Sun-Spots-Request@Rice.edu (William LeFebvre) (09/28/88)
SUN-SPOTS DIGEST Monday, 26 September 1988 Volume 6 : Issue 239
Today's Topics:
Re: mail between smtp and uucp sites
Re: 4.0 versions of various utilities
Re: Old Sunspots question: SPE
yp + 4.0 + named + ypserv FIX
Sun Domain Name Server & YP
Braindamaged SunOS Install Scripts
Rudeness of SunOS 4.0 getwd
crashing pascal under 4.0
sun graphics documentation
Need Help with graphics on a Sun
PATRAN on SUN 4s (particularly running 4.0 with X11)?
FootMouse?
8mm tape drives vendors?
Send contributions to: sun-spots@rice.edu
Send subscription add/delete requests to: sun-spots-request@rice.edu
Bitnet readers can subscribe directly with the CMS command:
TELL LISTSERV AT RICE SUBSCRIBE SUNSPOTS My Full Name
Recent backissues are available via anonymous FTP from "titan.rice.edu".
For volume X, issue Y, "get sun-spots/vXnY". They are also accessible
through the archive server: mail the request "send sun-spots vXnY" to
"archive-server@rice.edu" or mail the word "help" to the same address
for more information.
----------------------------------------------------------------------
Date: Thu, 22 Sep 88 13:39:09 PDT
From: brianc@daedalus.ucsf.edu (Brian Colfer)
Subject: Re: mail between smtp and uucp sites
In V6n229, Stephen Jay (ultra!shj@ames.arc.nasa.gov) asks responds on how
Tahsin Choudhuri's (choudhur@umn-cs.cs.umn.edu) question in v6n206 about
how to address mail from A to E when they are connected:
> I think you can also use uucp only type addresses: "B!C!D!E!ue".
Only if your sendmail.cf is set up to do this. Here is an exerpt
############################################################
## mc/daedalus.m4
##
## generated on Wed Jul 6 20:32:35 1988
############################################################
------------------------------
Date: Thu, 22 Sep 88 17:24:11 EDT
From: aad@stepstone.com (Anthony A. Datri)
Subject: Re: 4.0 versions of various utilities
If you mean the sun consulting tcopy, it's now part of the 4.0
distribution, called tcopy. I haven't tried it, but have been using the
copytape from comp.sources under 3.2 with much success.
------------------------------
Date: Thu, 22 Sep 88 14:02:25 CDT
From: drl@vuse.vanderbilt.edu (David R. Linn)
Subject: Re: Old Sunspots question: SPE
I am currently in the market for CL across several architectures and have
been doing some pertinent homework.
The candidates boiled down to Lucid, Franz, and Ibuki. Both Franz and
Ibuki offer good site licenses (Ibuki's is a bit lower). Lucid does not
deal with users directly and I was referred to the sales people for the
vendors (SUN and HP). I have heard rumors of an SPE type product for each
of Franz and Ibuki so Lucid is probably out of the running.
David
David Linn - drl@vuse.vanderbilt.edu
------------------------------
Date: Thu, 22 Sep 88 13:30:39 -0400
From: Ronald G. Minnich <rminnich@super.org>
Subject: yp + 4.0 + named + ypserv FIX
Any number of people have had trouble with the following: trying to get yp
on 4.0 working with named, optionally with 3.x hosts on the same net. The
sympton in my case was that yp would hang forever on host name look-ups if
i ever tried to use named. The shipped patch tape did not work either, and
when i ran in.named with a debug level of 50 i noticed that it was being
asked for garbage host names of the form 'r.super.org' and '|.super.org'.
Evidently garbage requests are normal behaviour for yp. For some reason
it has to 'prime the pump' with a garbage host name. I don't know why.
Think about the cost:
talk to ypbind to get a server.
talk to ypserv; hand it garbage name, it forks, calls named, fails.
talk to ypserv again, till you are happy.
There is an additional problem though. On our system the requesting
program (e.g. rsh) would never stop asking for 'r.super.org'. It was like
it never got a hard error on the name, just 'try again'-type errors. To
you it looked like the program (e.g. mount of an nfs partition) was
hanging. And the mount would never background, as it never got past
host-name resolution to the point where it COULD background. So in some
cases you never even booted.
I took a look at the old ypserv sources from our vax running mt. xinu.
(Copyright Sun Micro by the way, so i hope nobody hassles me about this).
SMALL FLAME
Sun, i am about to fix a problem for you. I am doing it even though we
STILL don't have our sun sources, and i had to use mt. Xinu source! Are
you listening? There are lots of people out here who WANT TO HELP YOU
OUT!!!!!! GIVE THEM SOURCE!
END OF SMALL FLAME
In the host name lookup part, there is a line like:
if (h_errno == TRY_AGAIN || h_errno == NO_ADDRESS) exit(0);
(this is not exact )
Now, take a look at /usr/include/netdb.h for the definition of the
NO_ADDRESS error:
#define NO_ADDRESS 4 /* Valid host name, no address, look for MX record */
What does this mean? That there is no inet address for the host, but it is
an ok host name. What yp maps are being searched? the byname and byhost
maps. So, if there is no address, what should the error be? Take a look at
the code in TryInterDomain. Seems clear to me that the error should be
YP_NOKEY, and it seems real clear that it should not be handled the way it
is currently handled. The results bear this out: the yp routines
continually ask for <garbage>.super.org, and never move on, because the
only error they get is 'try again'!
The fix is to change the line to something like :
if (h_errno == TRY_AGAIN) exit(0);
so that only TRY_AGAIN errors will be retried. That fixes it for me. The
ypserv seems to work fine most other ways.
(Two days later) If only my problems had ended here. The next problem was
with ypbind. The problem is that it would get 'stuck' on a 3.5 node; that
is, it would decide to be served by a 3.5 host and i could not
/usr/etc/yp/ypset it to my 4.0 host no matter what i did. A look at the
old ypbind code and some printfs added to it show that it would get bound
to the node, but then was unable to ping the node to get unbound. It
appears unwilling to bind to a new node until it successfully pings the
old one to get unbound there. But 3.5 hosts can't really talk to 4.0
ypbind for some reason (i guess rpc version number?) so the unbind could
never succeed. This is all guess work from a little time spent debugging
ypbind. I think it is basically right.
The fix was ugly. In ypbind_broadcast_ack i hard-coded a check to make
sure sin->sin_addr was the address 192.31.192.15, our 4.0 machine. I did
this because i am certain (ha ha) that the 4.0.1 tape will somehow fix
this problem. Or maybe the 4.1 tape. Anyway we will only ever have one
host ever running 4.0.x based on these wonderful experiences.
If only my problems had ended there.
The final problem was that ypserv would get screwed up on the maps. It
would suddenly decide a map was bad, or that someone was not in it, or
some other thing. In ypserv_proc.c, in ypclr_current_map, there is:
dbm_close(current_map);
This is completely bogus. dbm_close is for use with a DBM *, from ndbm,
and yp uses dbm. In addition, current_map is a char *. So i replaced it
with
dbmclose()
and all my problems went away.
Now it may be that the mt. xinu code is real old and obsolete, and the
dbmclose() bug has been fixed at least, but it is all i have had, and the
sun stuff doesn't work. If any one out there needs more info let me know.
But for now, i have a hacked ypbind/ypserv pair that actually work, which
is better than i got from sun ...
BTW, in many other respects 4.0 is not that bad. X11 compiled and ran, as
have many other things. I had to bring uucp over and it appears to work.
So all is not lost, but we still will not run 4.0 on more than this one
machine.
Also i can now boot with ypserv/ypbind/in.named, which i could not do
before. I also suspect that the dbm_close may be in the shipped-with-4.0
ypserv, as in.named had all kinds of problems getting service names until
i ran it with my hacked-up ypserv. It could not get any service name
entries right. The dbm_close failure certainly did cause a lot of weird
flakiness, to say the least.
ron rminnich@{udel.edu,super.org}
------------------------------
Date: Thu, 22 Sep 88 15:08:14 EDT
From: doug@icase.edu (Doug Peterson)
Subject: Sun Domain Name Server & YP
I have recently installed Sun's domain name server patch kit on my system,
and have observed the following abberrant behavior:
A user with accounts on machines in my YP domain and accounts on machines
which are not in any YP domain (i.e. non-Sun's) cannot execute rcp or rsh
when the remote machine is in my YP domain. For example, icase.edu (in YP
domain ICASE) and lrcflx.larc.nasa.gov (which is an indpendent entity on
another network). If the user on lrcflx attempts something like:
rsh icase.edu some_command
he gets 'permission denied'. Both accounts are properly equivalenced via
.rhosts files (This used to work when ICASE was host table driven. The
host tables included lrcflx.) ICASE has always run YP. We are currently
running Sun OS 3.5. The reverse case of rsh works just fine (i.e. from
icase the user types
rsh lrcflx.larc.nasa.gov some_command
which succeeds.
Has anyone else seen this? Sun doesn't have an answer. As far as I can
tell, my nameserver is working correctly.
Doug Peterson
Systems Manager
ICASE
Mail Stop 132C
NASA Langley Research Center
Hampton, VA 23665-5225
(804) 865-4090
FTS: 928-4090
------------------------------
Date: Thu, 22 Sep 88 13:41:59 -0400
From: Dan Lorenzini <sunrise!batman!gcm!dal@sun.com>
Subject: Braindamaged SunOS Install Scripts
The way Sun's install scripts get files off the distribution tapes is
really slow. For each file on the tape, it rewinds to the beginning and
forward spaces until it gets to the desired file. This makes an operation
that should take about half an hour into a several hour wait-a-thon.
Granted, it is something that is not done too often, but I regard it as
poor programming and I expect better from Sun. Does anyone else agree
with me? I would like to get together with some people and rewrite the
scripts that read the tape. These could be substituted during the
installation process to speed things up. If anyone is interested please
send me mail.
Dan Lorenzini
Greenwich Capital Markets
Greenwich CT
203-625-2740
uunet!philabs!gcm!dal
sun!sunrise!batman!gcm!dal
------------------------------
Date: Thu, 22 Sep 88 20:06:24 EDT
From: Jean-Francois Lamy <lamy@ai.utoronto.ca>
Subject: Rudeness of SunOS 4.0 getwd
The SunOS 4.0 getwd has bad manners: let it go for a few days and your
/tmp will get littered with hundreds of .getwdxxxx files (this seems to be
markedly worse on our multi-user Sun 4s than on our diskful/diskless
workstations).
Jean-Francois Lamy lamy@ai.utoronto.ca, uunet!ai.utoronto.ca!lamy
AI Group, Department of Computer Science, University of Toronto, Canada M5S 1A4
------------------------------
Date: 22 Sep 88 20:16:25 GMT
From: boulder!rosing@handies.ucar.edu (Matt Rosing)
Subject: crashing pascal under 4.0
The following program will crash the Pascal compiler running under 4.0:
__________
program crash(input,output);
type t = record
next : array [1..10] of ^t;
end;
var v : t;
begin
with v do
next[0] := nil;
end.
__________
This is the smallest that I can make this program and still crash the
compiler. The following changes will prevent this from crashing:
1) change the identifier "next" to something else.
2) make "next" a scalar.
3) explicitly change the scope with "v." instead of the "with"
statement.
On the bottom of page 10 in the Sun Pascal Programmer's Guide (revision A,
May 88) in small print it is said
"There are three statement keywords 'exit', 'next', and 'return' that are
not reserved words. Instead, these are predefined identifiers and can be
redefined, although it is probably not a good idea to do so. (These are
predefined identifiers to prevent collision with previously written
programs.)"
Statement keywords that can be redefined? I at least understand why this
might cause a problem.
Matt Rosing
rosing@boulder.colorado.edu
... hao!boulder!rosing
------------------------------
Date: Thu, 22 Sep 88 22:24:58 EDT
From: schwartz@shire.cs.psu.edu (Scott Schwartz)
Subject: sun graphics documentation
Does anybody besides me find it strange that Sun supplies no man pages at
all for it's graphics/window libraries? Offhand, I'd say the biggest
advantage to X over SunView (and NeWS too, I presume) is online man pages!
(no :-)) Sun, please take note!
-- Scott Schwartz
[[ I also find it odd. More than that, I find it to be a major hassle. I
have to track down the manual every time I want to do some graphics work.
--wnl ]]
------------------------------
Date: 22 Sep 88 17:58:25 GMT
From: ssc-vax!dmg@beaver.cs.washington.edu (David Geary)
Subject: Need Help with graphics on a Sun
I am in search of a way to do some basic rasterop functions inside of a
SunWindow under Suntools. I've skimmed through the Sun docs (no RTFM,
please), and I don't want to use panels, the notifier, and all that fancy
stuff. All I want is to be able to do the following in a Sun Window under
Suntools:
1) Draw lines.
2) Invert rectangular regions on the screen.
3) Draw graphic text in the window.
I have developed a set of routines that do "panels" on a Unix-based
machine that only had a very rudimentary set of *higher* level graphics
functions such as those above, and now I want to port it to the Suns.
Ideally, I'd prefer someone send me some simple code that shows how to do
any of the above. I promise to be eternally gratefull if anyone would
send me some code. If you have no code, then any info would help.
Thanx,
David Geary, Boeing Aerospace
------------------------------
Date: Thu, 22 Sep 88 13:42:46 EST
From: Robert M. Harris <harris@caf.mit.edu>
Subject: PATRAN on SUN 4s (particularly running 4.0 with X11)?
We are considering using PATRAN (with their X driver) on our new SUN 4
system and would like comments from current users. Please tell us which
version of SunOs you're running under and which version of X you're using.
We are particularly interested in the experinces of PATRAN users running
under 4.0 and X11.
Please respond to the address below.
Thanks,
Bob Harris (harris@caf.mit.edu ...!mit-eddie!mit-caf!harris)
------------------------------
Date: Thu, 22 Sep 88 08:53:50 PDT
From: Reed Hastings <hastings@spar.slb.com>
Subject: FootMouse?
Hi,
I'm trying to set up a foot-controlled mouse. My foot's coordinated
enough (barely) to move the mouse well, but not to click accurately.
---> Does anyone know how to bind L,F, or R keys to mouse clicks at a
very-low level, so that applications can't tell the difference.
Another approach would be to hook up two mice in parallel, one for the
hand, the other for the foot, where the footmouse might not have buttons.
Unfortunately, putting two Sun/Mouse Systems Corp mice in parallel does
not work. Anyone know why (power demands, etc)? Workarounds?
If I ever get this working, I'll summarize the results to the net. If
anyone else has already done this, I'd love to hear about it.
-Reed Hastings
hastings@spar.slb.com
...{amdahl|decwrl|hplabs}!spar!hastings
------------------------------
Date: Thu, 22 Sep 88 14:08:12 EDT
From: flynn@pixel.cps.msu.edu (Patrick Flynn)
Subject: 8mm tape drives vendors?
We are interested in acquiring an Exabyte 8mm drive to do our backups on,
but we don't know which companies are acting as vendors for the drive,
where the driver software comes from, what the prices are, etc. We're
running 4.0 on a 4/280 here, and we'd like to make sure the drivers
support that configuration BEFORE we order anything. I'd appreciate some
email regarding people's experiences with these companies, so we can make
an informed choice. Thx!
--
Pat Flynn, Dept. of Computer Science, Michigan State University (aka Moo U.)
flynn@cpsvax.cps.msu.edu
FLYNN@MSUEGR.BITNET
uunet!frith!flynn
------------------------------
End of SUN-Spots Digest
***********************