Richard Conn[SMTP:[log in to unmask]] sez:
> I think it depends upon how you choose to use the term.
> There are ISO and IEEE standards. There are industry standards.
> There are "defacto" standards. There are organization standards.
> When it comes to Microsoft languages, I can look in the MSDN Library
> and find the definitive references for their language standards.
I not only agree (Microsoft's documentation for developers is
superb) but would go even farther.
There's something to be said for single-source applications.
Nobody has to go spec diving to discover what PowerPoint 7 for
Win9x will display. Whatever PP *does* display is, by definition,
what PP *will* display.
By contrast, open standards (the one I've become most familiar
with in recent years is VRML -- Virtual Reality Modeling Language)
run the dangers of (a) burdening tool developers with conformance
testing that may keep potential tool developers from getting into
the game, and (b) annoying users with nonconformances caused
by gaps in the spec and gaps in the test requirements.
There are five VRML browsers in widespread use for the Win32
platform, an perhaps a dozen more that are less widely used or
hosted on different platforms. My judgment is that the five major
browsers have the geometry right, the animation right, and the color
and lighting a lot better than it was (I take some small credit for
raising the awareness of browser makers as to how bad the color
and lighting situation was, but Mary Brady, Maureen Stone, and
Eric Haines did the work that made it possible to get as close to
conformance as we've gotten). It's a bitch when the hardest thing
to do is the thing that everybody can see right away.
The existence of multiple browser has made conformance issues
inevitable. I can build a VRML world in 15 minutes that each of the
five major Win32 browsers will render differently in some respect
(admittedly, I draw on years of expertise to do that). In fact, it's
difficult to build an interesting VRML world that all browsers on all
platforms will render about the same (I've done it, and it was a first
Those of us who've been doing Ada for a number of years remember
that, even with mandatory conformance testing in Ada 93, when
you put your code on a different platform with a different compiler,
there was no better word for what you did than "porting." Oh, we
denied it, but that's what we were doing.
Ada 95's common compiler code base has largely saved us from
that, and we might do well to stress that Ada 95 has not only
theoretical but actual portability, and why.
[log in to unmask]