Sun-Spots Digest, v7n3
William LeFebvre
Sun-Spots-Request at Rice.edu
Fri Nov 4 02:32:46 AEST 1988
SUN-SPOTS DIGEST Wednesday, 2 November 1988 Volume 7 : Issue 3
Today's Topics:
Re: possible compiler bug on Sun 4, sys4-3.2
Re: nd question
Re: Problem with Textsw windows
Re: Hard vs Soft Mounts
Re: Project Planning Software
Re: 50's and SunOs 4.0
Re: locking down systems
Re: Frame Maker to (di)troff translator
Framemaker
sun 3/50 memory upgrade
Mail problem
Memory for Sun 4's
Wren V 94181 SCSI configuration?
fast communication over RS232 on Sun 3/60?
Differences among Sun4's, and what version are you running?
Starting Sunview in SunOS 4.0?
SunLink SNA 3270: How to transfer a directory with ibmftp?
SunLink 3270?
Postscript Interpreter?
Send contributions to: sun-spots at rice.edu
Send subscription add/delete requests to: sun-spots-request at 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 at rice.edu" or mail the word "help" to the same address
for more information.
----------------------------------------------------------------------
Date: Wed, 26 Oct 88 10:30:02 PDT
From: stevo at judy.jpl.nasa.gov (Steve Groom)
Subject: Re: possible compiler bug on Sun 4, sys4-3.2
Reference: v6n274
>From: mc%miranda.uucp at moc.jpl.nasa.gov (Mike Caplinger)
>
>Consider the following C program:
>
>typedef union {
> int *i;
> char *s;
> float *f;
>} vp;
>
>main() {
> int i;
> char *s = "hi there";
> float f;
>
> printf("%x %x %x\n", &i, s, &f);
> foo(&i);
> foo(s);
> foo(&f);
>}
>
>foo(a)
>vp a;
>{
> printf("%d %x %x %x\n", sizeof(a), a.i, a.s, a.f);
>}
This program is broken. Lint will hint at it, but it's a subtle and
oft-ignored error message - 'foo, arg. 1 used inconsistently...' Broken
programs are not guaranteed to be portable.
The problem is that 'main' passes one type of argument and the 'foo'
expects another. A union is not the same type as anything but an
identical union.
Structures and unions, when passed as arguments, are guaranteed to be
passed by value. However, all that means is that the function gets a copy
of the value, and not a reference to the caller's copy. It does not mean
that you can pass a value of one type and expect to receive a value of
another type.
In the 680x0 compilers, pass-by-value was implemented by stuffing the
value onto the stack. This applied to structures and unions as well.
However, the Sun4 C compilers pass structure and union arguments in a
different way. The pass-by-value semantics are preserved, but the method
of deliberatly mismatching parameter types to perform implicit type
conversions no longer works.
By the way, there is a discussion of commonly used non-portable constructs
such as these in "Porting Software to SPARC Systems," Sun P/N 800-1596-10.
In particular, the section "Passing Mismatched Parameter Types" will
probably be of interest. This manual should have been included with your
Sun4.
Just to beat the subject to death, here is a fixed version of the above
program, which works as expected on both Sun3's and Sun4's under SunOS
4.0.
typedef union {
int *i;
char *s;
float *f;
} vp;
main() {
int i;
char *s = "hi there";
float f;
vp u;
printf("%x %x %x\n", &i, s, &f);
u.i = &i;
foo(u);
u.s = s;
foo(u);
u.f = &f;
foo(u);
}
foo(a)
vp a;
{
printf("%d %x %x %x\n", sizeof(a), a.i, a.s, a.f);
a.i = 0;
}
-steve
[[ Thanks to Richard O'Keefe, Mike McCants, and others for also making
this point. --wnl ]]
Steve Groom, Jet Propulsion Laboratory, Pasadena, CA 91109
Internet: stevo at elroy.jpl.nasa.gov UUCP: {ames,cit-vax}!elroy!stevo
Disclaimer: (thick German accent) "I know noothingg! Noothingg!"
------------------------------
Date: Thu, 27 Oct 88 06:55:00 EDT
From: Bennett Todd <bet at bent.mc.duke.edu>
Subject: Re: nd question
I stand corrected, you in fact don't have to repartition, just re-mkfs to
leave some spare room at the end off an extant partition. Indeed, if you
are out of free partitions (as someone else pointed out, the reason for
Sun suggesting the 'c' partition) that is all you will really want to do.
However, since the take-it-down/back-it-up/mkfs/restore cycle is all the
work, that is really the same effort as repartitioning (I discovered to my
relief that using stand/diag to repartition a disk doesn't trash any
filesystems whose partitions remain the same, which is as it should be).
Since I so far haven't run out of partitions I like the protection of
having to get the partition sizes and boundaries right *once*, then
working nd.local off a separate designated partition so as to protect
filesystems on other partitions from demolition due to typos.
-Bennett
bet at orion.mc.duke.edu (Reply *still* doesn't work, however I have
heard rumors of the existence of a genuine sendmail wizard in these
parts. I hope to track him down and see if I can't bribe him into
helping us out).
------------------------------
Date: Thu, 27 Oct 88 08:10:11 EDT
From: Chuck Musciano <chuck at trantor.harris-atd.com>
Subject: Re: Problem with Textsw windows
I had a problem similar to this in contool, and solved it by doing deletes
at the appropriate time. Since you are clearing the whole window, why
don't you just call textsw_reset() to clear the whole thing out? This is
more likely to free the internal memory buffers. You can, I think, fix
things so you don't get an annoying confirmer when you call
textsw_reset().
Chuck Musciano
Advanced Technology Department
Harris Corporation
(407) 727-6131
ARPA: chuck at trantor.harris-atd.com
------------------------------
Date: Thu, 27 Oct 88 13:05:14 BST
From: James Davenport <jhd%maths.bath.ac.uk at nss.cs.ucl.ac.uk>
Subject: Re: Hard vs Soft Mounts
In v6n272, jimc at math.ucla.edu says:
> forever. Thus if the client is writing, the write absolutely will take
> place (after the Sun FE shows up, replaces the board, etc. on the
> server), provided the client isn't rebooted by an impatient user.
This is almost true. It will NOT take place (ESTALE being returned) if the
FE replaces the disc. The new disc will have a different file system
identification on it (even if it mounted in the same place). If you think
about it, this is an inherent consequence of the implementation: the inode
number of a file is part of the file handle, and restore does not preserve
inode numbers.
James Davenport
jhd at maths.bath.ac.uk [via any gateway you can think of]
------------------------------
Date: Thu, 27 Oct 88 10:05:24 CDT
From: schultz at mmm.3m.com (John C Schultz)
Subject: Re: Project Planning Software
>I am using the SunTrac program from Sun to do project planning with....
>I would definately recommend it....
Unless SunTrac has improved a lot since I investigated, it is missing a
rather essential feature for many projects. It does not permit tracking
individuals. For example there seems to be no way to keep from assigning
Mr. Foo Bar to several hundred hours per month.
Along similar lines, there is no way to put varying personnel costs in
(senior personnel "cost" more).
Such accounting is required by many large companies and the Federal
Government. I was forced to use a spreadsheet (SC) which was
non-trivial.--
------------------------------
Date: Thu, 27 Oct 88 11:44:55 -0400
From: Henry B.J. Krempel <krempel at pacrat.npac.syr.edu>
Subject: Re: 50's and SunOs 4.0
I've heard a lot of talk about 3/50's not being able to run 4.0 well due
to memory limitations. I run a cluster with 6 50's and a number of other
8MB systems on it, and we don't have any speed problems. What's worse is,
the stations are all running GENERIC kernels!
Henry B. J. Krempel <krempel at pacrat.npac.syr.edu>
Syracuse University
250 Machinery Hall
Syracuse, N.Y. 13244
------------------------------
Date: Thu, 27 Oct 88 01:01:27 EDT
From: attcan!utzoo!henry at uunet.uu.net
Subject: Re: locking down systems
>... How would one go about
>chaining a workstation down? Are there products on the market? R.B.
Stopping a really determined thief is next to impossible. However, you
can stop the less motivated types merely by making life more difficult for
them. You can buy kits that are basically a cable, a lock, and two large
and very sticky pads. One goes on a desk or something else nice and
solid, the other goes on the machine. This has the advantage that it's
not tied to a specific physical design, like some of the PC-lockdown
products are.
One very sneaky trick that used to be quite useful was to just chain the
power cord to something. Easy enough to do with a light chain (i.e. one
that could wrap tightly around the cord near the plug) and a small padlock
or two. Unfortunately, practically all hardware nowadays uses removable
cords, so this doesn't work any more.
Henry Spencer at U of Toronto Zoology
uunet!attcan!utzoo!henry henry at zoo.toronto.edu
------------------------------
Date: Thu, 27 Oct 88 08:42:59 EDT
From: scott at cs.rochester.edu
Subject: Re: Frame Maker to (di)troff translator
We, too, are interested in a WYSIWYG document system that could dump its
files into a text description language that could be edited on Ascii
terminals. Dave FitzGerald asks specifically about Frame Maker to
ditroff, but we would be equally happy to see any of the commercial
systems dump to latex, troff with macros, or any other description
language that a human being would be willing to use outside the WYSIWYG
world. This appears to be a very difficult problem; producing raw tex or
troff is probably not too tough, but producing something you'd be willing
to edit by hand is considerably harder.
Some commercial systems are able to take input *from* tex or troff, but
none seem to be able to produce it. I think the first vendor to offer the
inverse function is going to make a mint.
------------------------------
Date: Thu, 27 Oct 88 04:40:32 EDT
From: g73!grant at lownlab.harvard.edu
Subject: Framemaker
As an up to now satisfied user of Framemaker I just wanted to point out a
fairly serious problem I ran into today. It does not seem posible to make
"floating anchors" to text figures within a document. What is a floating
anchor? If you are composing a document with embedded figures you often
want to anchor the figures as close as possible to where you reference the
figure ie. " see Figure 3" and have the figure bump up to this reference
at the first available chunk of space. Framemaker allows only fixed
offset points from the original reference. Where this is really a problem
is when you edit any text in front of the reference to the figure--you end
up with all sorts of holes in the text. What one has to do is fiddle
around with all of the text cut and paste, cut and paste, cut and paste
around the figure. The people at Frame said they were aware of the
problem but... Anybody writing a grant on this product will be tearing
their hair out, as I am now.
....Grant Balkema at Boston College grant at g73.harvard.edu
------------------------------
Date: Wed, 26 Oct 88 16:38:02 PDT
From: pjg at unrvax.unr.edu (Paul Graham)
Subject: sun 3/50 memory upgrade
i assume people are starting to get their postcards from helios about the
memory upgrade.
i had heard that they were looking for someone to do board maint. when you
installed this and became ineligible for sun maint. at that time it was
supposed to involve some soldering to make the upgrade. now it appears to
be some plug-in daughter card (in the cpu socket?) that regular users can
pull off.
a friend notes that this is cheaper than 4 meg. for his 3/60. such a deal.
maybe motorola will take a stab at maintaining these things.
pjg at solstice.unr.edu
------------------------------
Date: Wed, 26 Oct 88 20:25:05 PDT
From: versatc!tran at sun.com
Subject: Mail problem
We recently decided to mount our mailhost's /usr/spool/mail directory to
each and every subsidiary machine's /usr/spool/mail {or /var/spool/mail on
the SUN 4.0 system}, and we noticed that some of the mail files under
/usr/spool/mail/user_name was created with the ownership of NOBODY and
group WHEEL. As a result, mail was sent but users cannot read them.
Any ideas what went wrong?
Tony Tran
Versatec, Inc.
------------------------------
Date: Tue, 25 Oct 88 21:47:41 EDT
From: hsi!stepstone.com!aad at uunet.uu.net (Anthony A. Datri)
Subject: Memory for Sun 4's
As for memory and SunOs 4.0, I called Sun about prices on Sun 4's and
RoadRunners, and it seems that they won't sell you a roadrunner with only
4 meg, since SunOS 4.0 is too slow on it. They quoted $3400 more for the
additional 4 meg -- even with a 40% discount, can better be had?
Anthony A. Datri @SysAdmin(Stepstone Corporation) aad at stepston.com, stpstn!aad
------------------------------
Date: Mon, 24 Oct 88 17:53:01 EDT
From: <wzd at geovision.uucp>
Subject: Wren V 94181 SCSI configuration?
Has anyone out there configured a Wren V 94181(702mb) disk on a Sun 3/60.
If so, I would appreciate the cylinder/head/track information.
Thanks
Walter Dolegowski
------------------------------
Date: Thu, 27 Oct 88 03:02:12 EDT
From: mboasson at ajpo.sei.cmu.edu
Subject: fast communication over RS232 on Sun 3/60?
We try to use our Sun 3/60 in an almost real-time application.
Communication with other machines takes place over one or both RS232 ports
at 19200 baud. Unfortunately, Unix scheduling slows the actual datarate to
something like 1500 baud. Thta is, with relatively short messages: < 100
bytes. It is impossible to communicate with long messages through
buffering. Is there a trick to get around the scheduler - or anything
else that will speed up communication? When running this application, the
machine does not have to run other programs.
mboasson at ajpo.sei.cmu.edu (Maarten Boasson)
------------------------------
Date: Tue, 25 Oct 88 21:51:51 EDT
From: hsi!stepstone.com!aad at uunet.uu.net (Anthony A. Datri)
Subject: Differences among Sun4's, and what version are you running?
I'm investigating obtaining a Sun4 for purposes of product support, and I
have a dim memory of people saying that the 4/110 (the cheapest, right?)
is somehow different than the other models. How wrong am I?
Also, I'd like an idea of which version of SunOS Sun4 owners are running
-- are you running the 3.2 version, 4.0, or something I haven't heard
about yet?
Anthony A. Datri @SysAdmin(Stepstone Corporation) stpstn!aad
------------------------------
Date: 27 Oct 88 08:19:07 EDT (Thu)
From: osu-cis!n8emr!lwv%killer.Central at sun.com (Larry W. Virden)
Subject: Starting Sunview in SunOS 4.0?
I am attempting to start a program up in my .sunview file which needs a
parameter with embedded blanks. For instance, it might look like:
less '-P%f Hit return to continue' myfile
When I put something like this at the end of a shelltool command line, I
get the shelltool window, a flash of a set of error msgs from Less, who
appears to be attempting to open files -P%f, Hit, return, to, and
continue, and then I get myfile opened.
Yet, if I say " fgrep less .sunview | ksh" I get the window as I really need it.
Any suggestions as to how best to handle this problem?
------------------------------
Date: Thu, 27 Oct 88 17:12:01 SET
From: Danielle Heinzer <ESC1298 at ESOC.BITNET>
Subject: SunLink SNA 3270: How to transfer a directory with ibmftp?
Is there any possibility to transfer more than one file at a time between
the IBM MainFrame and the SUN workstation via "ibmftp" ?
SunOS 3.2, SunLink SNA Rel.5.0
Danielle Heinzer
ESA Computer Department/Computer Services
European Space Operations Centre
Robert-Bosch-str. 5
6100 Darmstadt
West-Germany
Tel int : 49-6151-886540
------------------------------
Date: Thu, 27 Oct 88 12:25:51 EDT
From: larryd%bnrmtl.UUCP at larry.mcrcim.mcgill.edu (Larry Dunkelman)
Subject: SunLink 3270?
Does anyone have any experience using SunLink 3270 ? I would appreciate
any information regarding its use, hardware required and whether the
software runs on 4.0
Larry Dunkelman Bell-Northern Research (Montreal)
bnrmtl!larryd at larry.mcrcim.mcgill.edu
------------------------------
Date: Thu, 27 Oct 88 08:57:08 SST
From: TNG TH <ISSTTH at NUSVM.BITNET>
Subject: Postscript Interpreter?
Does anyone know if there is a public domain Postscript Interpreter
running on SUN with full source?
[[ There's "ghostscript" from the gnu project. Sorry, I don't know any
more about it. --wnl ]]
------------------------------
End of SUN-Spots Digest
***********************
More information about the Comp.sys.sun
mailing list