Porting mh to SCO V/386

[Steve Dyer]

In article <168 at seacbc.UUCP> wain at seacbc.UUCP (Wain Dobson) writes:
>Has anyone out there attempted to get mh 6.5 running under SCO V/386. Would
>like to know.

Yes, it's real easy.  I haven't tried the latest MH 6.6, but it should be
similar.  I had a couple of problems which *may not* still be there, since
when I built it almost two years ago it was on a beta test copy of
XENIX 386 2.2.  XENIX 2.3 may be different, and I haven't had the chance
to try it yet.

The minor problems I can remember are:
	I had a 286 small-model "make" which couldn't handle the very
	deep tree traversal which the top-level makefile has to perform.
	A 386 "make" fixed that.

	Some data structures have arrays in them which are sized with
	manifest constants which ultimately evaluate to zero; i.e., after
	the preprocessor pass, you end up with something like:
	struct {
		int a;
		int b;
		char c[0];
	}
The XENIX C compiler at the time didn't like the 0.  I just replaced these
with a '1'.

Finally, you have to choose a delivery vehicle.  I suppose you could use
the XENIX mail programs or MH delivery itself, but there is built-in support
for sendmail, and sendmail (sans DAEMON or IPC) runs just fine under XENIX 286
and 386, so I used sendmail+smail.

SCO's TCP/IP product will include sendmail, I believe.  It's also available
from anonymous FTP from ucbarpa.Berkeley.edu.
-- 
Steve Dyer
dyer at ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer
dyer at arktouros.mit.edu