Sun-Spots-Request@Rice.edu (William LeFebvre) (10/23/88)
SUN-SPOTS DIGEST Friday, 21 October 1988 Volume 6 : Issue 269 Today's Topics: Re: How many clients on a 3/60 (also configuration advice) Re: Popping up an independent window in SunView Re: text: table full messages Re: Some Benchmark Results Re: nd question Re: .cshrc and $?prompt (2) moving /dev/zero for ftpd to work C preprocessor Bugs Suns and projection systems NFS server vegetable not responding LaTeX and TeX environment on Suns? communication with IBM? Nominating board of directors? 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: 17 Oct 88 20:35 GMT+0100 From: Michael Stumpf <stumpf@cogsys.psychologie.uni-freiburg.dbp.de> Subject: Re: How many clients on a 3/60 (also configuration advice) We are running a very small initial configuration consisting of a 3/60 (327MB) and two diskless 3/50 (SunOS 3.5). Our experience is that the "server" has a quite heavy load, even if the clients are running programs which do not use the disk directly. But most of our applications (e.g. QuintusProlog, FrameMaker, ...) force the clients to page all the time and the server becomes very slow. In fact, if the clients are running such programs, it is very uncomfortable to work at the server's console screen (the mouse is responding to the third click only). Maybe the situation is better if you do not make use of the graphics, maybe just number crunching does not force paging in this way. In our configuration, clearly the disk drive is the point (at the moment, the 3/60 has got only 4MB memory, which is to be expanded to 12MB in the next weeks), but I recommend also at least 12 to 16 MB memory. The disk drive itself is as loud as the cpu - I won't like to have them in my room without some special arrangements. We decided to put most of the cpus and disks (we will have two additional 3/60s with 327 disks in some days) into a special room. In the two rooms beside of this rooms the screens, keyboards and mouses are placed. At the moment, our df looks like this: /dev/sd0a 7608 3461 3386 51% / /dev/sd0f 4897 4240 167 96% /pub.MC68020 /dev/sd0h 139342 119214 6193 95% /usr.MC68020 /dev/sd0d 52836 18273 29279 38% /usr.MC68020/psisun1, that is, there are only 30MB left - and there are only 53MB max. for the ordinary users (clients have 16.5MB swap - I recommend at least 20MB, the server has 50.4MB, which seems to be sufficient also for LISP. Michael. ------------------------------ Date: Mon, 17 Oct 88 16:10:48 PDT From: weiser.pa@xerox.com Subject: Re: Popping up an independent window in SunView See the source code for my game sdi (available for anonymous ftp from host arisia in pub/sdi.shar, or in the Sun User Group tape, volume 2) for code which pops up windows. The file 'helpers.c', and the routines `popup_warning', 'popup_msg', `easy_pop', and `easy_warn' are good places to start. Or, if your sun has the sdi game installed, do this: run the game, go to the "things to read" pie menu and select "source", then use textedit search for one of the names above. -mark ------------------------------ Date: Mon, 17 Oct 88 20:38:54 PDT From: walker%skat.usc.edu@oberon.usc.edu (Michael D. Walker) Subject: Re: text: table full messages Reference: v6n253 With regard to text: table full messages: This is common (under SunOS 3.n). X isn't more 'demanding' per se, but X will use up more text table entries under normal usage (i.e., several different programs). Each different program image (X or otherwise) uses up a text table entry. Suntools doesn't have this problem because most SunView programs are built into suntools (using the toolmerge program). Do an ls -l /usr/bin and you will see that programs like shelltool are linked to suntools. Thus, suntools uses only one text table entry except for programs running in windows and tools not merged into suntools. Why do this? Because without shared libraries this allows one copy of the SunView libraries to be shared by several "programs". The fix? Under SunOS 3.n, reconfigure your system with maxusers increased from 4 to 10 (I don't know about licensing problems over this). Under SunOS 4.0, this problem goes away. I guess it just allocates space as needed. [[ That and the fact that 4.0 has shared libraries. --wnl ]] Mike Walker :-] arpanet: walker@oberon.usc.edu USC University Computing Services uucp: sdcrdcf!oberon!walker Phone: (213) 743-2957 bitnet: walker@kylara ------------------------------ Date: 18 Oct 88 03:17:07 GMT From: werme@alliant.com (Ric Werme) Subject: Re: Some Benchmark Results Reference: v6n257 Dan Ehrlich <ehrlich@blitz.cs.psu.edu> notes: > What is interesting to note is that a Sun 4/260 running 3.2 is > significantly faster that the same hardware running 4.0. Anyone have any > thoughts... This generally means that the clock interrupt routine in 4.0 is taking more time. (And I hope doing more....) Back in my PDP-10 days, I could see the PI 7 light on the front panel glow faintly, taking time away from everyone. I bet the light on your virtual front panel is glowing a little brighter these days. Eric J Werme uucp: decvax!linus!alliant Phone: 603-673-3993 ------------------------------ Date: Tue, 18 Oct 88 11:09:23 EDT From: Bennett Todd <bet@bent.mc.duke.edu> Subject: Re: nd question On re-reading I noticed a couple of things I left out. First, while I mentioned that I initially installed 4 hosts, I gave as an example our current /etc/nd.local, which shows more. Sorry about the inconsistancy. Second, I forgot to mention that before you boot the critter you are going to have to play around with /mnt/dev/MAKEDEV (assuming you are frobbing the new /dev/ndl0 root on /mnt) to build the device files on the new root. For a diskless 3/60 the following should cover most of them: # cd /mnt/etc # ./MAKEDEV std # ./MAKEDEV nd # ./MAKEDEV pty0 # ./MAKEDEV pty1 # ./MAKEDEV pty2 # ./MAKEDEV bwtwo0 # ./MAKEDEV win0 # ./MAKEDEV win1 # ./MAKEDEV win2 # ./MAKEDEV win3 I probably missed a few, and might have something wrong in there, but I think that is what we are using right now, and it seems to work.... -Bennett bet@orion.mc.duke.edu ------------------------------ Date: Mon, 17 Oct 88 10:10:44 EDT From: Bennett Todd <bet@orion.mc.duke.edu> Subject: Re: .cshrc and $?prompt (1) Darin McGrew <McGrew@Sun.COM> writes: >Bennett Todd <bent!bet@mcnc.org> writes: >>..., it is considered wise to start off .cshrc with the following >>incantation: >> >> if ($?prompt == 0) then >> exit >> endif >> >>This will cause non-interactive shells to exit .cshrc immediately... > >The exit can cause scripts to exit prematurely. A better solution... > > if ( $?prompt ) then > # the following applies only to interactive shells... > endif Sounds like a reasonable argument... but I don't see quite how the exit makes scripts exit prematurely. I just tried a little one and it in fact didn't exit prematurely. I thought exit was interpreted relative to the script and not the surrounding csh invocation. That's how it seems to behave when I test it. Of course, I really had no idea until I tested since csh isn't viewed as a programming language. Give me a /bin/sh with ecsh-style history editing and job control and I'll rm /bin/csh in a flash.... I expect I'll keep mine the way it is; it works right (even if it shouldn't) and is much quicker (as well as, in my opinion, cleaner) than having the whole body under an if(). -Bennett bet@orion.mc.duke.edu ------------------------------ Date: Tue, 18 Oct 88 13:47:47 PDT From: Darin McGrew <mcgrew@sun.com> Subject: Re: .cshrc and $?prompt (2) In a recent posting, I recommended using if ( $?prompt ) then ... else in one's .cshrc, rather than using if ( ! $?prompt ) exit to avoid premature termination of scripts. Several of you have corrected me. Both of the above methods work on SunOS Release 4.0, although I'm sure I've seen the second method fail elsewhere. The second method is faster because the rest of the .cshrc is not parsed. Darin McGrew mcgrew@Sun.COM I speak for myself, not for my employer. ------------------------------ Date: Mon, 17 Oct 88 17:45:25 CDT From: dave@wubios.wustl.edu (Dave Camp) Subject: moving /dev/zero for ftpd to work Thank you for all your responses on getting anonymous ftp to work under SunOS 4.0. I have written a dumb replacement for 'ls' so that I get a minimal functionality, but I would like to use the real one. One of the requirements was that I copy the device driver /dev/zero onto a subdirectory of ~ftp/ . Would someone please tell me how this is done? If I simply use 'cp' I get an unending file full of zeroes :-( Thank you, -David- [[ Yes, because /dev/zero is a device. You can only create it with "mknod". Find the major and minor device ID with "ls -l /dev/zero". Instead of a size, you will see "3, 12". You will also see a "c" in column 1, which indicates "character special". "cd" to the directory you wish to create zero in (probably "~ftp/dev") and, as root, type the command "/etc/mknod zero c 3 12". That'll do it. Read mknod(8) for more details. --wnl ]] (314) 362-3635 Mr. David J. Camp Room 1108D ^ Box 8067, Biostatistics 706 South Euclid < * > Washington University Medical School v 660 South Euclid Bitnet: david@wubios.wustl Saint Louis, MO 63110 Internet: david%wubios@wucs1.wustl.edu uucp: uunet!wucs1!wubios!david ------------------------------ Date: Mon, 17 Oct 88 15:24:04 cdt From: wsmith@m.cs.uiuc.edu (Bill Smith) Subject: C preprocessor Bugs The C preprocessor on the sun does not accept expressions as an arguments to #ifdef or #ifndef. In specific, the Andrew code uses expressions such as #ifdef 0 #endif or #ifdef (DEBUG) #endif Neither of these may be compiled successfully under 4.0 SunOs with out editing to be #if 0 or #ifdef DEBUG. [[ Sorry, but K&R, page 208 is pretty clear about the syntax of a #ifdef: it takes an identifier and not an expression as an argument. Neither "0" nor "(DEBUG)" are proper identifier names. A little bit of playing indicates that the C compiler under 3.x doesn't complain about them, but always assumes that the condition is "false". --wnl ]] Bill Smith uiucdcs!wsmith wsmith@cs.uiuc.edu ------------------------------ Date: Mon, 17 Oct 88 10:57:07 MST From: mc%miranda.uucp@moc.jpl.nasa.gov (Mike Caplinger) Subject: Suns and projection systems I've had some experience with projecting Sun color displays using the Electrohome CAD/CAM projector. The Electrohome can adjust to a wide range of scanning rates and resolutions, and it can handle a Sun's output. But you won't find the results very satisfying. The image never gets better than blurry, even with an on-site consultant constantly adjusting and tuning the thing. The projector is pretty unstable as far as its fine settings goes, and requires retuning practically every time you turn it on. Normal sized text is nearly unreadable; for the demos I was doing, I found I had to use about a 20-point bold font for anyone to even make a stab at reading the display. And as with most curved-screen projection systems, anybody more than a few feet off the optic axis can't see much of anything. All in all, I found the experience frustrating and unsatisfactory. Another alternative for small audiences is to slave multiple monitors to the same machine and spread them around the room. Sun provided us with some kind of specially-built video amplifier for the purpose; I've no idea if such things are commercially available. We were doing these demos in the Bay Area so you might check with the San Francisco sales office for more info. I did all this while I was with Bellcore, so maybe you have to be as big as the phone company to get such consideration. Mike Caplinger, mc@moc.jpl.nasa.gov ------------------------------ Date: 17 Oct 88 17:05:21 GMT From: ksr!benson@uunet.uu.net (Benson Margulies) Subject: NFS server vegetable not responding I am using a Sun 3/60 running SunOS 3.5, diskless, with quite a lot of memory. I talk to two NFS servers, a Sun 4/260 and a Sun 3/160. Every very small interval (a minute to five minutes), I get NFS server foo not responding still trying NFS server foo ok there is no pause between them. I get these for both servers. I have my packet size in fstab turned down to 1024. If I set timeo to a large value, my machine hangs for the amount of time I specify, but I get less messages. If I set timeo to a small value (currently its 5), I get plenty of messages but I can otherwise more-or-less get work done. Can anyone elucidate and provide some palliative? --benson Benson I. Margulies Kendall Square Research Corp. harvard!ksr!benson benson@ksr.com ------------------------------ Date: 17 Oct 88 21:28:27 GMT From: bonham@ucalgary.ca (Michael Bonham) Subject: LaTeX and TeX environment on Suns? I have recently been asked to update our TeX and LaTeX environment under SunOS 4.0. I'm sure the subjects have come up before, but for the last couple of months I haven't seen any discussions in the TeXhax or Sunspots digests regarding the following questions: 1) Is there an undump for SunOS 4.0? Where is it or how do I build one? We have both Sun 3's and 4's and wish to support LaTeX on both. Right now tex and latex are shell scripts which invoke virtex but it would be nice to have a binary, undumped core image version. [[ The sun-spots archives has a version of undump for 4.0. It is in the "sun-source" area and is called "undump4.shar" (11574 bytes). It can be retrieved via anonymous FTP from the host "titan.rice.edu" or via the archive server. For more information about the archive server, send a mail message containing the word "help" to the address "archive-server@rice.edu". Because of a bug in the 4.0 include files, you may need to compile undump with "-DFPU". --wnl ]] 2) Is there a dvi previewer which doesn't use voluminous, obsolete .pxl fonts? We have dvisun, dvitool, dvipage, texsun (and probably a few others) kicking around, but these are always bombing out when one or another font family, size or style is not installed. If .pxl fonts are obsolete, why do the tools require it? We currently don't run X windows, but do have NeWS. Does anybody use dvi2ps to produce something NeWS can display interactively? [[ The newer version of dvitool (from the VorTeX package) uses PK files and handles missing font files gracefully. The version we have is 2.0. --wnl ]] Please reply by electronic mail to bonham@ucalgary.ca Thank you! ------------------------------ Date: Thu, 20 Oct 88 14:09:16 SET From: J-P Lauque <C4@ESOC.BITNET> Subject: communication with IBM? Here at ESOC (European Space Operation Center) would like to connect SUN to the IBM channel in order to do program to program communication. Has some body done this? Reply to : JP LAUQUE / ECD / 5 robert bosch strasse D6100 darmstadt tel germany (39) 6151 886 537 ------------------------------ Date: Mon, 17 Oct 88 22:21:48 EDT From: dpz@hardees.rutgers.edu (David P. Zimmerman) Subject: Nominating board of directors? G'day all - I've been over and over this notice of Sun's 27 Oct 1988 annual stockholder meeting, and although it details quite clearly how to cast a vote for a nominee for the board of directors, and even how to withhold a vote, it isn't entirely obvious to me how these people get nominated in the first place. Anyone with any insight? From what I _can_ see: - the term of a director is essentially unlimited - for a willful vacancy, the board can keep its ranks closed by taking proxy votes and applying them to their replacement nominee and in fact, "In the event that additional persons are nominated for election as directors, the proxy holders intend to vote all proxies received by them in such a manner in accordance with cumulative voting as will assure the election of as many of the [current director] nominees listed below as possible and, in such event, the specific nominees to be voted for will be determined by the proxy holders." Hmm. Sounds a little paranoid to me. Either that, or overly protective? Might be interesting to get a student type (nudge nudge wink wink :-) in that crowd.... David ------------------------------ End of SUN-Spots Digest ***********************