Newsgroups: comp.lang.smalltalk
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!math.ohio-state.edu!uwm.edu!fnnews.fnal.gov!gw1.att.com!nntpa!ssbunews!ssbunews.ih.att.com!lgm
From: lgm@polaris.ih.att.com (Lawrence G. Mayka)
Subject: Re: C++ Productivity
In-Reply-To: jim.fleming@bytes.com's message of 27 Feb 1995 02:35:45 GMT
Message-ID: <LGM.95Feb27174357@polaris.ih.att.com>
Sender: news@ssbunews.ih.att.com (Netnews Administration)
Nntp-Posting-Host: polaris.ih.att.com
Organization: AT&T Bell Laboratories, Naperville, Illinois, USA
References: <D3zrpn.Hxq@research.att.com> <3htf7i$psa@news1.delphi.com>
	<LGM.95Feb26150506@polaris.ih.att.com> <3irdq1$sua@News1.mcs.com>
Date: Mon, 27 Feb 1995 23:43:57 GMT
Lines: 51

In article <3irdq1$sua@News1.mcs.com> jim.fleming@bytes.com (Jim Fleming) writes:

   In article <LGM.95Feb26150506@polaris.ih.att.com>, lgm@polaris.ih.att.com (Lawrence G. Mayka) says:

   >In article <3htf7i$psa@news1.delphi.com> jsutherland@BIX.com (Jeff Sutherland) writes:
   >   We build our compiler technology in C++.  Tools are built in a combination
   >   of C++ and Smalltalk.  Applications are all built in Smalltalk.  This seems
   >   like a pretty obvious way to do things and is analogous to the
   >   assembler/C/COBOL mix seen in most large legacy applications.

   >The problem with this approach is that it ill-fits the frequent desire
   >of companies both small and large to "standardize" on One True
   >Language That Solves All the World's Problems.  If you say that
   >Smalltalk is not realistically appropriate for compilers or other
   >systems programming, you have essentially ruled yourself out.  At
   >least one other very-high-level programming language (which I won't
   >name in this newsgroup) can indeed demonstrate ample capability for
   >most systems programming, but it, too, has difficulty answering the
   >perennial question: "Is your language suitable for toasters and
   >microwave ovens with 4 KB of RAM?"  The lowest-common-denominator
   >effect is difficult to overcome.

   Are you saying that Smalltalk is "not realistically appropriate for
   compilers or other systems programming"?

No, but Mr. Sutherland appeared to imply that by saying that "We build
our compiler technology in C++" and that "This seems like a pretty
obvious way to do things."  I was, in fact, warning of the unfortunate
implications of such statements.

   What other "very-high-level programming language" are you talking
   about and why would you not name it in this newsgroup?

CLOS, a.k.a. ANSI Common Lisp.  Its optional type declarations (which
can "boil away" dynamic type-checking) and extensible syntax (which
can employ unsafe or nonportable constructs in a disciplined manner)
can bring it within striking distance of C's performance when
necessary.

   Is your discussion here "academic" or commercial? I ask because
   you refer to "toasters and microwave ovens" which to me seem sort
   of commercial.

You can answer that yourself by asking major universities and
companies about their plans for programming-language usage.
--
        Lawrence G. Mayka
        AT&T Bell Laboratories
        lgm@ieain.att.com

Standard disclaimer.
