Luc Saint-Elie wrote:

>  Bonjour (Hello) Craig R. McClanahan, vous avez �crit le (you wrote
> on) 12:02 01/09/99 -0700 :
>
>
>> If you can crash a JVM (Dr. Watson under NT, core dump under Unix)
>> with a Java
>> program -- servlet, JSP page, applet, or application doesn't matter
>> -- then there
>> are three possible culprits:
>
> First to be clear I didn't intent to start the classical OS war, just
> facts : my company does a pretty nice business in the intranet market
> where NT is the de facto  standard (been used with Micro$oft software
> or Lotus software).
> And I'm trying to get our developers to show interest for Java
> technologies.
> Until now I succeeded but this JSP weakness makes me wonder....
>

I'm not after an OS flame war either -- NT is real, and I've got apps
running under it as well, even though it would not be my first choice.

What I was objecting to was your assumption that JSP *caused* this
behavior, and is therefore "weak" or "dangerous".

I will restate my view with even a little more generality -- if an OS
crashes for ANY reason (perhaps excluding hardware-caused failures,
unless you're in a high-availability scenario), with ANY application, in
ANY language, it is broken.  The fact that you have to use it, like it,
take advantage of its capabilities ... that's all beside the point.
OS's should never crash, and occurrences when they do should be reported
to the vendor and fixed.  NT4 has had several service packs since its
original release -- are you current on all the patches?  This can (and
has) happened to other OSs as well -- the fact that it was NT in your
case is a fact, not flamewar bait.

Within a particular interpreted language environment (such as Java, or
VB, or PERL, or PHP, or whatever), if a legal program written in that
language causes that environment to crash (Dr. Watson, or core dump),
then the environment has implementation bugs that need to be addressed.
Report them back to the vendor, with your example cases, so that they
can be dealt with.  Sun's JDK has had lots of implementation bugs
addressed through its various generations - it's possible that there's
another one here.

If you draw the incorrect conclusion that JSP should be avoided because
this happened to you here, then whatever the real bug is remains, and
will bite you later whenever one of your apps does whatever the JSP
environment did to make it surface.  A much better approach would be to
help isolate it to a repeatable test case, then report it to Sun's
JDK team, and Microsoft's support team, to get it fixed.

Craig





>
>
>> * Faulty JVM implementation
>
> I use SUN JDK 1.2.2
>
>
>> * Faulty native libraries (if you are using them) such as a type two
>> JDBC driver
>
>
> I use for my tests a 100% java database and JDBC driver (InstantDB)
>
>
>> * Faulty OS
>
>
> Well :-))
> But the point is that NT does exist nevertheless what you/we/I can
> think about it. And there is a large market share in the intranet
> stuff for the NT OS
>
>
>> I'd suggest focusing on one or more of these areas.  The fact that
>> you're seeing
>> these problems under JSP is coincidental -- they've always been
>> possible (unless
>> you recently changed one of the three variables above), but didn't
>> get triggered
>> until whatever the JSP pages are doing under the scenes started
>> getting executed.
>
>
> I'm not sure of that at all. For example import a java class (for
> example <%@ page language="java"
> import="java.sql.*,DBConnectionManager" %>), then compile the jsp
> (call the jsp to have the jspengine compile the code).
> Move the java class elsewhere : result JVM crash, Dr Watson and co....
>
> I mean in this case its normal that the jspengine crashes, perhaps
> normal that the whole webserver crashes, but crashing the OS...
>
> If you stop the server, delete all the produced code (in the work
> directory) then start the server, you get an error but not a general
> crash.
> So I suppose that when something really strange appears for a code
> that has been previously compiled without problem, under NT, the
> jspengine becomes crazy, and dangerous
>
>
>> Craig McClanahan
>>
>
> +------------------------------------------------+| Luc
> Saint-Elie                                 || Development and
> marketing manager              || Pictoris - Paris -
> France                      ||
> http://www.pictoris.com                        || email :
> [EMAIL PROTECTED]
> |+------------------------------------------------+| Java Server Side
> Open Source technologies      ||
> http://www.interpasnet.com/JSS
> |+------------------------------------------------+

===========================================================================
To unsubscribe: mailto [EMAIL PROTECTED] with body: "signoff JSP-INTEREST".
FAQs on JSP can be found at:
 http://java.sun.com/products/jsp/faq.html
 http://www.esperanto.org.nz/jsp/jspfaq.html

Reply via email to