RE: [wsrp] [WSRP-IMPL] Thoughts on WSRP Open Source Implementatation...

[Thomas Schaeck <SCHAECK@de.ibm.com>]

Referring to a statement below - I think an a-priori catalog is required to
create a complete test suite.

If multiple parties participate in creation of the test suite, we can then
divide up the implementation of the test cases and have cross-reviews to
make sure the implementations of the test suite are correct.

I think the spec needs to contain all the MUSTs, MAYs, SHOULDs and MUST
NOTs to be complete, apart from that the compliance test suite definition
should be a separate document (that always needs to be kept exactly in sync
with the spec itself)

Best regards,

Thomas



Larry Cable <larry_cable@yahoo.com> on 08/21/2002 06:07:17 PM

Please respond to larry_cable@yahoo.com

To:    Rudnicki Joseph G CONT NSSC <RudnickiJG@NAVSEA.NAVY.MIL>
cc:    "'wsrp@lists.oasis-open.org'" <wsrp@lists.oasis-open.org>
Subject:    RE: [wsrp] [WSRP-IMPL] Thoughts on WSRP Open Source
       Implementatat ion...




--- Rudnicki Joseph G CONT NSSC
<RudnickiJG@NAVSEA.NAVY.MIL> wrote:
> Hello,
>
> I am very interested in assuring that WSRP allows
> total "plug and play."
> Anything else will make WSRP similar to early
> versions of CORBA, more of a
> philosopy than an interoperability specification.
>
> Will creating conformance test definitions make it
> easier to assure that
> WSRP is "plug and play?" Often looking at the
> problem from different
> directions assures a more consistent product. This
> would mean that the
> conformance test definitions could both provide a
> basis for WSRP conformance
> tests and provide feedback to the WSRP
> specification.
>

that was my thinking on the matter

> Take care.
>
> Joe Rudnicki
>
>
> -----Original Message-----
> From: Larry Cable [mailto:larry_cable@yahoo.com]
> Sent: Wednesday, August 21, 2002 7:53 AM
> To: Alan Kropp; 'Thomas Schaeck'; 'Michael Freedman'
> Cc: 'wsrp@lists.oasis-open.org'
> Subject: RE: [wsrp] [WSRP-IMPL] Thoughts on WSRP
> Open Source
> Implementatat ion...
>
>
>
> --- Alan Kropp <akropp@epicentric.com> wrote:
> > I agree with Thomas.  I believe there are
> precedents
> > for the spec
> > implementors also being the compliance suite
> > implementors.
> > In JSR 168, and in fact all JSRs, the spec lead
> (who
> > almost
> > always implements the spec) is also in charge of
> the
> > compliance
> > suite.
> >
> > Given that, can we find a way of sharing the
> > compliance suite
> > implementation across a few TC members, whether or
> > not they are
> > also implementing the spec?
> >
> > Besides the implementation of the compliance
> suite,
> > there are
> > other issues, that the TC needs to decide on, such
> > as:
> > * The process for the TC to validate that the
> > compliance suite
> >   is in fact correct
>
> would there be any value in creating a compliance
> test suite definition; essentially a document that
> catalogs (apriori) the set of conformance tests;
> what
> they test, "how", and what the conformant result(s)
> would be?
>
> Although this is certainly "unusual" this could be
> valuable; it could either be a separate document
> or "inline" conformance statements in the spec
> itself
> as long as they did not significantly impact the
> readability of the spec itself.
>
> Rgds
>
> - Larry Cable
>
> > * Are there any penalties for failing the suite,
> or
> > benefits for
> >   passing?
> > * Do companies administer the test on themselves?
> > If it is the
> >   honor system, what exactly are we expecting to
> get
> > out of it?
> > * If it's not the honor system, what's the process
> > for
> >   administering the test:  does anyone in specific
> > administer it,
> >   how do you schedule a time to do it, do you have
> > to bring any
> >   hardware/software to a certain place.  This
> would
> > mean there's
> >   an ongoing cost to maintaining a compliance
> > program, and are
> >   there TC members willing to shoulder it?
> >
> > -----Original Message-----
> > From: Thomas Schaeck [mailto:SCHAECK@de.ibm.com]
> > Sent: Monday, August 19, 2002 4:30 AM
> > To: Michael Freedman
> > Cc: wsrp@lists.oasis-open.org
> > Subject: Re: [wsrp] [WSRP-IMPL] Thoughts on WSRP
> > Open Source
> > Implementatat ion...
> >
> >
> >
> > I think that while the validation compliance suite
> > should be done by
> > different people, it dosn't necessarily mean
> > different companies.
> >
> > It would be ok if some companies have teams
> working
> > for implementations for
> > their products and contribute different people,
> e.g.
> > from their QA
> > departments to the compliance suite.
> >
> > Since IBM is already providing significant
> resources
> > for editing the spec
> > and providing a free implementation it would be
> good
> > if other companies
> > contribute resources to work on the compliance
> > suite.
> >
> > One thing we should also consider is something
> like
> > a "plug fest" (hope
> > that's the proper term), i.e. a meeting where
> > different companies bring
> > servers with their portals and WSRP producers and
> we
> > test that they work
> > together properly.
> >
> > Best regards,
> >
> > Thomas
> >
> >
> > Michael Freedman <Michael.Freedman@oracle.com> on
> > 08/17/2002 08:52:35 PM
> >
> > To:
> > cc:    wsrp@lists.oasis-open.org
> > Subject:    Re: [wsrp] [WSRP-IMPL] Thoughts on
> WSRP
> > Open Source
> >        Implementatat    ion...
> >
> >
> >
> > We should try and find out if there are sufficient
> > members of the TC
> > willing to
> > work on this that aren't also tainted.  We are in
> > the same position as IBM
> > in
> > that we are beginning early implementations hence
> > need to disqualify
> > ourselves
> > as well.  Who out there isn't planning on
> > implementing a WSRP
> > producer/container over the next 4-6 months and
> > would be willing to work on
> > on
> > a validation/compliance suite?
> >
> > As for calling the Apache work a "reference
> > implementation"  I think we
> > need to
> > be careful.  WSRP has a variety of usage patterns
> --
> > one of which seems to
> > be
> > the target of this implementation.  As "reference
> > implementations" often
> > define/imply a coded version of the specification
> > i.e. the code defines the
> > spec particularly where the spec is unclear -- it
> > seems inappropriate in
> > this
> > situation.  Would "sample implementation" be a
> > better term?
> >    -Mike-
> >
> >
> > Alan Kropp wrote:
> >
>
=== message truncated ===


__________________________________________________
Do You Yahoo!?
HotJobs - Search Thousands of New Jobs
http://www.hotjobs.com

----------------------------------------------------------------
To subscribe or unsubscribe from this elist use the subscription
manager: <http://lists.oasis-open.org/ob/adm.pl>