And the problem is?
When creating a servlet/jsp engine you implement to spec. If the spec is
unclear the RI has an implementation of the contract of that spec. If there
are additional features beyond the spec.
Class interfaces need to conform to the spec for the RI. If people are
coding to any given implementation, then they will be sorry when porting.
This will be an issue if future milestone releases are used to try new
features or extensions to the spec, but that should not impede Tomcat.
I'm not sure I see the concern here, or maybe it's just because I've been in
vendor land for so long, that I don't see it as such an issue. Certainly
the OMG and X/Open have never found this to be an issue, but rather a good
thing. Maybe coders should review the official specs, or when migrating
from JRun to WebSphere they will wonder why certain things work differently.
Thor HW
----- Original Message -----
From: Milt Epstein <[EMAIL PROTECTED]>
To: Thor HW <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Sent: Thursday, February 17, 2000 9:27 AM
Subject: Re: Tomcat-as-a-reference implementation?
> On Thu, 17 Feb 2000, Thor HW wrote:
>
> > However you look at it, a RI clarifies some ambiguities in the spec.
> > If the implementation api/contract respnds differently than the RI
> > (assuming the RI conforms to the spec) then it is not compliant.
>
> This is exactly the concern Vince was bringing up, and I think it is a
> legitimate one. If Tomcat is to be both a RI and a production
> implementation, then it may have some features that are not in the
> spec, so relying on it as a RI will be problematic. Somehow the
> division needs to be made clear. Maybe Watchdog does that, I don't
> know.
>
>
> > ----- Original Message -----
> > From: Milt Epstein <[EMAIL PROTECTED]>
> > To: Thor HW <[EMAIL PROTECTED]>
> > Cc: <[EMAIL PROTECTED]>
> > Sent: Thursday, February 17, 2000 8:58 AM
> > Subject: Re: Tomcat-as-a-reference implementation?
> >
> >
> > > On Thu, 17 Feb 2000, Thor HW wrote:
> > >
> > > > The 2 original posts had mentioned those things.
> > > >
> > > > 3. Vendor Y's servlet engine doesn't implement Feature X, or
> > > > > > implements
> > > > > > > it differently than Tomcat.
> > > > > > > 4. Therefore, Vendor Y's servlet engine isn't
spec-compliant.
> > > > If implement isn't a HOW-TO then I'm not sure what is.
> > >
> > > As I suggested previously, I'm 99% sure he meant "in terms of how it
> > > behaves", not "in terms of the algorithm it uses". That seems quite
> > > clear from the context.
> > >
> > >
> > > > ----- Original Message -----
> > > > From: Milt Epstein <[EMAIL PROTECTED]>
> > > > To: Thor HW <[EMAIL PROTECTED]>
> > > > Cc: <[EMAIL PROTECTED]>
> > > > Sent: Thursday, February 17, 2000 8:14 AM
> > > > Subject: Re: Tomcat-as-a-reference implementation?
> > > >
> > > >
> > > > > On Wed, 16 Feb 2000, Thor HW wrote:
> > > > >
> > > > > > I believe the Watchdog parts are the compliance tests.
> > > > > >
> > > > > > As for HOW vendor Y implements a feature, I doubt you would know
> > > > > > unless you decompiled it, which would be illegal. All you
really
> > > > > > care about is that a spec feature is implemented, not HOW. A
> > > > > > reference implementation is refered to when the spec alone is
not
> > > > > > crystal clear and an example of how the feature is to react is
> > > > > > needed.
> > > > >
> > > > > The posts you are responding to make no explicit reference to
"HOW" a
> > > > > feature is implemented, and any implicit reference is to the black
box
> > > > > behavior of the feature rather than the algorithm used (at least,
that
> > > > > seemed clear to me).
> > > > >
> > > > >
> > > > > > As to the production quality, if a vendor can't cut it against
> > > > > [ ... ]
> > > > >
> > > > > Similar, there was no mention of "production quality" in the
previous
> > > > > posts, so I'm not sure of the relevance of this.
> > > > >
> > > > >
> > > > > > ----- Original Message -----
> > > > > > From: Stark, Scott (Exchange) <[EMAIL PROTECTED]>
> > > > > > To: <[EMAIL PROTECTED]>
> > > > > > Sent: Wednesday, February 16, 2000 8:05 AM
> > > > > > Subject: Re: Tomcat-as-a-reference implementation?
> > > > > >
> > > > > >
> > > > > > I agree with this logic from a vendors perspective.
> > > > > > Conformance should be based on a well defined compatability
> > > > > > test suite and any servlet engine vendor can measure
> > > > > > themselves against. Does such a test suite exist?
> > > > > >
> > > > > >
> > > > > > > -----Original Message-----
> > > > > > > From: Vince Bonfanti [mailto:[EMAIL PROTECTED]]
> > > > > > > Sent: Wednesday, February 16, 2000 9:26 AM
> > > > > > > To: [EMAIL PROTECTED]
> > > > > > > Subject: Re: Tomcat-as-a-reference implementation?
> > > > > > >
> > > > > > >
> > > > > > > Yes, I'm concerned about this from the perspective of the
vendor
> > of a
> > > > > > > servlet/JSP engine that strives to faithfully implement the
specs.
> > My
> > > > > > > concern is that people will apply the following logic to
non-spec
> > > > > > > features:
> > > > > > >
> > > > > > > 1. Tomcat is the Reference Implementation of servlets/JSP.
> > > > > > > 2. Tomcat implements Feature X.
> > > > > > > 3. Vendor Y's servlet engine doesn't implement Feature X,
or
> > > > > > > implements it differently than Tomcat.
> > > > > > > 4. Therefore, Vendor Y's servlet engine isn't
spec-compliant.
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
>
> Milt Epstein
> Research Programmer
> Software/Systems Development Group
> Computing and Communications Services Office (CCSO)
> University of Illinois at Urbana-Champaign (UIUC)
> [EMAIL PROTECTED]
>
>
___________________________________________________________________________
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff SERVLET-INTEREST".
Archives: http://archives.java.sun.com/archives/servlet-interest.html
Resources: http://java.sun.com/products/servlet/external-resources.html
LISTSERV Help: http://www.lsoft.com/manuals/user/user.html
