Most of the work on SCO port of BIND 4.9.3 has been done by Michael A Meiszl.

The following represents my experience from building various BETA versions
on SCO 3.2v4.2.

The following procedure is known to build and install BIND 4.9.3 (as of
BETA11 patch 3) on SCO 3.2v4.2, using gcc 2.5.8.

	0) Backup old named and zone files. cd to named distribution's
           root directory.
        1) make links
	2) cd native.b
           mkdir include/sys
	   cat >include/sys/param.h <<EOF
	   #include <sys/types.h>
	   #include "/usr/include/sys/param.h"
	   EOF
	3) Edit Makefile, search for string SCO and uncomment SCO specific
           lines. Based on my personal experience (building perl on SCO
           system after the BIND's compat library and include files have
           been installed), I *strongly* suggest to leave the INSTALL_COMPAT
           line commented out.
	4) make 
	5) su
	6) kill -9 `cat /etc/named.pid`
	7) If you are secondary server, I suggest to always remove all
	   secondary zone files.
	8) make install
           You need the scoinst (BSD style install emulation) script. There
           are various such scripts on the net.
	9) /etc/named
       10) Be your CMOS battery with you.

You also probably need to update /usr/lib/libsocket.a with the new
resolver routines. This task is a bit difficult and I am currently
(7-Nov-94) trying to write shell script to automate this job.
Alternatively, you can stay with separate libresolv.a and change each
occurence of -lsocket in Makefiles to -lresolv -lsocket. Anyway, you
must relink your applications to take the advantage of new resolver
routines (if nothing else, sendmail is good candidate). Personally,
I am preferring libsocket.a update.

Eduard Vopicka, <eda@vse.cz>

On Dec 8,  2:38am, Paul A Vixie wrote:
} Subject: Re: b11p3 on SCO - problems and fixes
} thanks, this will all be in b11p4
}-- End of excerpt from Paul A Vixie

Thanks!

And another SCO specific one: SCO's kill executable requires *numeric*
specification of signal (!!!). This causes the ndc shell script to fail
miserably - things like `kill -HUP 12345' on SCO simply do nothing w/o
any error message. Maybe this results to kill(0,pid)???. So it would be
essential to have the following patch included in conf/Info.SCO.

Brgds,

Ed

*** named/ndc~	Thu Dec  8 11:29:10 1994
--- named/ndc	Thu Dec  8 11:31:27 1994
***************
*** 26,37 ****
  	echo $ARG
  	case $ARG in
  	status)	echo "$PS";;
! 	dumpdb)	kill -INT $PID && echo Dumping Database;;
! 	reload)	kill -HUP $PID && echo Reloading Database;;
! 	stats)	kill -IOT $PID && echo Dumping Statistics;;
! 	trace)	kill -USR1 $PID && echo Trace Level Incremented;;
! 	notrace) kill -USR2 $PID && echo Tracing Cleared;;
! 	querylog|qrylog) kill -WINCH $PID && echo Query Logging Toggled;;
  	start)
  		[ $RUNNING -eq 1 ] && {
  			echo "$0: start: named (pid $PID) already running"
--- 26,37 ----
  	echo $ARG
  	case $ARG in
  	status)	echo "$PS";;
! 	dumpdb)	kill -2 $PID && echo Dumping Database;;
! 	reload)	kill -1 $PID && echo Reloading Database;;
! 	stats)	kill -6 $PID && echo Dumping Statistics;;
! 	trace)	kill -16 $PID && echo Trace Level Incremented;;
! 	notrace) kill -17 $PID && echo Tracing Cleared;;
! 	querylog|qrylog) kill -20 $PID && echo Query Logging Toggled;;
  	start)
  		[ $RUNNING -eq 1 ] && {
  			echo "$0: start: named (pid $PID) already running"

-- 
"Eduard Vopicka, Computing Centre, Prague University of Economics,
W. Churchill Square 4, CZ 130 67 Prague 3" <Eduard.Vopicka@vse.cz>