Re: C/C++ programmer giving Ada95 a chance -- writing an emulator.

[Robert Dewar <robert_dewar@my-deja.com>]

In article <38E6971D.3A8435B9@acenet.com.au>,
  Geoff Bull <gbull@acenet.com.au> wrote:

> According to the OED, a standard is a "weight or measure
> to which others conform or by which the accuracy or quality
> of others is judged".

Sure, that is the informal meaning of standard, it is the
meaning that is appealed to when you here an advertisement
saying something like "moogoo gizmos, the standard by which
gizmos are judged".

But in the computer language field, it is helpful to understand
the word "standard" in a technical sense to talk about something
in the class of ISO or ANSI standards. Sure, Java has a standard
in the informal sense, and so does every other non-standardized
language. Since every language has a standard in this informal
sense, the informal sense is not much use.

So I reserve the word standard to mean a national or
international standard approved by an appropriate national or
international body. This is a useful usage, and corresponds
to common usage as well. Certainly if you see me use the
word standard in connection with a programming language it
is meant in this technical sense.

Obviously people can use words anyway they want, like
Humpty-Dumpty in Alice, so I will repeat the statement,
now that you understand it better, there is no standard
for Java.

Is this just semantics? or is it an important distinction?
It is indeed an important distinction! When there is an
international standard, there is a very strong pressure
on vendors to conform, for example, the Microsoft C compiler
is indeed compliant with the ISO standard, which has been
prepared in a consensus process to meet everyone's requirements
as best as possible.

The trouble with an informal "standard" like Sun's "standard"
for Java is that it is designed first and foremost to meet
Sun's commercial requirements. This is not a criticism in any
sense, for Sun to do otherwise would make no sense. Now to
a certain extent, trying to be reasonably universal may be
part of this commercial requirement, but there is no guarantee
of this.

The argument between Sun and Microsoft is a perfect example. An
international standards body would give equal weight to the
requirements of Sun and Microsoft (and IBM and ....). In fact
Microsoft's requirements that Java interface well with COM and
DCOM is a perfectly reasonable technical point to be considered.

It is not at all surprising that Sun objects to doing something
that would help Microsoft, and consequently it is not at all
surprising that Sun pulled out of the standardization process.

Proprietary standards are simply not the same kind of beast
as international standards arrived at by consensus. Microsoft
can't complain too much about Java, because they play the game
of using proprietary "standards" for their own technology with
the intention of maintaining commercial advantage.

Probably the most helpful terminology is to use standard in
this context to imply a standard reached by a consensus
process among experts, and "proprietary standard" to talk
about "standards" maintained by vendors to their own advantage.

Note once again that it is not a criticism of Sun and Microsoft
that they use these proprietary standards to their own
commercial advantage. You could even make the case that
their fiduciary responsibilities to their stockholders require
them to take this viewpoint!


Sent via Deja.com http://www.deja.com/
Before you buy.