al@escom.com (Al Donaldson) (12/31/89)
First, my apologies. I'd responded by Email to Ben Cox and had apparently
given him the wrong number for ordering 1.3 from Prentice Hall. And then
Ben posted my mis-information to the world.. My online phone list shows
the following, carefully cut-and-pasted this time using Suntools :-)
Prentice Hall 201-767-5937 orders
Sorry about the confusion, both for Ben and the rest of the world.
Now for some observations on 1.5.0, as if anyone will believe a word I say..
I haven't spent much time on it, but I'm quite impressed. A lot of loose ends
have been tied up.
- I only had one truncated file (#38), and after some help by Glen Overby,
finally got it with the following request:
From al Fri Dec 29 04:56:47 1989
To: LISTSERV@VM1.NoDak.EDU
get pc-upd 150-38 minix
My guess is that you can get other files by modifying the "get" line,
but I haven't confirmed that.
- libc.a: four missing files (unget.s, memmove.s, strcoll.s and strxfrm.s),
and warnings for {sincos.c, doprintf.c, printk.c). Run file excludes _exit.c
since it does not start with [a-z], so you need to compile it separately.
The ordering info was a nice touch (needed in 1.4a), but I found I needed
to start by extracting everything from 1.3 libc.a, then copying in the new
1.5.0 stuff, then building a totally new libc.a.
- After making libc.a (and checking CRCs on header files), the boot disk
was straightforward to build. Some minor confusion over location of lib
(kernel and fs makefiles refer to /lib; mm, tools and commands makefiles
refer to /usr/lib). Otherwise, I built and booted 1.5.0 OS with no problems
except having to set root image device from fd0 to hd2 on boot menu.
Bootup *seems* faster than 1.3PM on the same hardware (AST 386 with slow HD)
-- is this for real or just my imagination?
- Most of the commands compiled straight out of the box, but I had problems
with the following: ast, cc, fsck, more, nm, strip, uudecode.
Most of these were from missing a.out.h, which I'll check to see if I have
elsewhere. Cc probably needs special treatment, and fsck was missing fs/const.h.
As I understand, uudecode is history.
I haven't done any testing of new commands yet, but I found that old kermit
(Charles Hedrick's uuencoded executable) seems to get confused on line speed.
So I'm looking forward to getting the rest of the source, especially kermit.
Again, great job, Andy, Bruce, and everyone else who contributed to this!
Al Donaldson