[mod.computers.sun] SUN-Spots Digest, v4n26

Sun-Spots-Request@RICE.EDU (Vicky Riffle) (08/22/86)

SUN-SPOTS DIGEST           Friday, 22 August 1986      Volume 4 : Issue 26

Today's Topics:
		        Setup flames and Comments (4)
		     Long cable for monochrome monitors
		      Ringing the console bell in 3.0?
			         SCSI disks?
			     BIBTEX on a SUN3?
				KEE on SUN-3?
			   Serial Line IP Package?
			   SunUNIFY vs. SunINGRES?
	       How do I redirect input with WIN_INPUT_DESIGNEE?
			 Circuit-analysis software?
				    Dbxtool?
		      Request for Help with Sun Windows?
------------------------------------------------------------------------
Date: Fri, 15 Aug 86 19:07:39 EDT
From: Matt Landau <mlandau@Diamond.BBN.COM>
Subject: Setup flames and Comments (1) [Sun-SPOTS v4n25]

>From: Keith Cooper <cooper@rice.edu>
>Subject: Setup flames and Comments (1)
>
>                                                           Try using it 
>to install software on a working network.  Of course it doesn't work, it 
>only appears to be able to deal with internet addresses beginning with 
>the SUN default network.  If your site has class A or B addresses, well, 
>that's too bad.  

Actually, I've recently used setup to install 8 Sun 3's (3 servers, 5
diskless) on a running class B network.  One of those servers was being
installed tapeless using another machine on the same network as a
tapeserver.  Try *that* when you can't use you real network address!

It was no fun, but after doing it the first time, I came up with a hack
that works to get around setup's bogus requirement that you use a class C
network address.  When you're in mini-Unix, before invoking setup, create
an /etc/hosts file that contains the REAL network address of your servers,
clients, and any other important machines on the local network, and reboot
mini-Unix.  Then make your /etc/hosts file mode 222 and run setup.  When
setup asks for internet addresses, give it anything -- it doesn't matter
what, because when setup tries to create the new /etc/hosts file, it won't
be able to overwrite the (correct) file that you've already installed.
Since setup doesn't expect this step to fail, it ignore the failure and
proceeds with the installation, which works just fine!

The only other thing to know is that setup WILL build bogus /etc/hosts 
files for your client machines, so you'll need to replace them before 
you can boot the diskless machines.  Just mount the diskless clients'
root partitions (using /dev/ndl0, /dev/ndl1, etc.) on /mnt and do a
"cp /etc/hosts /mnt/etc/hosts" for each one.  Your clients should now
be able to boot and all your machines should run correctly on the local
(class A or B) network.

While we're on flames, has anyone figured out why Sun 3/75's get bogus
entries inserted into the kernel routing tables at boot time if you're
not running routed?  On each of our machines, netstat -r shows an entry
looking like "default 0.0.xx.yy" where xx and yy are two random numbers.
They're different for each machine, but always the same for a given
machine, so the first thing each diskless machine's /etc/rc.local does
is /usr/etc/route delete 0 <bogus-addr>.  But I sure would like to know
what's causing it.  I've spoken to Sun about it half a dozen times and
each time gotten a lot of head scratching and "We don't know"s.
-- 
 Matt Landau      	 		BBN Laboratories, Inc.
    mlandau@diamond.bbn.com		10 Moulton Street, Cambridge MA 02238
 ...harvard!diamond.bbn.com!mlandau     (617) 497-2429

-----------------------------

Date: Fri, 15 Aug 86 15:07:10 CDT
From: oddjob!fermi!sra@lbl-csam.ARPA (Scott Anderson)
Subject: 3.0 Setup (2)

> A Sun "customer support engineer" told me that Sun "does not support"
> running 68010 binaries on a 68020 cpu. This sucks---but at least it's a
> reason for automagically providing two full sets of binaries.

It is not true that 68010 binaries can't be run on a 68020 cpu; maybe
what you mean is that setup does not support 68010-binaries-only when
there are Sun-3's present.  But in any case you probably will need the
68020 binaries that go in the root partition of a Sun-3 client, i.e. for
it's kernel.

> Of course it doesn't work, it only appears to be able to deal with 
> internet addresses beginning with the SUN default network.  If your
> site has class A or B addresses, well, that's too bad.

This was a minor annoyance; we're a class B network, 128.135.*, and
setup insisted on the individual host numbers for the last two bytes
(the much-discussed lack of subnetting).  So, instead of 128.135.4.6,
I had to type in 128.135.1030.

My own beef with setup is that it would not allow me to designate an
over-lapping partition on the root disk.  We have all of our user files
on a second disk, so I had no use for the 'd' partition.  I therefore
wanted to put the client partitions there, rather than have them be
nebulously in the 'c' partition.  Not possible with setup -- at some
point I'll have to fire up diag, I guess.

It would also be nice if the default edit card in the client setup
would include the root parts of the ethernet and Internet numbers.  I
very quickly got tired of typing "8:0:20:1:" which is the same for all
Sun equipment.

Despite these inflexibilities, I was very impressed with setup.  It did
an awful lot of the work for me, especially since we have mixed binaries,
and generally it did it very well.  Sun obviously put a lot of effort into
the program; hopefully the next version will give the installer a little
more control over things.

				Scott Anderson
				ihnp4!oddjob!kaos!sra     

-----------------------------

Date: Mon, 18 Aug 86 20:42:12 pdt
From: Charles L. Athey III <athey@lll-lcc.ARPA>
Subject: 3.0 Setup (3)

Keith Cooper commented that Setup was and is nothing more than a demonstration
program.  This is a totally incorrect statement.  We at the Lawrence Livermore
National Lab participated in the design review and the beta testing of Setup.
It was NOT designed as a Demo program.  I would agree that Setup has some
problems, especially when us hackers would like to Fix certain things.  BUT
who do you think Sun is selling all of their machines to?  NOT UNIX gurus or
even computer programmers.  We have purchased roughly 75 machines during the
last 3 months and maybe 20 of them have gone to computer programmers -
of these 5 were knowledgable in UNIX.  Most were physicists, geologists,
etc.  These people don't know anything about setting up a machine and they LOVE
setup.  Yes it has problems but it is a whole lot better than what we
used to have to do with Sun release 0.3, remember THAT!!  Setup WAS written
for the NOVICE!  Come on Keith - you know better than to make rash
statements about things you have not been really involved with.

As for Keith's statement that he can't get anyone in Tech support who has
been there any length of time I would say HOGWASH!  Most of those actually
responding have been there awhile now.  Those that return your call, not
those that answer it initially.  I make about 3 software calls to Sun
each week and I have not had a new person in over 3 months.  PLUS - if
you ask for a specific person to be assigned your call you will probably
get them...I have done this at least 10 times in the last 2 months and have
ALWAYS had the requested person be assigned.  Keith, you have been working
with Suns for a long time, don't you know some of the Tech support staff?
Just ask for someone you know.

I know that Sun has some problems, will falsely stating things help though?
If you don't get good service go to the next person up the ladder - it
ALWAYS works for me.

-----------------------------

Date: Fri, 22 Aug 86 10:00:13 cst
From: Keith Cooper <cooper@rice.edu>
Subject: 3.0 Setup (4)

Chuck -

One of the problems with the written word is that sarcasm is sometimes
difficult to detect; certainly my remarks about Setup contained some severe
sarcasm.  Unfortunately, they also contain large grains of truth.  Setup has
dumped core for me in numerous ways: it has severe problems with termcap
entries it doesn't like, with remote tapes, and with internet addresses it 
doesn't like.  My whole point, Chuck, is that SMI is selling machines to 
people who aren't hackers; are they expected to figure these things out when
SUN tech support can't?  In one case, the local office finally loaned us a 
screen, keyboard, and mouse because we couldn't get a reasonable workaround to 
our problem three days of trying.  Sure, I know people in tech support.  
I still got six pieces of electronic mail telling me  "if you'd just read
page xxx, you'd see how to use it."  Unfortunately, Setup had undocumented 
restrictions that the manual never mentioned and the tech support people
hadn't yet discovered.  By and large, the problems I've seen haven't been 
with old, exotic, and unsupported configurations; they've been fresh out of
the box SUN 3 systems.

My real intent was to mention (again) that we might be better served if some
of the effort spent on Setup's user interface had been directed at its
function, instead.  Too many naive users in this town have called me up when
the combination of documentation and help from tech support failed to bail
them out of their problems with this externally beautiful program.

keith

-----------------------------

Date: Sun, 17 Aug 86 02:22:21 PDT
From: hoptoad!gnu@lll-crg.ARPA (John Gilmore)
Subject: long cable for monochrome monitors

I have done this, and it's not very hard.  Happily, the connector
pinouts are specified in the hardware installation manual.

The video cable is a 9-pin cable.  I found a 9-pin male to female
straight-thru cable at my local Zack Electronics store and it worked
fine as an extender.

The keyboard/mouse cable is a 15-pin cable in which only some of the
pins are used.   I tore up an unused Ethernet transceiver cable (I had
to move a few pins to make all the right pins have wire behind them).
It won't screw on to the connectors, since it has Ethernet style slide
lock posts, but I removed the posts so it would fit, and it only falls
off (aborting my system, thank you Tom) every 6 months or so.

-----------------------------

Date: Fri, 15 Aug 86 18:52:50 PDT
From: akkana@infidel.ARPA (Akkana)
Subject: ringing the console bell in 3.0?

Remember the ^G suntools dilemma in 2.x?  Sun consoles were perfectly
capable of beeping, but for some reason Sun decided not to include
that in suntools -- when a window received a ^G, it would flash
rather than giving an audible signal.  Well, now it seems that 3.0
has the opposite problem -- suntools has a bell option, but it is no
longer possible to ring the bell through ^G on the bare console
without suntools -- a ^G does nothing!

I have found ioctls which turn the console bell on and off, but I
haven't been able to find a way of telling the console to beep when
it sees a ^G.  This means that programs like vi, talk, etc. are no
longer able to communicate properly.  A quick call to Sun Tech
Support seems to indicate that this is a "feature", not a "bug".

Does anyone know whether there is a way to make ^G ring the console
bell, Tech Support notwithstanding?  (A way which does not require a
source license, that is.)  Leaving out the console bell after being
persuaded to add it to suntools seems too silly to be likely.

..
	...Akkana            Center for Nonlinear Studies, LANL
	akkana%cnls@lanl.arpa  hp@lanl.arpa  escher!cnls!akkana

-----------------------------

Date: Sat, 16 Aug 86 13:38:57 EDT
From: Barry Shein <bzs@BU-CS.BU.EDU>
Subject: SCSI disks?

I was at the MacWorld expo (I think that's what it's called, here
in Boston this week) and noticed one vendor was advertising a
nice range of SCSI disks (Mirror Technologies.)

The one that caught my eye was a 172MB (formatted) little box
with 30msec average access and approx 1MB/sec transfer rate
(my first remark was, omigod, it's an RP06!) I believe the
price to a University would be about $5500 (maybe a little less.)

Does anyone know any reason off-hand why this -wouldn't- work
with a SUN/SCSI? I assume there could be some trivia with
cabling and explaining it to diag (and maybe setup), but
otherwise it seems like it should just work.

Thoughts? I may ask them for a loaner and report back, but I
figured someone out there could save us both some time and trouble
if there's a fatal flaw I am missing.

	-Barry Shein, Boston University

-----------------------------

Date: Sat, 16 Aug 86 15:07:24 EDT
From: Barry Shein <bzs@BU-CS.BU.EDU>
Subject: BIBTEX on a SUN3?

Has anyone gotten bibtex from the UNIX TeX distribution working?
It seemed at first the problem was just that the language they
used felt that 'others' was the choice of the week for a default
case while the local choice of the week was 'otherwise'. A global
replace got it to compile but it sputtered messages and died
when it tried to imbibe plain.bst.

Thanks in advance, I have a user here who thinks fixing this will
improve his lot in life.

	-Barry Shein, Boston University

Hell hath no wrath like a language ill-defined

-----------------------------

Date: Tue, 19 Aug 86 11:01:22 -0200
From: Eric Ole Barber <Barber%vax.runit.unit.uninett@tor.ARPA>
Subject: KEE on SUN-3?

Does anyone know the status of KEE on SUN-3? I ask because a colleague
was at AAAI-86 and IntelliCorp refused to demonstrate either KEE or
Simkit on the SUN-3 which they had in the room. If KEE does in fact
work on SUN-3, is it a completely self-enclosed sub-environment, or are
'outs' to the operating system possible?

-----------------------------

Date: Tue, 19 Aug 86 08:20:48 MDT
From: uplherc!saturn!jr@utah-cs.arpa (J.R. Westmoreland)
Subject: Serial Line IP Package?

I'm interested in finding some one who has installed the Serial Line IP 
package on a SUN2 or SUN#.  I have tried to install the package with less
than success.

I have found at least one error and several possible incompatable
calls to functions.  Has anyone else found the same problems?

I would appreciate any helpfull suggestions that any one could give me.

J.R. Westmoreland
...!utah-gr!uplherc!jr
...!decwrl!esunix!uplherc!jr

-----------------------------

Date: Tue, 19 Aug 86 14:12:16 cdt
From: carol@pax.cs.UTEXAS.EDU
Subject: SunUNIFY vs. SunINGRES?

We are planning to buy database software from Sun, either Ingres or Unify.
Does anyone have comments on the merits of one system over the other?
First hand experience with bugs, performance, user interface, RPT vs.
Report Writer, etc. would be appreciated.  We'll be running it on 
Sun 3/160s, one with 4MB memory and the other with 8MB.

Carol Kroll
carol@sally.utexas.edu
ut-sally!carol

-----------------------------

Date: Tue, 19 Aug 86 19:50:19 PDT
From: jwm@renoir.Berkeley.EDU (Jeff Mc Carrell)
Subject: Help needed: how do I redirect input with WIN_INPUT_DESIGNEE?

My application has a large image window and a smaller message window
(both canvases) that is used only for ascii messages/prompts and the
like.  I'd like to have any input that the user might type in the small
message window offered to and handled by the image window.

I'm running SunView 3.0.  The documentation for the WIN_INPUT_DESIGNEE
(SunView Programmer's Guide, pgs 42 & 68)
parameter leads me to believe that it is exactly the facility I need;
however, I can't seem to make it work correctly.  When I try to hand
the window (the handle returned by window_create) as the argument to
WIN_INPUT_DESIGNEE as suggested by the documentation, I get an ioctl
error as in:

	message = window_create(frame, CANVAS, ... 0);
	image = window_create(frame, CANVAS, ... 0);
		...
	window_set(message,
		WIN_INPUT_DESIGNEE, image,
		0);

	yields:
	WIN ioctl number 806C6714: Invalid argument

handing window_set the image's file descriptor as the argument as in:

	window_set(message,
		WIN_INPUT_DESIGNEE, (int) window_get(image, WIN_FD),
		0);

gets rid of the ioctl error, but instead of appearing in my image
window, keystrokes are passed to the parent (the shell).  I surmise that
for some reason my image window is refusing them and they are then
being passed to the next link, which is presumably the parent of the
process, or the shell.

Note that my input masks for the image window are already set up to
accept keystrokes.  As long as the user types in the image window,
everthing is fine.

My question:  Has anyone succeeded in using the WIN_INPUT_DESIGNEE
parameter to pass input among tiled windows of a single frame?
If so, how?

				thanks,
					Jeffrey W. McCarrell
					Computer Science Division
					571 Evans Hall
					U. C. Berkeley
					Berkeley, CA 94720
					jwm@renoir.Berkeley.EDU

-----------------------------

Date: Wed, 20 Aug 86 09:29:33 CDT
From: wucec2!rfh0023@seismo.CSS.GOV (Ramon Fernando Herrera)
Subject: Circuit-analysis software?

I am looking for a program that does analysis of electrical circuits on
an Apollo or Sun computer.  The ideal program should make extensive use
of interactive computer graphics, for both input and output. Please mail
me any information you may have.

-----------------------------

Date: Wed, 20 Aug 86 15:19:30 +0100
From: enea!suadb!anders@seismo.CSS.GOV (Anders Bj|rnerstedt)
Subject: Dbxtool?

In the July 1986 issue of Software-Practice and Experience
there is an article by E. Adams and S.S. Muchnick with the
title: "Dbxtool: A Window-Based Symbolic Debugger or Sun
Workstations".

Since it seems to be a very useful tool I am interested in
whether it is available (with or without support), at what cost,
and on which Sun4.2 release ?

---------------------
  Anders Bj|rnerstedt (The vertical bar is actually an 'o' with two
		       dots over it, pronounced like: Bjurnerstedt)

  Department of Information Processing & Computer Science
  University of Stockholm
  S-106 91  Stockholm
  Sweden

UUCP:	{seismo,mcvax,cernvax,diku,ircam,prlb2,tut,ukc,unido}!enea!suadb!anders
ARPA: enea!suadb!anders@seismo.arpa

-----------------------------

Date: Fri, 15 Aug 86 03:51:42 edt
From: ihnp4!mb2c!edsdrd!web <William E. Bejcek>
Subject: Request for Help with Sun Windows?

I was wondering if someone could give me a hand with what seems like a very 
small and simple problem.  The problem is as follows:

     I have a program that uses stdin and stdout all over the place that
     also would like to pop-up a menu off of the window (ShellTool) that it
     is running in.  How can I get a handle on the window so that my program
     can call menu_show () ?

It looks like there is some stuff (WINDOW_ME) in the environment, but I'm
not sure what it gets me.  We are going to try to open it and use
menu_show_using_fd ().

I guess another possibility would be to create a window for my program to
run in, hence, I'll have the handle for the pop-up menu.  But, it looks
like all my stdin and stdout stuff won't work.  Is there a good way to
set this up and allow me to still use stdin and stdout?  If there is, this
seems like the best approach since I'll be able to create other window
type objects off my program's main window.

The bottom line is that I'd like to reduce porting effort, but still be
able to get at all the neat stuff on the SUN.  Obviously, I'll need to
port the window interface routines of mine, I just want to minimize the
rest of the porting effort (ie, I don't want to change all the fprintf
calls).

Any suggestions?

	ihnp4!mb2c!edsdrd!web <William E. Bejcek>

-----------------------------

End of SUN-Spots Digest
***********************