Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!mcsun!ukc!edcastle!as
From: as@castle.ed.ac.uk (A Stevens)
Newsgroups: eunet.micro.acorn,comp.sys.acorn
Subject: Re: Possible bug in C compiler
Message-ID: <7927@castle.ed.ac.uk>
Date: 19 Jan 91 13:03:48 GMT
References: <1991Jan18.150311.23599@cns.umist.ac.uk>
Organization: Edinburgh University Computing Service
Lines: 23

In article <1991Jan18.150311.23599@cns.umist.ac.uk> vanaards%t7@uk.ac.man.cs writes:
>
>   I've just compiled this source, only a few warnings, but unexpectidly
>I discovered that the executable is about 1Meg long!!! How can that be,
>is this really a bug in the C compiler - could someone else verify my
>results. 
>
>  Thanks. 
>
>

From my experience with compiling a Prolog system I suspect the problem
is those large global arrays.  As I recall ANSI-C allocates them
into the o. files!  You can cure this behaviour (as I recall) by
compiling with the -pcc flag.  This switches global arrays from
initialised to unitialised data.   I recall a ``volatile'' decalration
also helped - buy that might produce lousy code.   This bug is
fixed in the latest beta releases of C I am told.


This is all from memory ...

	  Andrew