Notes on the sequent symmetry (i386 cpu) system.
The operating system is DYNIX V3.0.12 NFS.
	Bill Dunlap
	20 Nov 89

dyn.load does not run properly on the Sequent Symmetry,
although it is not far from working.  I used dyn.load2
from contrib/dyn.load2, slightly patched, for dynamic
loading.  First copy
	cp $SHOME/contrib/dyn.load2/dyn_load.c $QPE
Then apply the patches in sequent.patch :
	cd $SHOME
	patch < sequent.patch
After making everything, you must install the S function
for dyn.load2 :
	$ S
	> dyn.load2<-
	function(names, userlibs = "", syslibs = "-Fats -lm -lc", size = 100000)
	{
        	if(length(names) == 0) return(char(0))
        	ld.tail <- paste(sep = " ", collapse = " ", names, userlibs, syslibs)
        	method <- 1     # use 'ld -A ...'
        	invisible(.Internal(dyn.load(ld.tail, size, method), "S_dynload"))
	}

The following notes refer to problems taken care of
by the patch file.  If you apply the patch, you can
ignore the rest.

1.
There is no ratfor on the sequent, compile the one in src/ratfor.
In src/ratfor, the makefile refers to '-ly' to get yacc stuff.
There is no yacc library, but it is not needed.  Just remove
it from the makefile.

2.
I copied the mach.m and cdefs.h from the sun versions.  The
sequent fortran compiler will recognize the \0 in the C way
if you use FFLAGS=-funix.  It will accept "string"//char(92)
but that does not work for char(0)!.  char(0) does nothing:
x=char(0) does not change the value of x.  So I left the otehr
char(i) stuff in there, but changed char(0) to "\\0".  I
will have to put FFLAGS=... -funix ... in $C/makehead and
$QPE/Makefile.
I took the R1MACH values from the ones installed with cmlib
on the sequent owned by UW Computer Science.  They seem to
work, by I didn't check them carefully.
