[Winstone-devel] Some problems running on Winstone...

[Martin C. <mar...@gm...>]

Hi Rick, I did run some tests today, and identified some minor problems:

1.- Here is the exact message printed while starting the webapp,
running on IBM JVM 1.4.2:

linux:/opt/winstone # [Winstone 2005/08/04 06:51:19] - XML Error (Line
4): cvc-elt.1: Cannot find the declaration of element 'web-app'.
[Winstone 2005/08/04 06:51:23] - HTTP Listener started: port=3D80
[Winstone 2005/08/04 06:51:23] - Winstone Server v0.6.4 running:
controlPort=3Ddisabled prefix=3D webroot=3D/opt/resin/webapps/sampleapp

2.- Error page for HTTP code 500:

I have a centralized error handler, configured via web.xml following
the servlet spec. It has worked fine on Resin2, Resin3, WebSphere,
Orion/Oracle9i and Jetty, but did not work on Winstone, here is the
stacktrace, please read my comments below:

linux:/opt/winstone # [Winstone 2005/08/04 06:48:46] - Untrapped Error in S=
ervle
t
java.lang.Throwable: Invalid path to resource: /WEB-INF/action/xxx/config.x=
ml
        at dinamica.StringUtil.getResource(StringUtil.java:323)
        at dinamica.StringUtil.getResource(StringUtil.java:163)
        at dinamica.Controller.service(Controller.java:121)
        at javax.servlet.http.HttpServlet.service(HttpServlet.java:47)
        at winstone.RequestDispatcher.forward(RequestDispatcher.java:295)
        at winstone.RequestHandlerThread.processRequest(RequestHandlerThrea=
d.jav
a:232)
        at winstone.RequestHandlerThread.run(RequestHandlerThread.java:150)
        at java.lang.Thread.run(Thread.java:534)

javax.servlet.ServletException
        at dinamica.Controller.service(Controller.java:353)
        at javax.servlet.http.HttpServlet.service(HttpServlet.java:47)
        at winstone.RequestDispatcher.forward(RequestDispatcher.java:295)
        at winstone.RequestHandlerThread.processRequest(RequestHandlerThrea=
d.jav
a:232)
        at winstone.RequestHandlerThread.run(RequestHandlerThread.java:150)
        at java.lang.Thread.run(Thread.java:534)

[Winstone 2005/08/04 06:48:46] - Error using error page /action/error/gener=
al fo
r code 500
java.lang.Throwable: Invalid path to resource: /WEB-INF/action/xxx/config.x=
ml
        at dinamica.StringUtil.getResource(StringUtil.java:323)
        at dinamica.StringUtil.getResource(StringUtil.java:163)
        at dinamica.Controller.service(Controller.java:121)
        at javax.servlet.http.HttpServlet.service(HttpServlet.java:47)
        at winstone.RequestDispatcher.forward(RequestDispatcher.java:295)
        at winstone.WinstoneResponse.sendError(WinstoneResponse.java:437)
        at winstone.WinstoneResponse.sendUntrappedError(WinstoneResponse.ja=
va:55
6)
        at winstone.RequestHandlerThread.processRequest(RequestHandlerThrea=
d.jav
a:239)
        at winstone.RequestHandlerThread.run(RequestHandlerThread.java:150)
        at java.lang.Thread.run(Thread.java:534)

javax.servlet.ServletException
        at dinamica.Controller.service(Controller.java:353)
        at javax.servlet.http.HttpServlet.service(HttpServlet.java:47)
        at winstone.RequestDispatcher.forward(RequestDispatcher.java:295)

Winstone reports "Untrapped Error in Servlet java.lang.Throwable:.."
It's not that it was not trapped, the framework controller (the
servlet is this case) trapped the exception with try-catch, and then
throws a ServletException containing the original exception, by
definition, the servlet container would print its own message unless
there is an error-page directive in web.xml, as in my case. Later the
stacktrace prints the following, the core of the problem:

"[Winstone 2005/08/04 06:48:46] - Error using error page
/action/error/general for code 500.." I think it says "I had some
problem trying to pass control to the error page defined in web.xml",
but I cannot figure out why? that would be my question. I think you
mentioned something about request error processing being fixed in last
snapshot?

3.- I tested Winstone on Linux using three commercial grade JVMs: Sun,
IBM and Blackdown. All worked fine, except for the IBM
parser/validation silly message.

But while testing with open source JVMs, I got -sadly as expected- all
kind of problems that inhibited Winstone from running. I am sure that
all of this problems are inherent to the early stage of these JVMs,
however I managed to run Resin 2 with all of them, and even Resin 3
with one of them, so that's why I am asking your opinion, because
Winstone could not run with any of them. Here are the details, all
tests with latest releases of each JVM and the GNU Classpath library:

SableVM: Did not run, complains about system properties it does not
support. Never mind.

Cacao VM: Same as above.

JamVM: Complained about JNDI, but this does not happens with Resin, so
I wonder if there may be something we can do on Winstone code to
override this problem.

linux:/opt/winstone # [Winstone 2005/08/04 06:46:09] - Error binding
object jdbc/dinamica to JNDI Context
javax.naming.NoInitialContextException: Can't find property:
java.naming.factory.initial
   at javax.naming.spi.NamingManager.getInitialContext (NamingManager.java:=
90)
   at javax.naming.InitialContext.getDefaultInitCtx (InitialContext.java:19=
2)
   at javax.naming.InitialContext.getURLOrDefaultInitCtx
(InitialContext.java:227)
   at javax.naming.InitialContext.lookup (InitialContext.java:257)
   at winstone.jndi.WebAppJNDIManager.setup (WebAppJNDIManager.java:138)
   at winstone.WebAppConfiguration.<init> (WebAppConfiguration.java:626)
   at winstone.Launcher.initWebApp (Launcher.java:378)
   at winstone.Launcher.<init> (Launcher.java:143)
   at winstone.Launcher.main (Launcher.java:538)
   at java.lang.reflect.Method.invokeNative (Native Method)
   at java.lang.reflect.Method.invoke (Method.java:355)
   at jamvm.java.lang.JarLauncher.main (JarLauncher.java:50)

Well, that's all, thanks for reading this long report.

Regards,
Martin Cordova
--=20
Dinamica - Open source J2EE framework
Free, easy and powerful
http://www.martincordova.com