Sun-Spots-Request@RICE.EDU (William LeFebvre) (04/16/88)
SUN-SPOTS DIGEST Friday, 15 April 1988 Volume 6 : Issue 53 Today's Topics: Administrivia Re: TCPDUMP for SunOS 4.0 Re: Filters for fig Re: Does anyone have Calctool V2.1? Other calculators? Re: Implementations of PHIGS and GKS Re: Problems configuring cua0 on a 3/160 Dialin/out on one port newfs/mkfs inode shortage problem Likely autoinitialization bug A question about NFS hierarchies screen dumping? Adding a third Eagle? Conference room sized color displays? 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: Fri, 15 Apr 88 13:54:25 CDT From: William LeFebvre <phil@Rice.edu> Subject: Administrivia My apologies to all who have been patiently waiting for me to process requests. I have just processed all the pending requests (and some of them dated back to the middle of March, I'm afraid). Everyone has now been added or deleted as they requested, with the exception of the Bitnet folks---I'm still working on that. Once again, I'm sorry that I took so long to get around to it. The archive's copy of Volume 6 issue 50 was empty. This has been fixed. The indexes are all up to date now. I hope to have a marathon sun-spots party this weekend and digestify the entire backlog of messages. Then I can put them out at my leisure during the coming week. There are 213695 bytes waiting to be processed, the oldest one is from April 6. I was down to a week turn-around time, but then things got busy and I went a few days without assembling any digests. Thank you for your patience. William LeFebvre Department of Computer Science Rice University <phil@Rice.edu> ------------------------------ Date: Mon, 04 Apr 88 22:09:06 PDT From: Craig Leres <leres%lbl-helios@lbl-rtsg.arpa> Subject: Re: TCPDUMP for SunOS 4.0 Reference: v6n41 The nit interface changed quite a bit between 3.X and 4.0 Sun OS. In any case, tcpdump probably won't get converted to 4.0 until we start running 4.0 on our Suns... Remember, until Sun says it's ok, we can't give you the tcpdump source (under any circumstances). Please don't ask. Craig ------------------------------ Date: Tue, 5 Apr 88 14:05:39 BST From: everson%COMPSCI.BRISTOL.AC.UK@cunyvm.cuny.edu Subject: Re: Filters for fig Reference: v6n37 > I have just retrieved Fig from the archive. My difficulty is the filter > supplied (f2p) is for pic. Does anyone have a suitable filter for TeX or > LaTeX?... What is needed is a fig to postscript filter as the postscript can then be included into the TeX/LaTeX file using the \special command. We have such a beast here; however, it is encoded by a program atob which we do not have. If you have a copy of atob or alternatively if you have got a copy of a filter from somewhere else please send me a copy of the source! This multiplicity of formats is *very* frustrating! Phill Everson University of Bristol, UK [[ The new version of Fig (1.4) comes with a fig->postscript converter (called (f2ps). Now if only I could find the time to bundle it up and put it in the archives.... By the way, all you fig fans will be "pleased" to hear (I'm being sarcastic) that the fig file format changed with version 1.4 and is incompatible with the old format, although I understand that 1.4 will read old format files. --wnl ]] ------------------------------ Date: Tue, 05 Apr 88 16:03:07 -0400 From: mesard@bbn.com Subject: Re: Does anyone have Calctool V2.1? Other calculators? > Does anyone out there have the complete and correct sources?... I have Calctool 2.1. Let me know if you haven't gotten it yet, and I'll mail it (or the parts you're missing) to you. > In a more general light, can anyone recommend a good *scientific* > calculator program (not necessarily just for SUN's)?... I also have a command line (i.e. non visual) calc program that I wrote which I use all the time and love dearly. It does things like % calc "tan 45 / (5 % 3)" 0.5 Support exists for trig and other functions, math and bitwise operations, hex, octal, binary and ascii representations, precision as high as your C math library and hardware will go. (Which is ~16 decimal digits as I [mis]understand the IEEE double precision standard.) I just uploaded a shar with the source files and man page for this program to titan.rice.edu. I'm cc'ing this message to wnl to let him know and to tell him to do what he will with it. [[ It has been moved to "public/wsm_calc.shar". I decided that, since it was not Sun-related, it didn't really belong in sun-source. But don't worry, archive server users can still retrieve it with "send public wsm_calc.shar". It is 21234 bytes in length. --wnl ]] MESARD@BBN.COM BBN Labs, Cambridge, MA ------------------------------ Date: Wed, 6 Apr 88 11:11:59 EDT From: bob@allosaur.cis.ohio-state.edu (Bob Sutterfield) Subject: Re: Implementations of PHIGS and GKS Reference: v6n42 Look in the directory doc/PEX in the release 2 distribution of the X Window System, Version 11. Therein is the PEX Protocol Specification, the result of work by the X3D committee on 3-D extensions to the X11 protocol. Their description of a PEX resource called a "PHIGS workstation" begins on page 101. It may take you in a direction you want to go. Bob Sutterfield, Department of Computer and Information Science The Ohio State University; 2036 Neil Ave. Columbus OH USA 43210-1277 bob@cis.ohio-state.edu or ...!cbosgd!osu-cis!bob ------------------------------ Date: Mon, 4 Apr 88 15:35:24 MST From: hyder@blueox@hub.ucsb.edu Subject: Re: Problems configuring cua0 on a 3/160 The detailed information provided about modems and cua0 in this reply is interesting but not quite correct for SunOS users. It is quite possible, thanks to the device driver supplied by Sun, to use the same modem for dial in and dial out. Honest, it works just fine. (The discussion of modem lines that was given relates mostly to more generic 4.nBSD implementations.) The bi-directional mode works IFF(sic) you configure everything EXACTLY as described in the adding a modem section of the reference manuals. (It may also help to find notes from the Sun System Administrator's course and look at them but the reference manual has better detail.) YES, a minor kernel reconfig is required. THE CATCHES: 1. The modem !MUST! be configured properly. Almost all of the problems I've seen have been the result of incorrect modem switch settings. If you have a real Hayes modem the switch settings are in the Sun manuals. 2. The instructions are for the ttya and ttyb ports and some of the add-on RS-232 cards may not be as easy to configure for use. (believe the initial problem report is related to one of these) Give it a try. A modem set up this way is particularly useful for restricted line environments and on outbound modems that are rarely used, like only for UUCP at 3am. Paul Hyder Blue Ox Software, Inc. ------------------------------ Date: Mon, 4 Apr 88 22:22:16 PDT From: ho@tis-w.arpa (Hilarie K. Orman) Subject: Dialin/out on one port Reference: v6n43 In v6n43 Ron Hitchens gives an analysis of the /dev/cua0 mechanism ending with the depressing conclusion: "Unfortunately there is no simple method to use one modem for both dialin and dialout like you're trying to do. The setup as it exists doesn't allow it ..." Having followed the instructions in the Sun system administrators' guide for adding a modem to the system, I can testify that one modem can most definitely be used for both dial-in and dial-out. The trick is the "flags" parameter in the kernel configuration for the device. This activates a mutual exclusion mechanism between devices with matching major device numbers and minor device numbers differing by 128 (just one bit). Init waits for activity on one device, dial-out uses the other. Now, there must be a little bit of magic going on here, because somehow init manages to get away with not doing an open on its device, merely getting notified when carrier is detected, then doing the open. Nonetheless, the bottom line is, if you have followed the instructions and are still having problems, then the fault is in the modem. It must be able to use the signal lines to notify the CPU about carrier detect, and it must be able to drop carrier when the CPU drops DTR. Usually a little in-line test device will tell the whole story. ------------------------------ Date: Mon, 04 Apr 88 19:06:39 PDT From: Craig Leres <leres%lbl-helios@lbl-rtsg.arpa> Subject: newfs/mkfs inode shortage problem Over the months, several people have complained that they don't end up with enough inodes when they make a new filesystem under Sun OS. I ran into this difficultly when I put news/nntp up on my file server. The problem seems to be that newer drives tend to have many more sectors per cyclnder. For example, the standard 70 Mbyte Micropolis 1325 that comes with a "shoebox" only has 136 sectors/cyclnder while my 600 Mbyte Fujitsu 2344's have 1890 sectors/cyclnder. With these newer big drives, the default number of cylinders per cylinder group (16) is too large. Since you're limited to 2048 inodes per cylinder group, if you don't have enough cylinder groups, you're out of luck. The workaround we came up with was to specify 8 cylinders/group to newfs (using -c 8). We really wanted 4 cylinders/group to get the desired inode density of 2 Kbytes/inode, but there's some other filesystem limitation that force cylinders/group to be a multiple of 8. Still, this yields twice as many inodes. Hope this helps. Craig ------------------------------ Date: 5 Apr 88 21:36:06 GMT From: ncar!boulder!stcvax!stc-auts!kak@rutgers.edu (Kris Kugel) Subject: Likely autoinitialization bug [[ This was "cross-posted" to several Usenet groups. There may already be discussions about this going on in comp.lang.c. --wnl ]] /* showbug * compile: cc showbug.c -o showbug -lm */ main() { float mean = 0.035000; int samples = 200; { double sigma = sqrt( (mean*(1 - mean))/ samples); printf("sigma = sqrt( (%f)/%d) ", (mean*(1 - mean)), samples ); printf("= sqrt( %f ) ", (mean*(1 - mean))/ samples ); printf("= %f ", sqrt((mean*(1 - mean))/ samples )); printf("= %f (!)\n", sigma ); } } the Autoinitialization of sigma fails on every system I've tried it on so far, (sun, bsd2.10, ATT3b1/3.51, ultrix) (even on bsd2.10, sigma is different, although both are wrong) Does anyone know what is going on here? Is this a bug? Kris A. Kugel Storage Tek: ...{ uunet!nbires, ncar, ihnp4 }!stcvax!stc-auts!kak ------------------------------ Date: 6 Apr 88 02:00:33 GMT From: helios!lalonde@uunet.uu.net (Terry Lalonde) Subject: A question about NFS hierarchies Consider the following: A diskless machine (A) mounts a filesystem on machine (B) at /usr and another filesystem on machine (C) at /usr/machc. i.e. mount machineB:/usr/sun3 /usr mount machineC:/usr/machc /usr/machc Question: Is this an inefficient to do things? Does this make I/O to a file on /usr/machc dependent on both (B) and (C) and their loads? Thanks, Terry ------------------------------ Date: Tue, 05 Apr 88 14:18:23 EDT From: jfjr@mitre-bedford.arpa Subject: screen dumping? I am a new user. If this deserves an RTFM I'll take it but try to do it nicely I am sensitive. We have a Sun 3 and we are doing some "interactive" plotting. At the end of the plot program we just go system(screendump | rasfilter8to1 |lpr -v). Most of the time this works but if someone else is logged in from another terminal somethings get all confused. Sometimes it tells me that "valloc" failed. Sometimes there are problems with "pix_rect", sometimes with /dev/fb. Is there anything I can do?? or should I live with it?? Jerry Freedman,Jr jfjr@mitre-bedford.arpa ------------------------------ Date: 5 Apr 88 14:30:42 GMT From: roy%phri@uunet.uu.net (Roy Smith) Subject: Adding a third Eagle? Has anybody tried putting more than 2 drives per controller on a Sun-3/180? We've got two 180s, each with two eagles and a single XY-450 controller. I've also got a fifth eagle sitting in a box (long story) waiting to be made useful. The XY board can supposedly handle up to 4 drives but the kernel driver won't support than many. Is there some good reason why not? Lacking any way to wedge a third drive onto an existing controller, what suggestions do people have for a new controller? For now, all I want to put on it is a plain old eagle, but at some time in the future we'll probably want to support bigger, faster drives. Is it really worth it to get one of those fancy new buffer-a-cylinder-at-a-time controllers as opposed to something like a plain old XY-451? Roy Smith, {allegra,cmcl2,philabs}!phri!roy System Administrator, Public Health Research Institute 455 First Avenue, New York, NY 10016 ------------------------------ Date: Tue, 5 Apr 88 17:23:03 EDT From: mstan!s2!dpk@uunet.uu.net (Douglas P. Kingston) Subject: Conference room sized color displays? We are looking for a large high resolution color display suitable for use in a conference room. We would like it to be a projection system of some sort. We have heard of the GE Lightvalve (?) but I believe that is a bit large (and expensive for our needs (a 50 seat conference room). The standard Sun display options are too small. Any pointers would be welcome. It would be a plus if it could also display the lower resolution PC graphics, but this is not a requirement. -Doug- dpk@morgan.com (or dpk@brl.arpa) ------------------------------ End of SUN-Spots Digest ***********************