Sun-Spots-Request@Rice.edu (William LeFebvre) (10/04/88)
SUN-SPOTS DIGEST Sunday, 2 October 1988 Volume 6 : Issue 247 Today's Topics: Re: Force periodic password changes Re: TAAC-1 application accellerator Re: Getting a dead 386i serviced Re: Problem: UUCP and detached processes Re: Need Help with graphics on a Sun Re: cc:Fatal error in ccom: Killed Problems with VME address spaces Booting SunOS 4.0 Collected info on Sun 386i How to make a device driver awake with Ctrl-C? Follow up question on Bigger Icons? interactive spelling checker? gettimeofday() bug? Coral-66? 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, 29 Sep 88 02:45:06 EDT From: attcan!utzoo!henry@uunet.UU.NET Subject: Re: Force periodic password changes > Does anyone out there have software to force users to change password > every so often on a SUN? Are you sure you want to do this? It is likely to result in much poorer passwords, especially if it comes as a surprise at login time. (Grampp and Morris, in the second Unix BLTJ, say: "...the most incredibly silly passwords tend to be found on systems equipped with password aging".) Our own policy is to *encourage* people to change password every six months or so -- a simple shell script sends them mail if they haven't changed in that long -- but we don't force it on them. Henry Spencer at U of Toronto Zoology uunet!attcan!utzoo!henry henry@zoo.toronto.edu ------------------------------ Date: Thu, 29 Sep 88 12:22:49 EDT From: sorensen@hstbme.mit.edu (Gregory Sorensen) Subject: Re: TAAC-1 application accellerator We have a beta release TAAC in a Sun 4/260. I've only spent a month or two programming it, so I am probably on the worst part of the learning curve, but let me give you my impressions of it. First, it does provide nice color. Our 24 bit images are easily displayed, and it is easy to take R,G, and B bytes and load them in and get a display. And, from their demos, it appears to be nice and fast. However, in my opinion, it has two extremely serious flaws. First, the display is "video keyed", meaning two things: first, there is a sizeable subset of colors which you cannot display, as it is used for keying. For many people, this is not a problem, but I want to display arbitrary data sets, with potentially arbitrary colors, and the TAAC will not do this. second, you cannot have another non-TAAC window open using that color -- so in effect, you lose part of the color spectrum -- negating some of the TAACs benefits, in my opinion. Second, it does not multitask. This means only one process can use the TAAC at once -- so, I can't have two display windows open, nor can I even start two jobs, or have two diskless workstations utilize the TAAC at the same time. A non-fatal flaw is how difficult it is to program. While the software library is getting better, even with the new release it is not free from assembly-code like programming. Also, as of right now, it does not handle double precision arguments. And, of course, it does no I/O itself, so you have to write special code to handle all of that. You have to treat it as a very fast, very picky asynchronous processor that is kinda dumb. So, in summary, we have one, and so I am not going to let those 12 MFLOPS (or so they claim!) sit around, but if I was going to spend $75K for a Sun4 with a TAAC, I might see if I could get a deal on a Stellar GS1000 instead. After all, it supports X-windows... Greg Sorensen sorensen@hstbme.mit.edu or sorensen@athena.mit.edu PS: I would happily be proven wrong on any of my points above, but even after asking the TAAC gurus at Sun, they say that's the way it is. Then they mutter promises about the TAAC-2.... ------------------------------ Date: Thu, 29 Sep 88 10:19:26 PDT From: paladin@sun.com (Joe Davis) Subject: Re: Getting a dead 386i serviced Dan Davison writes: >Calling Sun (local TSEs and sales, and two separate calls to the hotline) >has resulted in a machine dead for the past 8 or 9 weeks. The local >hardware people know nothing about the machine; apparently the Right Coast >Sun people do not talk to the Left Coast Sun people. Dan, I have two questions for you, (1) are you under warranty? and if not (2) are you under contract. You stated the the system failed after only 2 or 3 weeks, this would put you under warranty and someone from the Western Support Center would be helping you after you had put a service call through the Sun 800 number. If your on onsite contract support a Sun FSE should have come onsite and (1) checked the SCSI fuse (2)checked the SCSI terminator (3) replaced the CPU/Mother board. If warranty, then you will have to send in your system for repair with a turn-around of about 3 to 10 days. If you have a service order open please email it to me and I will look into it. Joe Davis Western Support Center Engineer Sun Microsystems, Inc. Milpitas, Ca. ------------------------------ Date: Wed, 28 Sep 88 12:03:45 PDT From: Vaughan Pratt <coraki!pratt@sun.com> Subject: Re: Problem: UUCP and detached processes >Scene is this: User logs in over the modem, starts up a process in the >background, then exits. Uucp tries to dial out, and gets a NO DEVICE >error.... >I want UUCP to get in. Is there any easy way of doing this? Start up your background process from another session, e.g. by rlogin'ing (rloggingin?) back to yourself. -v ------------------------------ Date: Thu, 29 Sep 88 07:42:11 EDT From: Chuck Musciano <chuck@trantor.harris-atd.com> Subject: Re: Need Help with graphics on a Sun If all you want to do is draw simple objects, create a window with a canvas in it, and start drawing using the pw_* routines. Unfortunately, you will have to read the manuals; no one ever got off easy when writing a windowed application. Forget the notifier stuff; just read the chapter on canvases, and the chapter in the pixrect manual on the pr_* routines. There are pw_* analogs for canvases. Really, you should take the time to learn the new window system. It is fairly powerful, not too hard to learn, and your end users will benefit from your windows expretise in the form of better user interfaces. Chuck Musciano Advanced Technology Department Harris Corporation (407) 727-6131 ARPA: chuck@trantor.harris-atd.com ------------------------------ Date: 28 Sep 88 00:05:04 GMT From: munnari!spinifex.eecs.unsw.oz.au!gordon@uunet.uu.net (Gordon Rowell) Subject: Re: cc:Fatal error in ccom: Killed Reference: v6n235 I had the same problem with our Sun 3/50, notably when using dbx within X windows and trying to do a make from inside dbx. If you only have one or two xterm windows up, no problem. With more than that, cc or cpp die continually. The death is usually accompanied by a "Text: table is full" message on the console window. The solution: I patched the kernel to allow more effective users on the Sun. The default value for MAXUSERS is 4, and I changed this to 10 and re-installed the kernel. I have had no further problems, and we can now happily have someone on the console with many windows and two remote users over the ethernet (although the Sun is somewhat slow as a result of all the activity). I would suggest this patch to all heavy X users. To do it: Read your "Installing UNIX on the Sun workstation", notably Section 8: "Configuring the System Kernel". It is relatively painless for simple modifications such as this (all you need to do is make a new template file and change the maxusers value from 4 to 10). You do not need a source distribution to do this. Gordon Rowell (gordon@spinifex.eecs.unsw.oz) Department of Computer Science, University of N.S.W., Sydney, Australia ------------------------------ Date: Tue, 27 Sep 88 13:38:00 GMT From: denis <mcvax!gec-rl-hrc.co.uk!denis@uunet.uu.net> Subject: Problems with VME address spaces We are considering writing a device driver for Sun 3 and 4 under OS3 and OS4. Sun told me it should be no problem. According to a recent posting from Scott Williamson <scott@memex.co.uk> there are incompatibilities in VME access between Sun 3 and Sun 4. Anybody else had any (good or bad) experience with this? Denis Howe uunet!mcvax!ukc!gec-rl-hrc!denis G15 GEC Hirst Research Centre denis@uk.co.gec-rl-hrc East La. Wembley MDDX HA9 7PP UK +44 (1) 908 9220 ------------------------------ Date: Thu, 29 Sep 88 11:14:10 PDT From: Richard Smith <smiddy@spam.istc.sri.com> Subject: Booting SunOS 4.0 Sounds to me like someone out on your net is running some derivative of the Wollengong software. We had this problem here, and Tim Ehrhart and I chased it for months. The basic problem is that the Sun client sends out a broadcast RPC request to the portmapper port (on all machines on the net). The RPC request is the portmapper "indirect" RPC call. That is, the client is asking the portmapper to do all of the work of talking to some other RPC program, in this case the "bootparamd" program. The client then expects the portmapper to return the information supplied by the other RPC program, if and only if the indirect call has succeeded. There is no reply allowed if the indirect call did not work. The problem is that Wollengong only has a partial implementation of the portmapper service, and it does not support indirect RPC calls. It returns an error, and the poor client freaks and dies. This is not really Wollengong's fault, btw, as this bug was in the original software we are running at SRI. Dave Kashtan was kind enough to fix it for us. We have told Sun about this, so SOMEONE there knows. It sounds like the info is just taking a little time to filter out. Thanks, smiddy@spam.istc.sri.com PS to Jerry. I don't think Sun has bothered to tell you folks about this either, so I thought I would CC this to you. I don't know if I am talking to the right person, but who ever ends up working on this can feel free to contact me for more info. ------------------------------ Date: Wed, 28 Sep 88 16:56:01 EDT From: gld@cunixd.cc.columbia.edu (Gary L Dare) Subject: Collected info on Sun 386i This machine has just come to my attention, and I'm a bit behind on any discussion that's taken place. Has someone collected a file of postings regarding this machine? I'm concerned about it's behaviour as: 1. A Sun station 2. A PC 3. How good is the DOS they run under UNIX? 4. Is the networking any good? Your responses will be gratefully appreciated. gld PS: RSVP to this account. Thank you! Gary L. Dare > dare@eevlsi.ee.columbia.EDU > gld@cunixd.cc.columbia.EDU @cunixc.BITNET ------------------------------ Date: Tue, 27 Sep 88 11:01:24 GMT From: Chris Brown <mcvax!aivru.sheffield.ac.uk!chris@uunet.uu.net> Subject: How to make a device driver awake with Ctrl-C? (I am posting this for a friend. If you reply, use the address given at the end, and NOT the one in the header!) I am currently writing a device driver for a simple VME device on a Sun3/140 running SunOS 4.0. The device consists of a simple byte orientated register interface (similar to the "skeleton" device described in the "Writing Device Drivers" Manual as supplied by Sun).The device also has interrupt capability when it is ready to send or recieve a byte. THE PROBLEM: One of the tests for the driver is that it is told to read some bytes from the device when there are no bytes present at all ! What I discovered during this test is that the driver was seen to "hang up" forever awaiting an input. I typed CTRL-C (again and again and again !!) but the driver still remained asleep. How do I get Sun device drivers to stop when the user supplies a CTRL-C interrupt signal. From what I have read in the manual there is a "PCATCH" flag available for the "sleep" kernel routine and this ORed with the priority, supplied to sleep, should allow the driver to pick up the user interrupt. My driver seems to be asleep in the "physio" kernel call for my read, it has the form: xxRead (dev, uio) dev_t dev; struct uio *uio; { int Unit = xxUNIT(dev); /* Use the UNIT macro to pull out the unit no */ return (physio (xxStrategy, &xxBufs[Unit], dev, B_READ, xxMinphys, uio)); } When the driver has gone into its forever sleep and I do a "ps -lax", this says that the driver is awaiting input on the "xxBufs" buffer. So, it looks as if the sleep is happening in the physio routine.Does this mean that I have to write my own "physio" routine and lower the priority in the "sleep" call ? I would be very grateful for any help with this problem. Thank you. * NAME: Richard Onyett * COMPANY: INMOS LTD * ADDRESS: 1000 Aztec West * Almondsbury * Bristol * BS12 4SQ * England * * TELEPHONE: (0454) 616616 * Email : rich@inmos.uucp * * PROBLEM: Using CTRL-C with a device driver. ------------------------------ Date: Thu, 29 Sep 88 09:40:19 EDT From: jwm@stdc.jhuapl.edu (Jim Meritt) Subject: Follow up question on Bigger Icons? Sridhar Acharya asks: > Is there a tool which takes the bitmaps > reated out of the iconedit, and forms icons of dimensions 128x64 (Width X > eight)?. I could easily produce icons of width -64 pixels and any height > by concatenating the two images and increasing the height, but found it > too painful to put together two icons side by side. Is this applicable? I made a tool that could extract and insert 64x64 icons into a rasterfile. You can build up the raster file by putting icons whereever you want them (use mouse) then save the rasterfile, then use the pic2icon program to turn that into an icon. A bit of a hack, but on hand. ------------------------------ Date: Wed, 28 Sep 88 17:09:21 EDT From: "Arthur Wouk" <wouk@brl.mil> Subject: interactive spelling checker? I am looking for an interactive spelling checker which runs under SUNOS 3.2. I have long used Fix2 which compiles under BSD4.x and Gould UTX/32 but fails under SUNOS, and the unix 'spellfix' which fails which does not run under SUNOS. An interactive spelling checker uses spell to compile a list of apparent errors, presents them for correction first on a word by word basis, to which responses and corrections can be made, and later in context for correction in context. As a last step it corrects the document and updates and maintains a private dictionary. Does anyone have such a working spelling checker, or can anyone point the way to one? ------------------------------ Date: 29 Sep 88 00:10:24 GMT From: Sven-Ove Westberg <mcvax!cad.luth.se!sow@uunet.uu.net> Subject: gettimeofday() bug? The routine gettimeofday always returns a pattern in the micro seconds part. Run this small test program. #include <sys/time.h> #include <stdio.h> main() { struct timeval tp; struct timezone tz; while(1) { gettimeofday(&tp,&tz); printf("s %ld us %07ld \n",tp.tv_sec,tp.tv_usec); } } and the output will be: dd00dd where d is any digit 0-9 and 0 is always 0. Sun is the only machine where I found this kind of timegap. Can someone explain the two 0 in the middle?? Sven-Ove Westberg, CAD, University of Lulea, S-951 87 Lulea, Sweden. ARPA: sow%cad.luth.se@ucbvax.berkeley.edu (only dumb ARPA mailers) Internet: sow@cad.luth.se ------------------------------ Date: Wed, 28 Sep 88 16:16:49 BST From: mcvax!gec-epl.co.uk!James_House@uunet.uu.net Subject: Coral-66? Does anyone know of a Coral-66 compiler for Sun (besides the one from Mach-One computing), or failing that, a Unix compiler that could be ported fairly easily? Thanks for any help, James House Systems Design Division GEC Electrical Projects Ltd Boughton Road, Rugby, CV21 1BU, England. Email: JAH@gec-epl.co.uk or ...mcvax!ukc!uk.co.gec-epl!JAH Tel: (+44 788 or 0788) 542144 ------------------------------ End of SUN-Spots Digest ***********************