Re: Conformance and Implementations

On Tue, 9 Oct 2001, Rob Lanphier wrote:

> Wait a sec.  While I think your logic for getting to the dreaded
> step 13 is o.k., I question whether or not you are characterizing
> it fairly.
> 
> A direction that I continue to think we should go in is a way of
> making certification possible.  This does not mean that everyone
> has to get certification, but rather that people have a choice.  
> Right now, people can't choose certified implementations because
> there's no such thing as certified implementations.  Big vendors
> see only cost associated with being spec compliant, not the
> marketing/sales benefit (because right now, it's impossible to
> characterize).

I have no problems with certification as long as certification is NOT
endorsed, promoted, sponsored, designed, or otherwise controlled by a
single entity such as W3C. I know of only one [weak] antidote to the
"dreaded step 13": availability of multiple, truly independent
certification options.

Let W3C design standards. Let the market decide which implementation
is better and how to compare/certify implementations. This environment
is far from ideal, and may still converge to "step 13", but I know of
no better large-scale alternative.

I agree that certification should be possible. I think it is possible
now. I question whether it is wise for W3C to be involved in
formalizing the certification process because I suspect it will lead
to introduction of legal and monetary barriers. Thus, I suggest that
W3C (QA) concentrates on testing (a practically useful aspect of
protocol compliance) rather than certification.
 
> I'll cautiously bring up POSIX as an example of how things can go.  
> While I don't feel the POSIX standard is an unqualified success, I
> think it was very helpful in the development of Linux.  Though
> most Linux distributions aren't POSIX certified to this day,
> having the certification carrot out there gave the Linux community
> a fixed goal to work toward.  This is in spite of the fact that
> the big companies involved in the creation of POSIX certification
> almost certainly didn't have that goal in mind.

I do not know what drives Linux community, but I can speculate that
Linux benefits primarily from the _existence_ of widely adopted
standards (such as POSIX) rather than from certification campaigns
behind those standards.

$0.02,

Alex.

Received on Tuesday, 9 October 2001 14:59:22 UTC