lesley evensen

If you're only exporting this data and don't need it to persist in your web app, make sure you're releasing all objects in your servlet (you don't have a reference to it somewhere and GC isn't removing it)
Do you get all the data in one buffer and then export to Excel? is there an exception happening somewhere prior to successfully exporting it?

Perhaps a simplistic question, but did you try compiling your jsp before deploying it? If you did then the issue of including the tag library would be surfaced. You can probably test the path as others suggested, putting just an .html page under /intranet and browsing to it.
You should also deploy the library/ies your page is dependent on (maybe you already know that!)

If you haven't yet, check the log file (under the logs folder of your jakarta-tomcat-4.1 installation) catalina.out - if your error handler is writing to sys out, the messages will be there. See if you've exhausted memory. You may have a memory leak in your application and that may explain why your server seems to run most of the time.

Maybe that's like saying Java or C++ is case-sensitive...you would probably expect that. Why do you need to get around this?

To respond to the original question (why use the Apache "Struts" framework), when you have tag libraries allowing you to build pages integrated with Java beans, sql (and less javascript!): you shouldn't necessarily use Struts. The overhead may not be worthwhile if you are only building a small app (small is relative, but if you have only a half-dozen classes or so to deal with that may be a measure of 'small', and only a few pages). If you have MANY pages, Struts can help organize control flow, which is a big issue. If you don't want to write a large controller yourself, Struts provides that framework

Has anyone tried running parallel-ized apps on Palm or other PDA's for instance? The J2ME last time I looked didn't contain packages that would necessarily support that.

Are you able to view a sample JSP page which doesn't rely on any Java classes (just html)? Is DefaultWebApp your document root, and did you add your lib to the Weblogic classpath?

Thank you Jorge - in your analogy, page-swapping makes sense but usually the app isn't handling pages - for VM mgmt, an OS is swapping pages in and out for you; by that token, should you let the EJB server handle 'passivation' and not the code in your application?
If you serialize out the data (write to disk), then you'd expect that bean to be 'passivated' If 'passivating' means the instance-data is cleared, but state is maintained, how much memory is actually released? Is there any reason to invoke 'passivate' in your app, and have the container also checking for beans that can be 'passivated'? I am wondering about the efficiency of this model..
thank you again for your quick response.

I thought WL supplies both .cmd and .bat files (for use on Windows or real OSs) - if not, rename .cmd to .bat and try again in DOS shell..

Is your WL 'compile' spec correct(not missing a slash after'bin'_)ie, you have binjavac.exe, not bin/javac.exe

I have only read about these topics, on the java Sun J2EE tutorial, so hope someone who's used them may answer: is the purpose of 'passivation' (ejbPassivate() ) simply to return a bean to the pool (resource management)? and since you haven't 'removed' it, why would you ever explicitly 'passivate' a bean- when you call 'create' or 'activate', presumably you will initialize values for that instance, establish a primary key, whatever; if your container is passivating objects that haven't been accessed for a period of time, isn't that consuming resources, too? And what is this 'secondary memory' storage the J2EE tutorial mentioned? If the object is not destroyed, memory is not released - if you explicitly invoke 'passivate', what is the benefit? ('passivate' my eye- not a fan of nebulous neologisms)
Thanks for your attention!

The difference between a 'regular' bean and Enterprise bean relates to scope of the bean object- being 'enterprise' wide probably means that objects/methods are interacting across clients and even servers, so must handle locking/threading, and remote identification of objects. Then, within enterprise beans, some are identified as 'session' beans (transactions, business logic dealing with a particular client) and 'enterprise' (persist over the 'life' of a server session, representing data in a database)

Perhaps you want only one reference/object, if that's the case you can synchronize the method or block of code returning the reference. If you create a pool or collection of instances, then you have to know when all instances are 'dereferenced' or not required anymore (I guess keep a count of how many instances have been created)

It would be nice if MS did not feel it had to impose its own flavour on everything. SOAP is a great idea for protocol, but MS will not leave it at that. For that matter, why cannot MS accept w3c standard XML instead, having MSXML? Perhaps they may take a leaf from the book IBM is reading and support open source (even if only to fight MS) software like Linux