Subject: Re: case confusion
From: Erik Naggum <erik@naggum.net>
Date: 2000/08/25
Newsgroups: comp.lang.lisp
Message-ID: <3176229973244563@naggum.net>

* Kent M Pitman <pitman@world.std.com>
| So a standard which says 
|  "You can do whatever you want and call your implementation conforming."
| gets points for being a good standard?

  Well, that's certainly _one_ implication.  I had other implications
  in mind, though...  Like, determining whether an implementation is
  or is not conforming is sometimes extremely hard because of a number
  of important issues that are reduced to fuzziness where you don't
  know how much you can do or how much you can't do.  Of course, the
  relative tightness of specification is just assumed to be reasonable.

|  (2b) Using just the standard, can one take a conforming implementation
|       and get work done without appealing to implementation-specific stuff.

  I'd label this "completeness", which is a different metric again.

| The Scheme goal was not much different--to unite a community of textbook
| writers.

  Heh.

| Note that I'm not being intentionally particularly defensive about
| CL's standard.  I don't mind if people say it's not up to par or
| what they expect in a standard.

  ... and here I was trying to argue that ANSI CL is more useful than
  the Scheme standard.

| I just absolutely don't believe the Scheme standard is fairly cited
| as a model of a "better" standard.

  Well, on that I argued that the model of better computer science
  standards have changed from the industrial standards of the past.

| In my opinion, it's a toy language.

  But there are many good standards for toys.  E.g., OSI (or "ISO
  reference model for Open Systems Interconnection" for completeness).
  Nobody in their right mind would even propose to use the whole set
  of OSI protocols (those who did weren't :), but the standards define
  a universe of their own that is fairly complete.  Pity it doesn't
  and can't apply to this universe.

| It achieves its "prettiness" and its "smallness" on the back of just plain
| leaving stuff out where it would appear to "clutter", and whatever you
| think of CL, I personally reject any claim that the Scheme is a model of
| improvement.

  I see that the point of my message failed to get through.  I have
  argued that Scheme is a better standard in that world of standards
  that isn't hampered by the requirement to fit the real world, and
  that, when hampered by the real world, ANSI CL is as good as you can
  get.  The real world is incredibly messy and a model of the real
  world that tries to brush the messiness under the rug can look nice,
  but not work nice.  What works nice is usually not very pretty.

  I'm not trying to hold Scheme up as a model.  I'm trying to explain
  that they have succeeded along _one_ metric.  Obviously, I don't
  like the result at all, or I would actually have used Scheme, and
  not Common Lisp.  There's a reason I wrote hyperspec.el for Emacs so
  I could jump right to the reference page, too.  Please don't forget
  that.

#:Erik
-- 
  If this is not what you expected, please alter your expectations.