[comp.lang.c] Standards and such

rob@raksha.eng.ohio-state.edu (Rob Carriere) (09/14/89)

In article <2127@dataio.Data-IO.COM> bright@dataio.Data-IO.COM (Walter Bright) writes:
>In article <11032@smoke.BRL.MIL> gwyn@brl.arpa (Doug Gwyn) writes:
><In article <2121@dataio.Data-IO.COM> bright@dataio.Data-IO.COM (Walter Bright) writes:
><<C is now mature, standard, and therefore obsolete.
><I generally agreed with your comments, except "therefore obsolete".
><There are some gaps in the chain of reasoning that I am unable to supply.
>
>Perhaps an analogy would help. As anyone who works on jet fighter aircraft
>design knows, as soon as you freeze the design in order to put the plane
>into production, it is obsolete. The reason is that the design stands still,
>while technological progress moves forward continuously.

1) This is wrong from any practical point of view: while the leading edge of
design may advance while you go into production, everybody else will have that
same problem to deal with.  So from my point of view as an aircraft user
(yeah, right :-), your product is on a par with the other aircraft on the
market, i.e. it is not obsolete.  Now you as the designer may be lamenting
about all the spiffy features you could stick in were you to design the thing
today, but I don't care, since I know if you were designing the thing today, I
wouldn't be buying it until the mid-nineties.  In fewer words, you have just
re-discovered the concept of development lag.

In the same way, C is obsolete from a *language*designers's* point of view,
but not from a programmer's point of view.

2) Even were it right, it only shows that there exist cases where
standardization == obsolesence.  It does not show that this is true in *all*
caes.  You have not closed the gap in your reasoning.

>The same goes for a computer language. [...] As evidence of this, look at
>languages that have been standardized in the past, and the weaknesses that
>have subsequently been discovered in them

Of the few languages I know, none were obsolete at the moment of their
standardization.  Can you supply a concrete example?

>New languages are invented to take advantage of new computer hardware, new
>compilation techniques, and new programming styles/techniques.

Any product has a finite lifetime.  That wasn't the point.

><117VAC, 60Hz in the USA is mature and standard, but by no means obsolete.
>
>I don't know much about power transmission, but are you so sure it's not
>obsolete? Given today's technology, wouldn't some other voltage or frequency
>be more efficient? I know that the inertia of existing equipment is too
>vast to contemplate changing this, but that doesn't not make it obsolete.

That choice is based on pretty fundamental physics, and the positon of the
optimum compromise isn't likely to change until the technology changes in a
major way.  (cheap AC/DC conversion would be one such major change)

>TV transmission format was high tech in 1950, but is now hopelessly obsolete.
>Backwards compatibility is a major problem for proponents of HDTV. But I
>do hope that the HDTV people plan a way for an HDTV++ when they settle
>on a standard!

Do you honestly think that the properties of HDTV could have been foreseen in
the 1950's?  This makes any attempt to build upwards mobility into TV
standards silly.  Such standards don't change until there is sufficient
technology for major change.  Almost by definition, such changes cannot be
foreseen. 

The point of a standard is to provide a reliable platform on which you can
build.  This means that the standard must be sufficiently far behind
state-of-the-dream that a perspective on what is wanted and what is not has
been gained.  This implies obsolesence to those who design platforms.  It does
not to those who build on them (most of whom wouldn't build on experimental
platforms for reasons of reliability).

SR

eric@snark.uu.net (Eric S. Raymond) (09/15/89)

In <3025@quanta.eng.ohio-state.edu> Rob Carriere wrote:
> Of the few languages I know, none were obsolete at the moment of their
> standardization.  Can you supply a concrete example?

ISO Pascal.
-- 
      Eric S. Raymond = eric@snark.uu.net    (mad mastermind of TMN-Netnews)