Kevin Yip

Ranch Hand
+ Follow
since Oct 17, 2000
Merit badge: grant badges
For More
Cows and Likes
Total received
In last 30 days
Total given
Total received
Received in last 30 days
Total given
Given in last 30 days
Forums and Threads
Scavenger Hunt
expand Ranch Hand Scavenger Hunt
expand Greenhorn Scavenger Hunt

Recent posts by Kevin Yip

Actually I've read that post mentioned above. First of all I think it barely constitutes a breach of the agreement (even if the candidate is bound by that agreement). Even if he/she is, by all means it should be determined by the point of law, not by instinct, as far as only an extract of a small part with totally different wording is concerned.
The quote by the moderator appears in 2 places as far as I can found - one in "SUN CERTIFICATION PROGRAM POLICY ON CANDIDATE MISCONDUCT"(the Policy), another in the "Pre-Test Agreement"(the Agreement). As you need to click to accept the Agreement only when you do the Test in an examination center, whether the Agreement is applicable to part 2 assignment is doubtful. We never need to click any button before downloading the assignment from galton website. So the following discussion only investigates whether the part 2 assignment is bound by the Policy.
Since it's more accurate to interpret the meaning of any single sentence or paragraph in a wider context, I try to interpret its meaning by reading the whole Policy. I believe the main rationale of the 'Pre-Test Agreement' and the Policy is clarified in a paragraph quoted from the Policy as below:
"The credential conferred by the Certification Program delivers its greatest value to you only as long as the Test is secure and continues to operate as a measure of personal knowledge and skill. Sun requires every candidate for Sun Certification to assist in maintaining the integrity and security of the Tests."
And the quote by the moderaor is actually only part of a larger paragraph, as follows:
"The content of all tests and test materials are proprietary and confidential information of Sun. Candidates may not disclose or share any of the content of these materials; nor may candidates ask for, write (in any media), publish, or otherwise disclose any test questions or answers."
Both of these confirm my assertion that Sun mainly concerns about the Test questions they designed for use in the examination centre. That's the reason why we only need to click the button when we do the test in the exam center. If these questions are released, the credential of the certificate will be in doubt.
For the part 2 assignment: Firstly there is no time-limit. Secondly, you can submit the assignment only when you paid Sun the fee, not at your wish. As such, even if a candidate get the assignment earlier it'll never DEVALUE the certificate as he/she is given an UNLIMITED time in the first place!
We understand it's the moderator's prerogative to delete any threads which he/she deems in breach of the javaranch policy. However whether it's actually the situation in this case I'm really in doubt. And the quality and popularity of this kind of forum to a certain extent depend on whether the moderator can exercise this discretion wisely. Being said so, alas... I agree myself that the easiest way to resolve a controversial issue like this is to remove it before it matures. Anyway, let's improve ourselves along this line.
[ April 28, 2002: Message edited by: Kevin Yip ]
In class diagram:
A dashed line has a open triangle on one end is implementation or realization
A solid line has a open triangle on one end is generalization
A dashed line has an arrow on one end is dependency
I'm not living in China now, but very glad to hear that there are more than 20 SCEAs in China. We better catch up!
May be I'm wrong, but I think in JDK 1.1 a signed applet enjoys unlimited access like a local application provided the correspeonding public key is trusted in the executing environment.
So the only difference between applet's access to the system in 1.1 and 1.2 is that in 1.2 the 'security principle of least privilege' is upheld by the introduction of policy file. As such, none of the 4 options provided is correct.
Network Variable (NV) are high-level objects that application devices use to communicate with one another.
But I'm not sure I understand the question. Generally the amount of network traffic is caused or determined by the application design or application requirement of data flow, not by the protocols which only provide the infrastructure for these data flow.
Should be (3) through the Home Interface.
I live in Toronto and I can't find any in all big bookstores here. So I ordered it from Amazon about 1 week ago and got it 3 days ago.
Peter Kao has a very detail study journal for Sun's PetStore. Take a look!
Architecture is both an art and a science. To be an architect has to master both of these.
Thus, what distingush an architect from a developer is his/her breadth of knowledge and his/her artistic talent.
Take the meaning of the word 'artistic' broadly to include handling creatively all kinds of situations like dealing with clients, teams, contractors, employers etc.
Even though multiple clients can share the same session EJB remote reference, only a single thread may access the EJB at any given time. If multiple threads try to access the same client context, an exception will be generated.
For entity bean, each client has its own instance of the EJB class in the container and is not multi-threaded. Whether the EJB Object is multi-threaded or not depends on the vendor implementation.
The issue here is performance. If you go ahead with the synchronization approach then it'll probably create a bottle-neck there because no matter what only one thread can run the doGet() at any instant of time.
Generally speaking, if the doGet() have a lot of things to do, it's good design to synchronize only those code blocks within the doGet() that cause non-deterministic behavior so as to mitigate the performance problem as much as poosible.
If your server can take high loading, it may be better, by using the SingleThreadModel approach, to create a pool of servlet instances so as to allow processing requests in parallel.
I've tried NumberFormatException ( a subclass of java.lang.RuntimeException), (a subclass of IOException), ServletException, and a subclass MyException, as defined below:

All works fine. Only UnavailableException is not working.
On a second thought, I agreed with Jim. You original code should be fine. However, there is something particular for the UnavailableExcpetion. Here is what SRV. of the 2.3 specification said:
"If a permanent unavailability is indicated by the UnavailableException, the
servlet container must remove the servlet from service, call its destroy method,
and release the servlet instance."
If the servlet is released immediately, even before throwing out the Exception object from its doGet() method, this may explain your situation.
Does the doGet() throw ServletException or UnavailableException? According to your description, I think it's the former.
To make your original code to work, you may try:
doGet(HttpServletRequest req, HttpServletResponse res) throws IOException, UnavailableException { ... }
It is unlikely that "structural correctness" as specified in SRV 13.2 includes the order of the sub-elements.
For example:
<!ELEMENT context-param (param-name, param-value, description?) >
<!ELEMENT security-role (description?, role-name) >
Why the "description" sub-element, which itself is optional, sometimes placed in the first position, even before the required sub-element??
The order seems a little arbitary and there is no reason to regard the order as specified is more "structurally correct".
If some container providers choose to follow this "strictly", we need to find out the logic behind it.