Re: SAAJ compatibility tests
by Alejandro Guízar
This is about the CTS for Java EE 5, specifically javaee5_0_tempb-b04-tck.
-Alejandro
Lance J. Andersen wrote:
> Alejandro,
>
> which CTS release are we talking about specifically?
18 years, 7 months
SAAJ compatibility tests
by Alejandro Guízar
Lance,
I am a member of the JBoss Web Services team. The project lead, Thomas
Diesler, indicated you as the person to go to when questions/issues
about the CTS emerge.
I have one such question with regard to SAAJ tests. I was hoping you
could help me, either directly or by forwarding my question to the SAAJ
team at Sun.
-----------
The test case saaj/api/javax_xml_soap/SOAPElement contains two tests,
addTextNodeTest1 and addTextNodeTest2, that verify the addition of
non-whitespace text to SOAPBody and SOAPHeader under SOAP 1.1 and SOAP
1.2. Of the four assertions, only adding Text to a SOAPHeader under SOAP
1.2 is expected to throw an exception.
No explicit requirement on Text children of Elements appears in the SOAP
1.1 specification. This is consistent with the CTS assertions: text
nodes are always accepted.
The table below summarizes explicit requirements on 'character
information item' children of 'element information items' defined by the
SOAP 1.2 specification.
EII CII content Significant?
--- --- ---
Envelope white space no
Header white space no
HeaderElement any character yes
Body white space yes
BodyElement any character yes
Fault white space no
Detail white space yes
DetailEntry any character yes
Given the table above, adding non-whitespace text to SOAPHeader or
SOAPBody under SOAP 1.2 is invalid. This applies to SOAPEnvelope,
SOAPFault and Detail as well. The CTS expects SOAPHeader to throw an
exception but not SOAPBody. Other elements are not even tested.
What is the rationale behind requiring SOAPHeader to throw an exception,
but not other elements?
----------
Thanks,
Alejandro Guizar
Senior Software Engineer
JBoss, a division of Red Hat
E: aguizar(a)redhat.com
M: 512-905-0841
18 years, 7 months
Re: [jbossws-dev] To SH or not to SH
by Jason T. Greene
Ok I am mostly convinced, although ant doesn't use ".sh"
On Tue, 2007-02-13 at 23:32 -0600, Alejandro Guízar wrote:
> Ant and countless other software uses .sh as well. When you see the .sh
> extension you immediately knows what the contents of the file are. The
> same does not apply to a file without extension.
>
> I'd go for keeping .sh.
>
> -Alejandro
>
> Jason T. Greene wrote:
> > Hi Thomas,
> >
> > I was planning on dropping the sh extension for JAX-WS tools. The reason
> > is so that the same command works for all platforms. This works since on
> > windows you don't need to type the .bat extension to execute the script.
> > However everything else in jboss uses .sh, and the old tools use .sh. So
> > perhaps it's bad to break convention.
> >
> > How do you want to do this?
> >
> > -Jason
> >
> > _______________________________________________
> > jbossws-dev mailing list
> > jbossws-dev(a)lists.jboss.org
> > https://lists.jboss.org/mailman/listinfo/jbossws-dev
18 years, 7 months
Re: [Fwd: FW: updated web service test report]
by Thomas Diesler
Heiko,
this is your domain, could you please share your results.
http://tddell.dyndns.org/mediawiki/index.php/Performance_Analysis
cheers
-thomas
William DeCoste wrote:
> Do you have any benchmarks we can share with Intuit? How much
> performance tuning/testing have you guys done? I know we EJB3 people
> have done absolutely zero ...
>
> Thanks -Bill
>
> Jason T. Greene wrote:
>> The reason why I point out #2, is that they are planning on doing a
>> major amount of security processing. WS-Security + custom handlers. So
>> the benchmark is not even representative of their own application.
>>
>> On Tue, 2007-02-13 at 10:21 -0600, Jason T. Greene wrote:
>>
>>> Yeah, I gave information to Paul about how this benchmark is invalid.
>>>
>>> 1. It's comparing JDK6 to JDK5 (you can't do this). If they really want
>>> to compare us to the sun stack they need to use the jax-ws 2.0 FC2
>>> RI on
>>> JDK5
>>>
>>> 2. These tests aren't using handlers. Handlers cause the entire message
>>> to be converted into DOM, once this happens the SUN stack will be
>>> signifantly slower.
>>> -Jason
>>>
>>> On Tue, 2007-02-13 at 07:53 -0800, William DeCoste wrote:
>>>
>>>> Latest performance results from Intuit on JBossWS1.2. If you guys
>>>> have any feedback, please let me know.
>>>>
>>>> email message attachment (FW: updated web service test report)
>>>>
>>>>> -------- Forwarded Message --------
>>>>> From: Barker, Cindy <Cindy_Barker(a)intuit.com>
>>>>> To: paul.robinson(a)arjuna.com, Bill Decoste <bill.decoste(a)jboss.com>
>>>>> Cc: Pedone, Tim <Tim_Pedone(a)intuit.com>
>>>>> Subject: FW: updated web service test report
>>>>> Date: Mon, 12 Feb 2007 17:32:37 -0800
>>>>>
>>>>> Bill/Paul,
>>>>>
>>>>> Please see attached for the web service performance report with
>>>>> JBoss WS 1.2. Although it is a bit better than WS 1.0.4, it seems
>>>>> still magnitude slower than the JDK web service stack. Any thoughts
>>>>> on next step?
>>>>>
>>>>> Thanks,
>>>>> Cindy
>>>>>
>>
>>
>
>
--
xxxxxxxxxxxxxxxxxxxxxxxxxxxx
Thomas Diesler
Web Service Lead
JBoss, a division of Red Hat
xxxxxxxxxxxxxxxxxxxxxxxxxxxx
18 years, 7 months
To SH or not to SH
by Jason T. Greene
Hi Thomas,
I was planning on dropping the sh extension for JAX-WS tools. The reason
is so that the same command works for all platforms. This works since on
windows you don't need to type the .bat extension to execute the script.
However everything else in jboss uses .sh, and the old tools use .sh. So
perhaps it's bad to break convention.
How do you want to do this?
-Jason
18 years, 7 months
