🚀 go-pugleaf

	 I don't think we try to support Nextstep version 2 any more,
	 so I will simply define HAVE_NEXT_NEXT_H unconditionally.
	 It is less work.


Then don't bother defining it at all, just change where it's used, which is
only once, in a conditional in getloadavg.c that either #include's
<mach/mach.h> or <mach.h>.  

Proof:
------
Monolith> find . -exec grep -F emacs {} \; -print
Monolith> pwd
/usr/local/src/gnu/emacs-19.25
Monolith>


Patch:  (file dates are misleading)
------
*** getloadavg.c        Sat Aug 27 23:17:43 1994
--- getloadavg.c.orig   Thu Sep  1 11:25:04 1994
***************
*** 382,388 ****
--- 382,392 ----
  #endif /* LOAD_AVE_TYPE */
  
  #ifdef NeXT
+ #ifdef HAVE_MACH_MACH_H
  #include <mach/mach.h>
+ #else
+ #include <mach.h>
+ #endif
  #endif /* NeXT */
  
  #ifdef sgi

----------end patch

Some more comments re NeXTStep 2.x: 

This code is in getloadavg.c:

#ifdef NeXT
/* NeXT in the 2.{0,1,2} releases defines BSD in <sys/param.h>, which
   conflicts with the definition understood in this file, that this
   really is BSD. */
#undef BSD

/* NeXT defines FSCALE in <sys/param.h>.  However, we take FSCALE being
   defined to mean that the nlist method should be used, which is not true.  */
#undef FSCALE
#endif


1) In 3.0 (and presumably 3.x), BSD is defined in /usr/include/bsd/sys/param.h.
Since 3.0 compiles fine without changing any code to do with param.h, it is
reasonably to assume that /usr/include/bsd is correctly set up as one of
the search pathes in 3.0.  Anyway, the comment could probably also read
3.{0,1,2}, especially if version 2 is no longer supported.

2) FSCALE is also defined in /usr/include/bsd/sys/param.h in 3.0, and
presumably 3.x.