Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site loral.UUCP
Path: utzoo!linus!decvax!ittvax!dcdwest!sdcsvax!sdccs6!loral!ian
From: ian@loral.UUCP
Newsgroups: net.lang.mod2
Subject: Re: one pass compiler
Message-ID: <452@loral.UUCP>
Date: Tue, 28-Aug-84 11:39:41 EDT
Article-I.D.: loral.452
Posted: Tue Aug 28 11:39:41 1984
Date-Received: Thu, 30-Aug-84 11:26:48 EDT
References: <4005@tekecs.UUCP>
Organization: Loral Instruments, San Diego
Lines: 45

This note is in reference to the portion of Pat Clancy's article 
reproduced below:

> Some two year old flyers I have from Volition do indeed state (in the
> "fine print") that Pascal-style forward declarations are required.
> I would have thought they would have fixed this by now, since this
> is not really Modula-2 (in fact, I think they called it "Modula-][").
> loral!ian's assertion that it is not a problem is surprising. 
> First, the Pascal-style ordering requirements can be very awkward
> and constraining; imposing them on Modula-2 is morally repugnant.
> Second, it means that a forward declaration of an identifier may have
> to be put *outside* the module in which that identifier is declared,
> which certainly violates the spirit of the module syntax.

  The Volition Systems Modula compilers were designed to run on
  microcomputers, notably the Apple ][e, Apple ///, IBM PC and the Sage
  line of 68000 systems.  Microcomputers, especially the Apple ][ and ///,
  have very limited resources and a one pass compiler is best for such
  systems.  There is no way to get rid of the forward references without
  using a two pass compiler.  If this is done, the entire symbol table must
  be kept for the second pass.  There are two places to store the symbol
  table from the first pass: in memory or on disk.  Until recently
  most microcomputers had only 64 K of RAM, so the symbol table clearly could 
  not be contained in memory.  Disk space was also very limited and disk
  access was slow.  Storing the symbol table on flex disk would have
  greatly slowed the compiler and consumed an unpredictable amount of disk
  space.  People who use minicomputers with virtual memory and fixed disk
  space measured in the gigabyte sometimes forget the problems faced by
  people designing software to run in the microcomputer environment.  Given
  the costs incurred in this environment by a multipass compiler one should
  be very sure that the forward reference problem really is that much of a
  problem.

  I am a reformed Pascal programmer, so this may be why I do not find 
  Volitions restrictions a problem.  It should be noted that except for
  forward reference they still allow CONST, TYPE and VAR declarations to be
  intermixed.  

  I can tell from Mr. Clancy's comment that imposing Pascal's declaration
  order restrictions on Modula is "morally repugnant" that he is a fellow
  language zelot.  Down with the evil typeless herisy of C!

			    Ian Kaplan
			    ucbvax!sdccsu3!loral!ian