This file documents some the machines that pdksh has been build
on and notes that apply.

Notes:
(1)	Built and tested by me (sjg), seems ok :-)
(2)	Reported ok (may mean earlier version)
(3)	Reported no good
(4)	Built with ./std/lib*
(5)	Built without ./std/lib*
(6)	No job control

If you succesfully build this shell on another system please let
me know. 

System, OS			Notes	Compiler/Flags
--------------------------	-----	--------------
sun386, SunOS 4.0.2		1,4	{cc,gcc} -D_BSD
sun4c,	SunOS 4.1.1		1,4	{cc,gcc-2.1} -ansi -D_BSD -DHAVE_SYS_STDTYPES
sun3,	SunOS 4.0.3		1,4	{cc,gcc} -D_BSD
sun3,	SunOS 4.1.1		1,4	{cc,gcc} -ansi -D_BSD -DHAVE_SYS_STDTYPES
Bull DPX/2, B.O.S. 2.00.45	1,5	{cc,gcc-2.1} -ansi -D_POSIX_SOURCE
Bull XPS-100			1,6	cc -D_SYSV -DUSE_SIGNAL
i386,	BSDI BSD/386		2

NOTES:
The table above sumarizes the config used.  {cc,gcc} indicates
either compiler can be used.  If gcc-2.1 rather than gcc is
indicated then gcc < 2 may not work.  This is at least true of
sun4c (sparc) systems.

Bull DPX/2:

pdksh is not needed on this system.  It is simply used as a
System V / POSIX test bed.  Build without ./std tree.  I only
tried with gcc-2.1. -D_SYSV must be left out of CFLAGS for POSIX
functions such as sigaction() to be used.

Bull XPS-100:	

Be sure to rm std/h/dirent.h std/h/sys/types.h and undef JOBS as
the kernel does not support it.  This machine has a sigaction()
implementation that appears to be faulty.  A SIGINT terminates
the shell, when using the system's sigaction().  Undefining
USE_SIGACT does the trick.  sigact.c can now be forced to build
by defining USE_SIGNAL, but this is not tested on the XPS.

