache.cxf.jaxrs.provider.PrimitiveTextProvider should be selected as its
> media type MUST be text/plain and the application provider's media type is
> */*.(x/y>x/*>*/*)
> So at this point the CTS expected response code is 200 instead of 406, the
> application provided provider should not be chosen.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
