Newsgroups: comp.lang.python,comp.lang.tcl,comp.lang.scheme,comp.lang.misc,comp.lang.perl
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!pipex!uunet!sytex!smcl
From: smcl@sytex.com (Scott McLoughlin)
Subject: Re: What language would you use?
Message-ID: <w1ycVc2w165w@sytex.com>
Sender: bbs@sytex.com
Organization: Sytex Access Ltd.
References: <SCHWARTZ.94Nov6001507@galapagos.cse.psu.edu>
Date: Sun, 6 Nov 1994 07:37:19 GMT
Lines: 107
Xref: glinda.oz.cs.cmu.edu comp.lang.python:2373 comp.lang.tcl:21407 comp.lang.scheme:10970 comp.lang.misc:18787 comp.lang.perl:38079

schwartz@galapagos.cse.psu.edu (Scott Schwartz) writes:

> Tom Christiansen <tchrist@mox.perl.com> writes:
>    A nice interpreted Pascal system, like the old UCSD pascal, might not
>    be a bad idea either
> 
> Sorry to pick a nit out of a long posting, but that statement strikes
> me as untrue.  The reason the old UCSD pascal system is extinct is
> that Turbo Pascal beat the daylights out of it.  Turbo v2.0 on a PC/XT
> was certainly much more pleasant to use than UCSD on a Sage 4 (10MHz
> 68000).  It seems obvious to me that cheap ($40) tiny (24K), fast
> (hundreds of lines per second on an 8086), native code compilers are
> The Right Thing (whatever language you pick.)
> 
Howdy,
        You know - you're absolutely right.  I am, I guess, a
complete madman - a non-academic, non-research, non-big industrial
corporation guy who thinks that Lisp/Scheme is a GOOD THING: gc,
cool macro systems, first class functions, latent typing with
optional declarations and/or type inference, etc.
        OTOH, coming out of "the trenches" I asked myself: "Gosh,
where is the $150 Windows version with a simple GUI toolkit,
debugger and native code compiler that runs < 4megs." And,
of course, the answer is that such a system does not exist.
Strange but true.
        Maybe it's "culture". If you don't _have_ to use 
DOS/Windows and you have $$ to buy whatever system you want,
then why would anyone _want_ to hassle with this platform?
        But increasingly I expect there's an economic 
concern. I quote from Fateman's "Fast Floating-Point
Processing in Common Lisp", section 5 "How Good are 
Modern Lisp Compilers":  "Can we build compilers that
generate appropriate optimized code for these
'simple' programs? We believe the answer is yes... Then
the question becomes one of economics: Is it worth 
implementing the tools in Lisp to perform the necessary
optimizations? This is the big sticking-point, as we
see it."
        I do not think the problem is a lack of a 
"market".  One of my partners spends three days working
in a fairly traditional MIS setting at U.S. Govt. The
horror stories are sad. Folks trying to implement
complicated control flow, data structures, reference
semantics using Visual Basic and Oracle's SQL and an
RDBMS.  Many projects are a nightmare, frequently
total failures. The languages have very little 
expressive power, so complex hacks are devised and
the projects collapse from the weight of their own
complexity.  No rocket science required, just simple
lists and trees in various "typical" configurations
for "typical" uses.
        "C"'s no good for these folks. They employ
contracted teams of programmers with fairly mediocre
skills, whatever else you want to say about C's
suitability to task.
        In other words, a Lisp/Scheme like language
_could_ be perfect for these folks. Much more
expressive power, but more "safe" and "high level"
than C's bit twiddling orientation.  A nice OODBMS
with good coupling would be _killer_ for these
dudes; many (not all) of their problems would 
simply disappear.
        I think the problem might be that 
getting all the expressiveness packaged up
into a nice tight optimizing compiler is too
expensive a task. You got the type inferencer,
the optimizer, the multiple entry points into
functions, the garbage collector, the dynamic
linker, etc. etc. etc.  You _can_ do all 
this; it has been done. The resulting systems
seem to be too large/too expensive FOR THE
SYSTEMS CURRENTLY POPULAR IN THE BUSINESS
COMMUNITY.
        Another "dynamic language", Smalltalk,
seems to be making inroads into this "market".
It's interesting to watch.  Digitalk/ParcPlace/
IBM all sell versions (IBM's is new). There's
another one who recently came out with a Mac
version. Of course, they all have extensive
toolkits to support GUI's, group programming,
RDBMS's and OODBMS's, "visual" programming
(really just screen painting), etc. They 
tend to be pricey though and used in the
"higher value added" sector of the MIS
computing hierarchy.  The price factor 
might just be a consequence of difficulty
of implementation/maintenance/support. In
other words, Turbo Pascal and Visual Basic
can be sold for $100 U.S. because they are
less costly to manufacture.  When these
were introduces, neither Pascal nor Basic
were setting the world on fire and TP was
introduced by a co. with relatively minimal
resources.
        It's sad if true, and I'd love
to hear otherwise.  I think one of the
best things the Lisp community could do
is write/publish the "Dragon Book" of
Lisp compilers and start teaching this
stuff in compiler courses to students.
I'd love to discover that I'm wrong.
        Sorry for the long post.

=============================================
Scott McLoughlin
Conscious Computing
=============================================
