Each test is an individual programs to prevent exactly that sort of
occurrence. When I ran portions that hung or failed I cycled the server and
reran them to take that extra step to prevent domino-style failures. I
don't think anything else would be necessary to reset Orion's JMS since
queues aren't persistent and therefore state's not outliving shutdown. The
failures ran the gamut from hangs due to incorrect server behavior,
incorrect exceptions (among other spec. breaks), to missing functionality.
I'm glad you're committed to a real JMS implementation and I look
forward to it being usable. As it stands now it's barely there.
Note that the JMSTest++ suite is not freely available. My Fiorano sales
rep specified that vendors are paying a license fee to use it.
--
Jason Rimmer
[EMAIL PROTECTED]
----- Original Message -----
From: "Karl Avedal" <[EMAIL PROTECTED]>
To: "Jason Rimmer" <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Sent: Wednesday, November 01, 2000 7:00 PM
Subject: Re: Initial JMSTest++ results, sort of
> Hello Jason,
>
> Interesting, we will look at that test suite to see what the results are.
If a
> very large number of tests fail, it might be because there is a bug that
> triggers failure in more than one case which leads to a great number of
> failures, but I can't tell for sure until I've run the tests.
>
> JMS hasn't been the main focus for Orion but with J2EE 1.3 and EJB 2.0,
JMS is
> becoming more important so we will obviously make sure we have a great JMS
> implementation (in J2EE 1.2, JMS was optional and not very tightly
integrated).
>
> Regards,
> Karl Avedal
>
> Jason Rimmer wrote:
>
> > JMSTest++ is a JMS v1.0.2 compliance test suite written by Fiorano
> > (http://www.fiorano.com) which publishes a competing JMS implementation.
I
> > performed the compliance tests on a machine with the following
> > specifications:
> > o Dell Intel Pentium 3 - 600 Mhz with 128Mb memory
> > o Windows 200 SP1
> > o Orion v1.4.0
> > o Javasoft's JDK v1.3 (running in mixed mode)
> >
> > The compliance suite consists of ~800 tests which includes the
breadth
> > of functionality detailed in the JMS v1.0.2 specification..
> > There is a caveat in that Orion's JMS implementation is only claimed
to
> > be JMS v1.0 compliant while the JMSTest++ tests are to determine JMS
v1.0.2
> > compliance. This isn't a large issue as the difference between the
spec's
> > are small and therefore only cover a small percentage of the tests.
> > This message was going to contain details of all the tests, their
> > results, and commentary. Unfortunately, Orion's JMS implementation was
> > unable to pass anything more than the rudimentary tests. The most
common
> > test result was a hang, though a functionality failure was right behind.
> > I spent some time looking over much of the JMSTest++ source and
while I
> > can't claim every test is 100% legitimate, it largely appears
reasonable.
> > The test suite is in active internal use at not only Fiorano, but also
Bea,
> > and Progress (home of SonicMQ).
> > Considering my results I believe it's false to claim that Orion
supports
> > JMS v1.0, let alone v1.0.2. The bottom line is that I would not depend
on
> > Orion's JMS implementation for anything: educational tool, toy, or
> > production application.
> > I don't plan to take up the testing again unless Evermind makes
major
> > JMS updates or the list membership considers specific results to be
> > valuable.
> >
> > --
> > Jason Rimmer
> > [EMAIL PROTECTED]
>
>
