Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!cmcl2!rna!dan
From: dan@rna.UUCP (Dan Ts'o)
Newsgroups: comp.sys.ibm.pc
Subject: MSC 4.0 problem
Message-ID: <640@rna.UUCP>
Date: Thu, 20-Aug-87 17:06:43 EDT
Article-I.D.: rna.640
Posted: Thu Aug 20 17:06:43 1987
Date-Received: Sat, 22-Aug-87 11:46:03 EDT
Organization: Rockefeller Neurobiology
Lines: 51


	While converting a large C program from small model to large model
using MSC 4.0, I encountered a problem not found on all other (UNIX)
C compilers. It went something like this:

	char *s0, *s1, *buf0, *buf1;

	buf0 = malloc(512);		/* Get two char buffers */
	buf1 = malloc(512);
	gets(buf0);			/* Fill with a line of text */
	strcpy(buf1, buf0);		/* Copy to second buffer */

	/* Some kind of lexical analysis, moving s0 to somewhere within buf0 */
	for (s0 = buf0; *s0 && *s0 != '/'; s0++);
	if (*s0 == '/')
/*WARNING*/	s1 = buf1 + (s0 - buf0);/* I want s3 to point in the same
						respective position in buf1 */

	The last line provoked 2 warnings from MSC 4.0, both something like:
Warning: conversion of far pointer to int.

	The binary result did not run. Fare enough, so I explicitly casted
an (int):
		s1 = buf1 + (int)(s0 - buf0);

	With this code, I got one such warning and the code still didn't run.
I didn't try casting to (long), figuring that you can't have objects
greater than 64Kb anyways, although (long) might have made sense. Instead
I settled for this, which removed the warning and worked:

		int n;
		...
		n = s0 - buf0;
		s1 = &buf1[n];

I believe,
		int n;
		...
		n = s0 - buf0;
		s1 = buf1 + n;

also works. Can anyone tell me the big difference between casting (int) which
didn't work and assigning explicitly to an int, which did work ?

	Thanks.
				Cheers,
				Dan Ts'o
				Dept. Neurobiology	212-570-7671
				Rockefeller Univ.	...cmcl2!rna!dan
				1230 York Ave.		rna!dan@nyu.arpa
				NY, NY 10021