Hi Jose,
Wow, this is a huge successful story, the email thread started with
"ouch" and ended with "Kudos", I count
total email thread on this is about 60, bugs got fixed one after
another, big thank you to you and to developer
Jan, Kin-Man, Sahoo, and other developers who looked at the issues and
provided the fix.
You have a great test. I will follow up with you and Jan on how to add
your tests into FishCAT test pool.
We may give your test a name as "Jose ouch test" :-)
Let me include your first email on 9/26 so others can get a whole story:
"I installed the promoted b65 build and using Netbeans 6.7.1 tried to
deploy IWebMvc2 (
http://code.google.com/p/internna/). It's quite a
complex application with Spring 2.5, Hibernate 3.3, DWR 3.0RC and many
other libraries (JAWR, BouncyCastle, Jasperreports and so on). It
deploys fine to Tomcat nonetheless. The results were not positive"
Thanks again (many times) for your time and effort to help improve
GlassFish quality!
Judy
Jose Noheda wrote:
> I was waiting for the final release of NB 6.8 / GFv3 to test things
> out. In the meantime I had placed a workaround for the Hibenate cache
> issue I was suffering. Today I finally was able to launch IWebMvc2 in
> GFv3 final and I have to publicly applause the work you've done. The
> logging issue is magically gone and the application fully works! No
> issues at all. None, zero. Kudos to all the team!!
>
> Regards
>
> On Tue, Dec 1, 2009 at 8:57 PM, Jan Luehe <Jan.Luehe_at_sun.com
> <mailto:Jan.Luehe_at_sun.com>> wrote:
>
> Hi Jose,
>
>
> On 11/25/09 09:40, Jan Luehe wrote:
>> On 11/24/09 23:30, Judy Tang wrote:
>>> "The second time (just pressing F5) the page works fine. IIRC
>>> this is the exact same behavior I got with Tomcat with versions
>>> prior 6.0.20 (with 6.0.18 for example). The bug is fixed in
>>> Tomcat now so it may just be a matter of identifying the patch."
>>>
>>> Wow, not only report the issue, but also points out what could
>>> be the cause, thanks Jose for continue testing, I am sure Jan
>>> will get back to you soon :-)
>>
>> Jose, thanks for picking up the nightly and testing it with your app!
>>
>> I just downloaded the nightly myself and deployed your app.
>> When I access the FILTERED GRID link for the first time, the page
>> compiles just fine, and I don't see the stacktrace you mentioned.
>>
>> The line where you are getting an NPE (ReflectUtil.java:101) looks
>> like this:
>>
>> public static void checkPackageAccess(Class clazz) {
>> checkPackageAccess(clazz.getName());
>> }
>>
>> which means "clazz" is null. I've cc'ed Kin-man in case he has
>> any ideas what
>> might be causing this. Thanks for narrowing down the range of TC
>> releases
>> for a possible fix.
>>
>> It looks like we're almost there, though. :)
>
> Kin-man has done some great detective work. I've updated
>
> https://glassfish.dev.java.net/issues/show_bug.cgi?id=11182
>
> with his findings.
>
> Looks like the problem is a user error, that is, a problem in the
> way the
> application's Document.tag is written, which contains this tag
> invocation:
>
> <c:when test="${}">
>
> where the EL expression evaluates to "null", which cannot be assigned
> to a boolean, which is the argument type of
> org.apache.taglibs.standard.tag.rt.core.WhenTag#setTest(boolean).
>
> Note that the above tag invocation was not present in the version of
> Document.tag that was bundled with the earlier version of the app
> which did not show any errors.
>
> We're still puzzled how the app works with Tomcat, though.
> It looks like Tomcat converts the EL expression evaluation result
> "null" to false, which would not be EL spec compliant.
>
> Thanks,
>
> Jan + Kin-man
>
>