Handling authorization exception from server to client

[André Salvati]

Hi,

I've been working with GWT 2.4 and trying to send an authorization
exception from server to client whitout success.

1) Server:

A - Raise ReportableException throught report() on my
ServiceLayerDecorator.

B – Send the following JSON to the client:

{"S":[false],"I":[{"F":true,"M":"Não
autorizado","X":"com.google.web.bindery.requestfactory.server.ReportableException"}]}

2) Client:

AbstractRequestContext doesn't enter on processPayload() → if
(response.getGeneralFailure() != null) because it doesn't find key
“F” (but “F” is there, isn't??)


What is wrong?

Is it the right way to treat this kind of exception?

Thank you.

[Thomas Broyer]

"F" is present in "I", meaning one particular invocation (service method call) failed, not the whole "batch request" (well, actually, "S":[false] says the one and only invocation failed, then the object in "I" gives information about the failure).

FYI, the response format is defined by the com.google.web.bindery.requestfactory.shared.messages.ResponseMessage AutoBean.

[André Salvati]

So, wasn't it supposed to enter the code bellow on
AbstractRequestContext? Is not it a "general" failure?? Maybe this is
an issue with RequestFactory...

if (response.getGeneralFailure() != null) {
ServerFailureMessage failure = response.getGeneralFailure();
ServerFailure fail =
new ServerFailure(failure.getMessage(),
failure.getExceptionType(), failure
.getStackTrace(), failure.isFatal());

fail(receiver, fail);
return;

[Thomas Broyer]

No, it's not a "general failure", it's only a failure for one invocation. Look at few lines down in AbstractRequestContext:

      Set<Throwable> causes = null;

      for (int i = 0, j = invocations.size(); i < j; i++) {

        try {

          if (response.getStatusCodes().get(i)) {

            invocations.get(i).onSuccess(response.getInvocationResults().get(i));

          } else {

            ServerFailureMessage failure = AutoBeanCodex.decode(

                MessageFactoryHolder.FACTORY, ServerFailureMessage.class,

                response.getInvocationResults().get(i)).as();

            invocations.get(i).onFail(

                new ServerFailure(failure.getMessage(),

                    failure.getExceptionType(), failure.getStackTrace(),

                    failure.isFatal()));

          }

response.getStatusCodes().get(i) reads from the "S" (in your case, returns 'false', for i==0), and response.getInvocationResults().get(i) reads from the "I" (in your case, a ServerFailureMessage).

To be crystal clear:

requestContext.myMethod(arguments).to(new Receiver<X>() {

    @Override

    public void onSuccess(X response) {

        // won't get here

    }

    @Override

    public void onFailure(ServerFailureMessage error) {

        // will get there: the myMethod invocation failed; its Receiver's onFailure is called

    }

});

requestContext.fire(new Receiver<Void>() {

    @Override

    public void onSuccess(Void response) {

        // will get there: the myMethod invocation failed, but the whole "batch" succeeded: the context could have had other invocations that succeeded

    }

});

[André Salvati]

Ok Thomas, understood.

Thank you.