Linus Nikander

I'm trying to invoke an event on ALL nodes participating in a clustered (websphere) environment. Regardless of which node triggers the event ALL nodes in the cluster need to be notified. Which technology should I choose ?

- I've looked att setting up JMS with a JMS engine for each of the nodes, but I haven't solved the problem of getting a single message to be duplicated out to the other engines. Also I want to deploy the same package to all nodes so node-specific listeners/topics isn't an option.

- I looked at quartz, but quartz doesn't let you target a single participating member in the cluster.

- I've been looking at JXTA and/or Shoal (still seems to be beta unfortunately), but I don't want to make to big of an effort to implement something that won't work.

Any suggestions ? Any container-runnable tech is viable.

I've been trying to find this information on the web but it seems nigh on impossible.

Does anyone know how much Bea charges for the normal version of the WLS and for the version with the portal framework ?

I know i could ask Bea themselves, I just don't feel like getting hooked by a salesperson right now

//Linus

Is there any way to prevent the browser from remembering information entered into a form filed in cached pages ?

Say for instance a customer purchases an item using his creditcard. After logging out from the commerce-site he leaves the computer forgetting to close the browser window.

Now user #2 sits down at the same computer. Clicking the back-buttom repeatedly he eventually reaches the form that usere #1 filled in with his cc-information. He quickly notes it and uses it for whatever illegal purposes.

Is there any way to prevent user #2 from obtaining the information user #1 entered ?

I've tested this on the system we are using, and since the browser never issues any new requests when you click "back" there really isn't any way to handle this serverside.

Possible solutions i've come up with so far:

- Use password-fields instead of html-text. Not a feasible approach since we don't want to hide the information the user enters with *'s during entry.

- Instead of submitting the textfield directly; when the user submits, use javascript to set a hidden form-field to the same value the textfield had, clear the textfield and then use the information in the hidden field. Seems to be a usable approach but i'm not sure if it really solves the problem.

I get the feeling there is an easy solution out there, I just can't seem to stumble onto it.

Any insights appreciated

//Linus Nikander - linus@nikander.net

Originally posted by Lasse Koskela:
I would just do
log.trace("foobar!");
and let the logging framework handle the log level. The method call is dirt-cheap -- it's the write operation to disk that makes logging so darn expensive.

I was actually referring to that which is passed TO the log.trace call, not the actual evaluation of if the call qualiifies for logging or not. For example in log.trace("trace " + this.veryHeavyMethodToExecute() ); it's the call to this.veryHeavyMethodToExecute() which could potentially incurr the penalty since it will always be evaluated before the call to .trace is made.

As far as I understood your answer you were referring to the .trace call.

Anyway, I'm still on your side Just have to convince my boss somehow.

//Linus

On my current project the coding guidelines mandate that we surround all log calls with (with the proper log-level ofcourse):

if (log.isTraceEnabled()) {
log.trace("Inside initLocationLists");
}

the motivation for the guideline is that if the logstatement includes any argument that has to evaluated first ( say log.trace("Value a = " + methodcallA(inputA)) ), then that methodcall will be made even though the loglevel is higher than the threshold for that particular log-statement. And because it will be evaluated we will incurr a performance penalty all the time regardless of loglevel.

Even though I can understand the reasoning I still feel that its a bit overkill, and worse that it clutters the code. What are your opinions ?

//Linus Nikander

I want to setup a server which will host NNTP newsgroups. I need the following features:
-The newsgroups should be reachable via a web interface aswell as a newsreader.
-I want to be able to force user to register, and to logon to be allowed to post.
-I want to be able to automatically mail out new posts to users who prefer the e-mail-push version of newsgroup participation.
-I want posts to be searchable via a web interface

Does anyone know if apache James can do all of the above ? If so, is it hard to configure ? if not, are there any other (prefrebly free) products out there that do ?

//Linus Nikander.

I think that they, by asking you to pick out the most restrictive modifier, are looking for the modifier which has the most restrictive properties OVERALL which will still fulfill the requirement of allowing you to access it from within the same class. In a comparison of overall properties, default is more restrictive than protected.

I agree with you that the question is somewhat ambiguous though, and that your interpretation is fully correct. I just don't think that it?s what they meant when they wrote it.

In a recent mock-exam that i attempted the following assertion was made:

String y = "test ";
String x;

assert(x > 1) : x = y + 1;

is this valid because x resolves to test 1 (i.e a value) ?

and will the following assertion be valid

String test = null;

assert (x > 1) : test;

Another question tested the functionality of a private constructor:

Given the following:

class Parent{

private Parent(){
//Code
}

}

class Child extends Parent{

Child(){
//Code
}
}


Now if I, in some other context, try to instantiate a Child:

Child myChild = new Child();

I should get an error since the call to super(); won't be allowed, right ?

A followup to this. Is the only way to circumvent the above problem to either make an overloaded (non-private) constructor in Parent and then call that explicitly from the Child() constructor, or to simply remove the private, or are there other ways around this ?

I've just been put in a project where WSAD is the development tool of
choice. As we are deploying to a websphere server there are definite
benefits to be had by working in that environment.

Being a big fan, and a pretty proficient user, of IntelliJ I was wondering if anyone has experience using IntelliJ in a project where other developers
use WSAD ? Although the step by step debugging offered by WSAD is a big
plus, I figure the added control (I don't like the wizardy style WSAD uses for a lot of its functionality) and speed of development I get by using
IntelliJ might make it worthwhile.

Is it very problematic, do the benefits of using WSAD only outweigh the
advantages of using IntelliJ concurrently with it ? Which issues can/will
arise ? All thoughts are welcome.

//Linus Nikander.