Hi Johan,
On 08/12/09 02:15, glassfish_at_javadesktop.org wrote:
> Hi Jan,
>
> I created
> https://glassfish.dev.java.net/issues/show_bug.cgi?id=9103
>
thanks so much for your efforts in opening an issue with a clear
and complete description! I've added a link to the forum thread to it.
> The root of the problem is clearly not caused by something in the webstack. I didn't figure out why the classpath contains a wrong jar file (foo-ejb.jar instead of the directory foo-ejb_jar).
> But now, the loading of the web-module logs a non-fatal warning, and the web-module is deployed (but unusable due to non-finished initialization). I see two options:
> 1. TldScanner.java catches the error about non-existing jar, and continue to do its work
> 2. If TldScanner can not complete its work, the module should not be loaded.
>
> Let me know if something is unclear.
>
I agree that any exception thrown by ServletContainerInitializer#onStartup
(note that the JSP container's TldScanner implements
ServletContainerInitializer)
should be considered fatal and prevent the web module from being loaded.
I am going to make this change (in the Servlet container).
In addition, we should make TldScanner more robust, by having it check
if a given
JAR file exists before attempting to open it. Under normal
circumstances, you would
expect for any JAR file that is present in the classpath to actually
exist, so the fact that
foo-ejb.jar is listed in the classpath, but does not exist, while its
contents have been
extracted to a foo-ejb_jar directory (is this directory present in the
classpath?),
is worrisome.
But now we have enough information to start investigating.
Thanks!
Jan
> - Johan
> [Message sent by forum member 'johanvos' (johanvos)]
>
> http://forums.java.net/jive/thread.jspa?messageID=360040
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: users-help_at_glassfish.dev.java.net
>
>