Port of MicroEMACS to OS/2
				Jon Saxton
				November 89

This is a summary of the patches needed to upgrade the 3.9e version of
MicroEMACS for OS/2 to be compatible with the 3.10 sources currently
being distributed.


ESTRUCT.H
=========

42> #define MSDOS	1	/* MS-DOS */
62> #define MSC		1	/* MicroSoft C compile version 3 & 4 & 5.1 */
96> #define	IBMPC	1	/* IBM-PC CGA/MONO/EGA/VGA drvr	*/

In the file OS2DEF.H which is normally included as a matter of course into
an OS/2 program Microsoft #defines COLOR.  This causes a problem when com-
piling MicroEMACS because ESTRUCT.H also #defines COLOR to be something quite
different.  The person who created the OS2.C and OS2NPM.C files distributed
with MicroEMACS suggested that Microsoft's #define of COLOR be edited out of
OS2DEF.H on the basis that no Microsoft stuff used COLOR.

Whereas that may be true for OS/2 1.0 there is no guarantee that later versions
of OS/2 will not use COLOR so editing a standard header file is potentially
dangerous.  Accordingly I have devised a simple scheme to resolve the #define
conflict without requiring modification to any header file supplied by Micro-
soft.

What I do is to #define a synonym for COLOR using the English spelling, COLOUR.
Later on, after OS2DEF.H has been included, I #undef COLOR and then #define
COLOR to be the same as COLOUR.


134> /* 
135>    In the following #define the English spelling of COLOUR has been
136>    used to avoid a conflict in os2def.h but it will eventually
137>    resolve to a #define of the American spelled COLOR.  In other
138>    words, if you set COLOUR then it will eventually set COLOR.
139> */
140> #define	COLOUR	1	/* color commands and windows		*/

Now OS2.H (and subordinate header files such as OS2DEF.H) get dragged in
at a point where we can resolve all the #define conflicts.

196> #if OS2 & MSC
197> #define INCL_BASE
198> #include <os2.h>
199> #undef COLOR
200> #undef PCOLOR
201> #undef HIBYTE
202> #undef NEAR
203> #undef PASCAL
204> #endif

218> #define	COLOR	COLOUR
 
EPATH.H
=======

Just need to provide something sensible.

2> 		MicroEMACS files on a system dependent basis.
3> 
4> 		The first entry in the table is the name of the startup
5> 		file.  The second is the name of the help file.  Each
6> 		subsequent entry is a path which MicroEMACS will search
7> 		for its auxiliary files (such as the startup file, the
8> 		help file, command files and so on.

60> #if	OS2
61> {
62> 	"emacs.rc",
63> 	"emacs.hlp",
64> 	"c:\\ue\\",
65> }
66> #endif


BIND.C
======

A couple of unused and undefined parameters were causing the compiler to
generate warning messages.

14> PASCAL NEAR help(/* f, n */)	/* give me some help!!!!

FILE.C
======

A line which said #if MSDOS | MSDOS was changed to:- 

355> #if	MSDOS | OS2

OS2.C
=====

deleted line 17> #define INCL_BASE
and line     18> #include "os2.h"
because these are now handled within ESTRUCT.H

Changed the declaration of "results" to avoid a pointer-to-nowhere condition.

315> 	RESULTCODES	 results;
316> 	char		 envBlk[2];
317> 
318> 	envBlk[0] = envBlk[1] = '\0';

All usages of "results" have to be changed to reflect the new declaration.

354> 	return (0 == DosExecPgm( failName, 1, EXEC_SYNC,
355> 					 args, envBlk, &results, prog));

Included the directory search routines similar to those found in MSDOS.H

473> #if	COMPLET
474> /*	FILE Directory routines		*/
		. . .
		. . .
	(not reproduced here)
		. . .
		. . .
567> #endif
 
OS2NPM.C
========

Deleted the #include of os2.h as this is now done within ESTRUCT.H

27> /* #define INCL_BASE */
28> /* #include	<os2.h> */

Fixed up the keyboard handling

212> 	static int	prefix = -1; /* -1 when not holding a prefix */
213> 
214> 	if (nextc != -1)
215> 	{
216> 		if (!prefix)
217> 		{
218> 			prefix = -1;
219> 			return nextc >> 8;
220> 		}
212> 

229> 	if (keyInfo.chChar == 0  ||  keyInfo.chChar == 0xE0)
230> 	{
231> 		nextc=extcode(keyInfo.chScan);	/* hold on to scan code	*/
232> 		return prefix=0;

Note:	I believe I can do much better things with the AT extended keyboard!