dave@micropen (David F. Carlson) (09/09/87)
Apologies to the flamers about Microport on this net: in another 2 years
you xenix people will have ATT UNIX too!
I received a beta of Microport's 2.30.10 yesterday. (I hope the number of
numbers they have isn't an indicator that they'll need that many revs!)
Well, people shl(1) and all it implies works! And well too. For those who
don't have shell layers, it is a control program that allows many (8) jobs
to be run from one terminal (or virtual console!!) and ^Z'd and restarted
just like our Berkeley people. ATT does this by specifiying a BLOCK on a
tty line of an inactive job so it waits in IO but otherwise runs. Shl(1)
allows creation, deletion etc., of these shells. The shells are named at
create time and may be called by name or number (number is the only way to
call up a shell under Berkeley job control.)
Other nice things are a massively reworked console/keyboard driver that works
now! Both the floppy driver and the wini driver appear reworked. Support
for the tape driver is hinted at but not included in my Beta copy. The old
beta tape driver doesn't work do to reworked and conflicting dma calls.
(The dma structure under 2.2 was a kluge for the tape drive as the floppy
driver already had dma stuff. The rework was a good thing in all likelihood,
however, it breaks code now.)
I haven't tested the tty drivers yet but they appear reworked also.
The console will now support 16 virtcons! (How about each with 8 shell layers!)
And has some fascinating ioctl's for attaching and detaching console(s) and
(!!!) windows. Undocumented but fascinating.
The big drag is a new boot code that hasn't been worked very well. I tested it
on Pheonix and AMI Bios and got *very* different results and Microport tested it
on AWARD and TVI and got different results from my machines! Oh well. Should
be worked out soon though. (The plus of the new boot is that it is supposed to
allow boot from any kernel plus dos from the dos partition of the hard disk.)
Things are getting better.
--
David F. Carlson, Micropen, Inc.
...!{seismo}!rochester!ur-valhalla!micropen!dave
"The faster I go, the behinder I get." --Lewis Carrolldyer@spdcc.COM (Steve Dyer) (09/10/87)
I am not interested in funding any kind of Microport/XENIX war that
some people might be interested in starting, but I think that the
non-SysV-ness of SCO XENIX is greatly exaggerated. SCO XENIX 2.2
for the 286 and 386 gives you shell layers (for those who care),
sdb, 12 virtual consoles, and all sorts of other neat stuff; it looks
like a 5.2 system without coff; don't know how better to characterize it.
The system administration environment is still somewhat V7ish, which
doesn't make too much difference to me. The programming model is
pretty much vanilla Sys V.
The primary criteria in choosing between Microport System V and SCO XENIX
seems to be in the areas of bugs (or lack of same), price, and support.
SCO XENIX is definitely more expensive, but I think it is also fair to
say that the SCO XENIX 2.2 release for the 286 is substantially more
bug-free than the equivalent from Microport, and this was also true
when comparing SCO 2.1.3 with the contemporaneous Microport release.
When you buy SCO XENIX 2.2 you get: working serial line support (9600
baud all day without crashes or data loss), working streaming tape drivers,
and a very robust crash-free system. It feels "mature", as if it actually
went through a thorough testing phase before being shipped. The same can't
be said for the Microport offering, judging by the public comments we've seen
in this group.
I also have to give high marks to SCO's support. It, too, costs money,
and you have to deal with the silliness intrinsic to a binary-only product
with a large installed customer base: you usually have to talk to a drone
who knows only enough to take your phone number and assign you a problem
number. HOWEVER, SCO folks do indeed follow up quickly, and I've been able
to get many of my problems solved, questions answered and updates received.
It's been well worth the money to me.
I think it basically comes down to price. How much are you willing to pay
for a reliable product with good support? Is it a better deal to get a
somewhat buggier UNIX system for $500 less? That's a legitimate question, and
the marketplace gives you an arena to choose.
--
Steve Dyer
dyer@harvard.harvard.edu
dyer@spdcc.COM aka {ihnp4,harvard,linus,ima,bbn,m2c}!spdcc!dyer