Conference turris::languages

    No, but I can think of one or two that I'd like too...
    
    q
    

  Just wait till they get the maintenance bill for changing all the programs 
  which deal with dates early next century. Might make them reconsider. That's
  assuming the standards committee pulls their collective finger out, stops
  playing silly games making programs incompatible by playing with EXIT PROGRAM
  for no apparent reason and finally recognises that years should be expressed
  as 4 digit fields.
						John Gillings, Sydney CSC

  RE: .2

  The FUNCTIONS addendum provide all sorts of 'real' date support for COBOL,
  and is now part of the standard.  VAX COBOL V5.0, now in FT, has these
  functions in it.   

    I have known languages that died for lack of use (Gogol comes to mind)
    but none that have achieved wide use.  Maybe IPL-V is dead, does
    anybody know?
    
    I think the U.S. DOD is trying to kill Jovial in favor of Ada, but I
    don't know how much success they're having.  The easiest way to kill
    a language is for everybody to stop using it.
    
    I assume you aren't referring to assembly languages.  If you are MACRO
    (assembly language for the PDP-1) is very close to dead; there is only
    one PDP-1 left in service.  I expect the assembly language for the IBM
    650 (SOAP?) is dead.  I hope the IBM 7000 and 1400 assembly languages
    are dead, but I suspect someone, somewhere, is still using them.
        John Sauter

	The Forth people are trying very hard to kill that language.
	The FIG 1978 standard lacked O/S support issues but that
	wasn't important on single-function, standalone machines in
	1978... you merely had to support disks or something.  It was
	powerful.  The restricted it more in Forth-79, but it wasn't
	a great leap.  Then they bent it into Forth-83, which floored
	division and some other stuff, and tightened the reigns a bit.
	This forced a lot of rewrites and now the world required O/S
	support.  The ANSI standard has transmuted the language beyond
	my desire to use it, and since a lot of vendors didn't make
	great any money off their Forth-83 rewrites, I don't see people
	chucking what they've got another time.  The ANSI standard is
	a decade late and overdrawn.

	Likewise what's happened to Common LISP.  LISPs were varied and
	widely asunder, but flourishing.  Committee-ify it, load it
	down with a bunch of junk, produce a language definition the
	size of Steele's huge Digital press book... and you'll run people
	off (those with any sense anyway).  Suddenly Scheme is rising
	in popularity as the elegant, sparse, tail recursive and really
	cool way to do things if you're of a LISP persuasion.  A lot of
	people were driven to Scheme from Common LISP's bloat, though.
	The Revised**3 Report on the Algorithmic Language Scheme (MIT
	AI Lab Memo 848a) says it best:

	"Programming Languages should be designed not by piling feature
	on top of feature, but by removing the weaknesses and restrictions
	that make additional features appear necessary."

	You can kill off a language by making artificial restrictions
	on it in the name of standardization that are contrary to the
	founding principals that made it a good language in the first
	place.

	Regards,
	rcs

In article <[email protected]> [email protected]
(OSF/1 Silver on MIPS !) writes:
                                       Suddenly Scheme is rising
   in popularity as the elegant, sparse, tail recursive and really
   cool way to do things if you're of a LISP persuasion.  A lot of
   people were driven to Scheme from Common LISP's bloat, though.

As an ex (due to job change) member of the Common Lisp standards team,
I basically agree.  Scheme is a smaller, cleaner, in many ways nicer
language.  Now that R^4S has macros, it's even complete enough for the
standard language to be useful (before that, virtually every
implementation had it's own macro facility and real programs tended
not to be portable).

  Unfortunately, Scheme still suffers from one serious defect compared
to Common Lisp: no standard library.  Common Lisp's error was to very
thoroughly confuse the standard library with the standard language.
Scheme has avoided that, but the cost is your right and duty to
reimplement *everything*, even something as basic as structures,
yourself.  This is a high (IMHO, unacceptably high) cost to impose on
the commercial user.  Possibly the largest argument in favor of using
Common Lisp is the enormous collection of built in components; I can
concentrate on solving my problem instead of reinventing hash tables
for the millionth time (or locating my previous favorite hash library,
etc.) and I can port my results without having to port my support
library.

Finally someone, Aubrey Jaffer, seems to be doing something about
Scheme's library problem.  Slib is a free (unfortunately GPL'ed) start
at a portable, standard Scheme library.  Aubrey has been asking for
and accepting community input on how to design pieces of this, while
being wise enough to ignore the academics who say that Scheme should
wait *another* ten years for each piece until the problems are
perfectly understood.  It's probaby foolish optimism, but I really
hope the Scheme user community will take Slib as an interface model
and produce implementations of it that can be used in commercial
products.  Maybe even a library standard so that commercial developers
can use this very nice language without reinventing wheels every time
they turn around.

                                            dan

    re .4
    
>The easiest way to kill a language is for everybody to stop using it.
    
    I read the other day that there are 100 billion (presumably billion
    being 10 to the ninth) lines of code in use in the US today - maintained
    at a cost of 30 cents per line per year. That represents plenty of
    inertia (or is that momentum?). )-:

    >> I hope the IBM 7000 and 1400 assembly languages are dead, but I suspect 
    >> someone, somewhere, is still using them.

    I heard recently that there are STILL many engines running SPS &
    AutoCoder.  Though most are via emulators, it does say something for
    the longevity of the breed.
    
                         Long Live the WordMark!