🚀 go-pugleaf

I am trying to use the GNU C compiler (2.6.3) with AT&T's nmake (3.0.1)
and get the following error message:

nmake
+ gcc -O -B/usr/local/nmake/lib/ -I- -c try.c
/usr/local/nmake/lib/include: cannot read
make: *** exit code 1 making try.o

The (trivial) Makefile is:

CC=gcc
try.o : try.c
	$(CC) $(CCFLAGS) -c $(*)

try.c is trivially:

#include <stdio.h>
main(){}

Apparently, gcc thinks that there should always be an include directory
under the directory that contains cpp. This is true in the default case,
but when you you the -B option to override with a different cpp, it
still expects an include directory below it. Of course, there is no
include directory under /usr/local/nmake/lib.

--------------------------------------------------------------

Q1: Why should gcc make this assumption for the -B option?

Q2: What is the proper way to remedy this?

--------------------------------------------------------------

One way is to put a soft link to the GNU include directory under its cpp.
I do not believe this to be an ideal solution.

-Jim.
gent@kingfish.ho.att.com

-- 
Bill Pennock               "Do not ever let anyone claim to be a 
AT&T Bell Laboratories      true American Patriot, if they ever 
(908) 949-2920  HO 2L-321   attempt to separate religion from politics."
billp@kingfish.ho.att.com           - George Washington