Path: utzoo!attcan!uunet!cs.utexas.edu!tut.cis.ohio-state.edu!mailrus!ames!pacbell!sactoh0!tree!stever
From: stever@tree.UUCP (Steve Rudek)
Newsgroups: comp.unix.microport
Subject: Conquer 4.0 on SysV/AT  (was Re: Strange compiler message)
Summary: Conquer compiles if you simplify lots of expressions.
	 But it requires a HUGE ARRAY so you still can't run it.
Message-ID: <344@tree.UUCP>
Date: 25 Jul 89 00:44:28 GMT
References: <34@jhereg.Minnetech.MN.ORG>
Organization: TREE BBS (916)-349-0385 Sacramento, Ca
Lines: 31

In article <34@jhereg.Minnetech.MN.ORG>, dab@jhereg.Minnetech.MN.ORG (David A. Bright) writes:
> 
> Can anybody explain this message from the C compiler under System V/AT?
> "npc.c", line 393: compiler error: segment overwrite - get help
> *** Error code 1
> 
You probably know by now that this type of error message is quite common
when trying to get source with complicated expressions to compile under
SysV/AT.  The work around is to break up the complicated expressions
--things such as deeply nested && and || conditionals--and replace things
like A+=B with A=A+B.

I spent many hours working through conquer 4.0, though, only to find that I
could compile it but not run it -- it uses a huge array and, of course, Sys
V/AT doesn't support huge arrays.  In some other cases it has been possible
to kludge around the need for a huge array by implementing a kludgy virtual
memory.  However, conquer's references to the huge array are pervasive and
it seemed an extensive rewrite of the code would be required; I backburnered
it until the year after next :-).
 
(1) Does anyone have more ambition than I?  That is, is anyone already
working on a conquer 4.0 hack?
 
(2) Why did Microport have 'no plans' to support the huge memory model?
Would it have been that big a deal for them to add that support?

(3) Has anyone worked on a generalized hack to fake virtual memory or, more
modestly, huge arrays?
-- 
----------
Steve Rudek  {ucbvax!ucdavis!csusac OR ames!pacbell!sactoh0} !tree!stever