Hi All: According to Kathy's "what we don't need to know" post, chapter 19 of the spec titled "Support for distribution and Interoperability" can be ignored. I recently came across a question in one of the test simulators that required knowledge of a point found in that chapter. The question asked to pick the true statements from a set of statements. One of the choices was as follows:
EJB containers are required to be able to publish EJBHome object references in a CORBA CosNaming service.
In section 19.7 of the spec I found the same quote:
EJB containers are required to be able to publish EJBHome object references in a CORBA CosNaming service [ 15 ].
If we can ignore chapter 19 as Kathy suggests, then I assume that this knowledge would not be required for the real exam, or is it? I would hate to lose points for not studying something that I am willing to study if needed. [ February 05, 2004: Message edited by: Keith Rosenfield ] [ February 05, 2004: Message edited by: Keith Rosenfield ]
Howdy -- we deliberatebly left the interoperability chapter in the spec *out* of the exam objectives, and if it's not on the objectives, there is almost *no* chance of getting any specific questions from that section. You ARE expected to know RMI-IIOP can be the wire protocol used by the server, which is why YOU should always do a narrow(), *and* you're expected to know that your arguments and return types must be RMI-IIOP-compliant, not just RMI compliant. However, the differences between legal RMI-IIOP types and RMI types are *not* covered in the exam because they're not actually *in* the spec and you're *very* unlikely to have problems with the differences in the real world. (and if you're making non-Java CORBA clients for your beans, you'll know it. ) So, if it's not on the real objectives, it's *extremely* unlikely that a question on this has snuck through. The main reason, though, that we have left this out is because these interoperability requirements are *not* absolute guarantees for a Container. They are requirements *only* for those vendors which choose to provide this level of interoperability. In other words, it's like the spec is saying, "You are not required to do this unless you support interoperability. But if you DO support interoberability, here's the way you must do it..." Which leads back to why *you* must assume that your Container is using RMI-IIOP, even if it is using only plain old RMI. You have to assume it because the Container is allowed -- but not required -- to support it. So, this is kind of a fuzzy point -- the stuff in that chapter -- but all you need to know about CORBA on the exam is that it *might* be supported by your vendor, and that you have to use RMI-IIOP, and that to assume RMI-IIOP means that we have to do the narrow(). Now, if the chapter on runtime enviornment (the one that covers the APIs that must be supported by EJB) had specified this, then it would be fair game for the exam, but these requirements specify only JNDI, not COSNaming. Hope that helps Cheers, Kathy
We find this kind of rampant individuality very disturbing. But not this tiny ad: