> -----Original
Message-----
> From: John Evdemon [mailto:jevdemon@vitria.com]
> Sent: Thursday, August 02, 2001 10:40 AM
>
To: 'Edd Dumbill'; Ron Schmelzer
> Cc: Mike.Champion@softwareag-usa.com;
xml-dev@lists.xml.org
> Subject: RE: Use of XML ?
>
> Good
call. This is pet peeve of mine - calling something a
> standard
when it
> really isn't. The "s" word gets tossed about much
too
> casually these days.
>
> UDDI is best referred to as a
project or an initiative (it is
> definitely not
> a
consortium). The UDDI api documentation is a specification, not a
>
standard.
I'm of two minds on this ...
The "right thing" is
probably to insist that "standards" come from recognized international standards
bodies (there are only 4, right -- the ISO, ITU, CCITT, ? ... and the national
standards organizations affiliated with them. Everything else is just a
commonly-supported / de-facto specification. But almost no one seems to
actually use this definition ... and it does exclude things that have some
legitimate claim to be "standards", such as W3C Recommendations and IETF
Internet Standards. Also, it enshrines as "standards" such things that
made it through the ISO process but don't have significant numbers of real-world
implementations or cross-vendor support (FOSI and DSSSL come to
mind).
The path of least resistance is to surrender, and go with
journalist-speak and define "standard" as any specification that is either
implemented by a dominant vendor (e.g., the Windows API) or agreed upon by any
significant group of vendors (SOAP 1.1, the W3C specs). But then the word
becomes meaningless, and the press releases from BozoCo and flunkies about the
new "standard" they proposed to the W3C have the same status as ISO
documents.
Is there much point in trying to carve out a definition
in the middle that would make W3C Recommendations "standards" but would exclude
UDDI? UDDI may well be half-baked, but (ahem) a casual study of the
xml-dev archives would suggest that this characterization can be applied to a
number of W3C Recommendations!
Maybe "standard" is too overloaded to be
useful,and we need a new word that describes specifications that are a)
thoroughly documented, b) arrived at through some sort of rigorous (and
hopefully open) deliberative process; c) widely supported by interoperable
implementations; and d) extensively field-tested and found understandable and
useful in real-world settings. (Hmm, what in the SGML/XML world meets
these criteria? My list would be: SGML itself maybe, HTML (not sure which
version), XML 1.0, XSLT, SAX 1.0, DOM Level 1, XPath, .... I'm thinking ....
???)