Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!mcvax!guido
From: guido@mcvax.cwi.nl (Guido van Rossum)
Newsgroups: comp.sys.ibm.pc,comp.lang.misc
Subject: Re: Rebuttal to Mr. Russell Nelson's complaint about SPL
Message-ID: <7374@boring.mcvax.cwi.nl>
Date: Wed, 6-May-87 05:15:45 EDT
Article-I.D.: boring.7374
Posted: Wed May  6 05:15:45 1987
Date-Received: Sat, 9-May-87 01:02:27 EDT
References: <224@helm.UUCP> <3997@teddy.UUCP>
Reply-To: guido@cwi.nl (Guido van Rossum)
Organization: "Stamp Out BASIC" Committee, CWI, Amsterdam
Lines: 44
Summary: No, John!
Xref: mnetor comp.sys.ibm.pc:3818 comp.lang.misc:380

Regarding the difference between translators and preprocessors:
I haven't tried SPL (not SPR, Jonh!)  so I don't know which it is, but I
do have a criterion to distinguish translators from preprocessors:

	A translator does *full* lexical and syntactical analysis, while a
	preprocessor leaves part of this kind of checking to the processor
	(compiler/interpreter/...) of its output language.

For example, the C++ front/end by Bjarne Stroustrup is a translator,
because it finds all bugs in the C++ source.  The "C preprocessor" is
what its name says, a preprocessor.  (Let's not pick nits; there may
always be bugs in translators that can make it produce output which
causes errors in the next phase; but this can happen even to a C
compiler: on some systems, code segments > 32K can generate assembler or
linker warnings, and on most systems linker warnings are the only way to
detect unsatisfied external references.)

The SPL implementation at hand could be either, but I would like to
believe its author who says that SPL is a translator (and whose vehement
reply seemed to indicate that he knows the difference).  One problem
could be with the documentation: if it relies on the user's knowledge of
BASIC instead explaining all the syntactical and semantical details of
the language in their own right, it is probably guilty of starting the
translator/preprocessor controversy.  Another problem could be with
semantical restrictions imposed by the underlying BASIC system; I
presume SPL's string operations are translated into the corresponding
BASIC string operations, and this means there are all kinds of nasty
limitations such as a maximum string length of 255 bytes or the
requirement of frequent garbage collections.  But as I say, I don't
know, as I haven't inspected SPL.

Quite a different point is the question whether SPL as a *language*
differs sufficiently from BASIC to deserve all the hype it receives.
When I glanced through the documentation about a year ago, I said to
myself: Gasp! a dialect of Algol-60 with BASIC string handling and a few
other goodies added.  It's good that the users of BASIC get access to a
more structured language; but it's bad that this is the structure of the
sixties.  C'mon folks, programming language design has evolved quite a
bit beyond that in the past 20 years!  I wish Dennis Baer had spent his
time in writing a true microcomputer implementation of something like
Icon, or our own ABC, for that matter.
--
Guido van Rossum, Centre for Mathematics and Computer Science (CWI), Amsterdam
guido@cwi.nl or mcvax!guido or (from ARPAnet) guido%cwi.nl@seismo.css.gov