PAGESWAPPER The VAX Migration Working Group by James Downward KMS Fusion, Incorporated Post Office Box 1567 Ann Arbor, Michigan 48106 (313)-769-8500 At the Spring DECUS meeting, it was felt that the SIG membership should be made aware of the status and functions of the various working groups and that the PAGESWAPPER articles/letters were a good way to start the process. I am chairperson for the VAX Migration Working Group. I must admit we do not really have a good name, certainly not one which conveys the wide range of the group's interests. The areas we were initially interested in included: 1. RSX to VAX migration issues. 2. VAX to RSX migration (some folks need this too) issues. 3. VAX to RSX code transportability issues. 4. RSX/VAX compiler conformity. 5. Using the VAX as a host development system for RSX or POS. 6. The VAX Application Migration Executive. However, it is now apparent that the migration issues between VAX systems and other operating systems (RSTS, RT, DEC 10/20, IBM??) also need to be considered. At this time, I am soliciting participation in this working group from users currently worrying about cross system migration issues, or doing host development work. Some of you may have developed migration aids. Others may have documented all the problems (and successes) encountered along the way. Others may be using VAXs as host development systems for a variety of other target systems. Others may have tried and given up in disgust when it 1 PAGESWAPPER - July 1983 - Volume 5 Number 1 The VAX Migration Working Group almost (but not quite) worked the way it should have. Well, here's your chance. Join the migration working group and share what you have learned. Send me a letter stating your areas of interest, current status of efforts, and perchance any neat things you have done which others might be interested in. In return, I will send out periodic mailings to the working group, informing you of who the other members are and what their interests are. The group will also try to provide a forum for coordinating SIR activity relating to migration issues. A Solution to the Priority Problem by David P. Schumacher Tri-University Meson Facility (TRIUMF) 4004 Wesbrook Mall Vancouver, B. C., Canada Our VAX 11/780 system currently utilizes 80-85% of its CPU time. To distribute the CPU resource more equitably between interactive and batch users, we allow interactive processes 10 minutes at priority 4 or 60 minutes at priority 3. All batch queues, except for FAST at 4, have a priority of 2. A detached job called SHUFFLEQUE randomly selects a batch job and raises its priority to 4 for 30 seconds. This technique speeds throughput of batch jobs without unduly degrading interactive response times. 2 PAGESWAPPER - July 1983 - Volume 5 Number 1 In this issue... In this issue... The VAX Migration Working Group . . . . . . . . . . 1 A Solution to the Priority Problem . . . . . . . . . 2 In this issue... . . . . . . . . . . . . . . . . . . 3 Editor's Workfile . . . . . . . . . . . . . . . . . 4 PSI Users Unite !!! . . . . . . . . . . . . . . . . 5 RTP LUG . . . . . . . . . . . . . . . . . . . . . . 7 LUG Agenda . . . . . . . . . . . . . . . . . . . . . 8 INPUT/OUTPUT . . . . . . . . . . . . . . . . . . . . 9 Standalone BACKUP for 750 Simplified . . . . . . . 18 How We Rounded Cape Horn . . . . . . . . . . . . . 18 Spring 1983 VAX Advanced Q & A . . . . . . . . . . 20 VAX CI Software Systems . . . . . . . . . . . . . 34 UNIX Hints and Kinks . . . . . . . . . . . . . . . 39 Writing C Code for VAX/VMS and UNIX Systems . . . 42 Q-Bus/UNIBUS Hardware Hints and Kinks . . . . . . 44 Security Mechanisms for VAX/VMS . . . . . . . . . 47 VAX Security Panel . . . . . . . . . . . . . . . . 49 Fall 1983 System Improvement Request List . . . . 53 Fall 1983 System Improvement Request Ballot . . . 69 INPUT/OUTPUT Submission Form . . . . . . . . . . . 71 System Improvement Request Submission Form . . . . 73 General material for publication in the Pageswapper should be sent (US mail only -- no "express" services please) to: Larry Kilgallen, PAGESWAPPER Editor Box 81, MIT Station Cambridge, MA 02139-0901 Preference is given to material submitted as machine-readable text or Runoff source. Material for "The DBMS Monitor" section of the Pageswapper (pertaining to VAX-11 DBMS) should be sent to: Julie Llewellyn United Technologies Microelectronics Center 1365 Garden of the Gods Road Colorado Springs, CO 80907 3 PAGESWAPPER - July 1983 - Volume 5 Number 1 Change of address, reports of non-receipt, and other circulation correspondence should be sent to: DECUS U.S. Chapter, MRO2-1/C11 Attention: Publications Department One Iron Way Marlborough, MA 01752 USA Only if discrepancies of the mailing system are reported can they be analyzed and corrected. Editor's Workfile by Larry Kilgallen, Pageswapper Editor Starting with "VAX CI Software Systems" you will find results of the "DECUS Scribe Service", marking either a great step forward in member service or else the beginning of the DECUS road-to-ruin. The work was done by Saint Louis journalism students and computer science students hired by DECUS to report on sessions. Although they did not all have years of VAX experience, they also did not have conflicting sessions to attend, and were able to concentrate on producing reports on the sessions they were assigned. The DECUS member coordinating the effort was Ralph Stamerjohn, former editor of the Multi-tasker and well-known leader in the RSX world. Ralph certainly deserves all our thanks for putting together such a great production. What disturbs me though, is that in printing these articles I make a major admission - that we did not get any articles on these sessions from SIG members. Perhaps those attending were so entranced with the idea of buying cassette tapes that they did not feel any other medium was required for making the record of those sessions (I doubt that many consciously thought about the Scribe service). But for whatever reason, when we give up on hoping for volunteer efforts, I think something has gone wrong. Larry Kilgallen, Pageswapper Editor 4 PAGESWAPPER - July 1983 - Volume 5 Number 1 Editor's Workfile ************************************************************ ************************************************************ ************************************************************ **** **** **** Topic for September **** **** **** **** Chargeback and Billing **** **** **** **** Send YOUR Article **** **** **** ************************************************************ ************************************************************ ************************************************************ PSI Users Unite !!! June 7, 1983 Larry Kilgallen, editor The Pageswapper Box 81, MIT Station Cambridge, MA 02139-0901 Dear Larry, I would like to take this opportunity to make the readers of the Pageswapper aware of a small group within the DECUS VAX community -- the PSI user's group. PSI, or Packetnet System Interface, is a Digital software product that allows a VAX/VMS system to connect to various networks, PAD boxes, and other hardware using the CCITT communications recommendation X.25. PSI is also available for TOPS-20 systems. PSI requires additional synchronous communications hardware, such as DUP-11s or KMC-11s. PSI is relatively new in the United States, although it has a strong following in Europe, and was developed in England by DEC England. PSI can allow access to various public packet switched networks in America such as Telenet, Tymnet, CompuServe Network, Uninet, and RCA Globcomm as well as most PTT networks in Europe. Although the current version of PSI (V1.2) does not provide totally transparent bi-directional access, the new version (V2.0, which is shipping now), has support by DEC for both incoming and outbound terminal connections. PSI runs much as DECnet does (in fact, DECnet can use PSI as its physical/link 5 PAGESWAPPER - July 1983 - Volume 5 Number 1 PSI Users Unite !!! level protocol), as an ACP hanging around in space. Digital's PSI (which competes with several other vendor's products serving the same functionality) supports access by remote terminals as per recommendations X.3, X.28, and X.29. PSI uses RS-232-C (V.24) at physical level and LAPB at frame/packet level. The PSI User's Group is a very loosely affiliated set of PSI users and future users scattered over America, Canada, and England. As of this writing, fifteen sites have become members. The term "member" is perhaps misleading, since no organized membership rolls are kept. Rather, sites may become members by sending some media -- either a floppy disk or magnetic tape -- along with a self-addressed, stamped mailer, to me. Each member is asked to contribute something -- software, advice, bugs, hints, kinks, even inside information -- which is then collated and redistributed to all. We have just finished the first mailing, and I hope to create a second distribution in July or August. If response by the membership is enthusiastic, distributions will probably be made on a quarterly basis. If response is overwhelming and the magnitude of copying floppies becomes too large, then future distributions will probably be made through the VAX SIG tape at each DECUS. Indeed, I have submitted the first PSI distribution floppy to the VAX SIG tape from St. Louis (Spring 1983) in hopes of reaching the widest audience possible. At this point, we are trying to determine whether distribution through the networks that most of us are connected to is feasible or not; it certainly seems the most appropriate method of distributing software for a group of this nature. Readers of the Pageswapper are invited to join by contacting me at: Joel M Snyder CompuServe Research and Development Centre 5055 East Broadway, Suite A-110 Tucson, Arizona 85711 (602) 790-5061 Sincerely, Joel Snyder 6 PAGESWAPPER - July 1983 - Volume 5 Number 1 RTP LUG RTP LUG We have 4 meetings a year. We have excellent support from our local DEC office. They provide at least 1 speaker for each of our meetings. I can request someone or some topic and they will obtain the technical speaker, usually flown in from Massachusetts (to North Carolina) for the meeting. Not very many DEC employees usually show up at our meetings - held 3-6 pm. Our meeting attendance ranges from 50 to 130. I do not have a concrete reason why. Our LUG contains 400 members (150 active ones). Our local DEC Software Services has supplied our LUG with an account on their local VAX 11/780. We have had that account for almost 3 years. In reality it has seen almost no use for mail or access to our automated database of members. We have about 10 different LUG officers, but most of the work falls on my shoulders. I have received excellent support from my DEC Regional Coordinator to date. I would like to see a list of speakers/topics that go over well at other LUGs published regularly. Our next meeting is June 13. Joe Ryan Regional LUG Coordinator LUG Chairman (RTP LUG) 7 PAGESWAPPER - July 1983 - Volume 5 Number 1 LUG Agenda LUG Agenda The following LUG meeting information is published with the understanding that Pageswapper readers will receive it too late to attend the meetings. The purpose, rather, (in accordance with Joe Ryan's suggestion above) is to share meeting topic ideas among LUGS. o VAX LUG (Albuquerque, NM) - Wednesday, June 22, 1983 The meeting this time will be held at the Signetics plant. The program is to be mainly a report on the recent DECUS meeting in St. Louis. Any further information on the 3rd Annual NM DECUS meeting will also be discussed. There will be a tour of the Signetics plant following the meeting. Submitted by: Donald Robbins, LUG Secretary Sandia National Laboratories Division 9263 Post Office Box 5800 Albuquerque, NM 87185 o BAYVAX LUG (San Francisco, CA) - Thursday June 30, 1983 The meeting will be held 9 am to noon at the Auditorim at SLAC. The speaker will be Ken O'Mohundro of ABLE computers, talking about Ethernet connections and port selectors. 8 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT INPUT/OUTPUT A SIG Information Interchange A form for INPUT/OUTPUT submissions is available at the back of the issue. INPUT/OUTPUT 156 Caption: PDP-11 Assembly language interpreter wanted Message: Our Computer Science department teaches PDP-11 assembler as part of a course that compares computer architectures and assembler language on a number of different computers. They would like to have an interpreter for PDP-11 assembler which runs on a VAX, under either VMS or UNIX. The interpreter should be able to provide a trace of the last 10 or so instructions executed and a dump of relevant memory at program termination and a diagnostic message if the termination is due to an error. Contact: Peter W. Day Computing Center, Uppergate House Emory University Atlanta, Georgia 30322 USENET EMORY!PWDAY (404) 329-7651 Date: May 13, 1983 INPUT/OUTPUT 157 Caption: Reading HP-3000 STORE Tapes onto VAX Message: We are phasing out an HP-3000 in favor of a VAX and consequently have many HP-3000 STORE formatted tapes. We are looking for a program that will enable us to read ASCII files from these tapes onto the VAX. Contact: Salvatore Mazzotta SCI 1033 Massachusetts Avenue Cambridge, MA 02138 (617) 868-1200 9 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT Date: May 18, 1983 INPUT/OUTPUT 158 Caption: Deming Software -- REPLY TO I/O # 132 Message: I have a number of small BASIC programs for SQC. One calculates the upper and lower control limits for R/XBar charts when given sufficient data. User options include the lot size to be used. It does not plot data or control limits, but I could add it fairly easily. The constants used in the programs are taken from "Fundamentals of Statistical Quality Control", by Jerome D. Braverman, published in 1981 by Reston, a Prentice-Hall Company. Contact: Milt Boyd DEC 110 Spit Brook Road Nashua, NH 03061 (603) 881-1351 Date: May 27, 1983 INPUT/OUTPUT 159 Caption: SIG-provided User System Services Message: The VAX Systems SIG Internals Working Group would like to know what VMS user-written system services members might like to see provided on the SIG tape. Send information to: Contact: David J. Slater Mantech International 2320 Mill Road Alexandria, VA 22314 Date: May 25, 1983 10 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT INPUT/OUTPUT 160 Caption: PLZ Cross Compiler Message: We would like to compile PLZ source code, or any language closely resembling PLZ, on our VAX or PDP-11 to produce Z80, Intel 8080, or Intel 8085 object code. Does anybody know of such a cross compiler? Contact: Donald Forest Decision Data Computer Corporation 100 Witmer Road Horsham, PA 19044 (215) 674-3300 Date: June 6, 1983 INPUT/OUTPUT 161 Caption: Tape Reading Software Needed Message: We are converting our Engineering workload from a Burroughs mainframe to a VAX 11/780. We have over 8,000 Burroughs library maintenance tapes that we must read on the VAX. Please contact us if you know of any VAX software that will read Burroughs library maintenance tapes. Contact: Phil Henderson United Technologies Corporation 1050 E. Arques Avenue Sunnyvale, CA 94086 (408) 730-6315 Date: June 3, 1983 INPUT/OUTPUT 162 Caption: Virtual Memory from Fortran -- REPLY TO I/O # 136 Message: After allocating memory with LIB$GET_VM, passing its address to a subroutine with %VAL is the method of choice for accessing the memory. The trick is to do this just once, not repeatedly. This is achieved by restructuring the program to be a subroutine of a new routine (or mainline) which just allocates the memory and passes its base address and size to the inner routine. The inner routine (i.e. the rest of the program) accepts the virtual memory as an adjustable array, and can thereafter reference it freely just as 11 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT any other array. The overhead of array indexing can be eliminated for specific frequently-used elements by passing their addresses separately as additional arguments which are picked up by the inner routine as scalar variables. Contact: Richard I. Katz Hughes Aircraft Company Building 12 M/S V197 Post Office Box 92426 Los Angeles, CA 90009 (213) 305-2355 Date: June 6, 1983 INPUT/OUTPUT 163 Caption: Zeroing real memory Message: We may be processing Department of Defense classified information on our 730. The 730 is in an unclassified area, so we are developing procedures to bring VMS up in a classified mode and to declassify the machine after running classified. One problem is how to clear main memory. We would like a way that doesn't require powering down the VAX. Does someone know how to zero real memory? We don't care if it crashes VMS, as a reboot is necessary to bring up different page, swap and system dump files. Also, we are interested in exchanging information with others doing similar things. Contact: Jim Kitchen Scientific Simulation, Incorporated Post Office Box 9331 Albuquerque, NM 87119 (505) 247-3776 Date: June 6, 1983 INPUT/OUTPUT 164 Caption: Using Auto-Call with DF03-AC's -- REPLY TO I/O # 125 Message: To call my VAX from my home is a long distance call, however, it is not a long distance call for my VAX to call me. I have 4 DF03-AC's and wanted to use the auto-call feature to save money on my phone bill, and here is how I did it. 12 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT First I created a file on the VAX with just one line in it, B (CTRL/B) then a 9 (to get an outside line) followed by my home phone number. At home I call one of the modems and logon. Then I set one of the other modems "/NOAUTOBAUD/SPEED=1200". I then PIP the second modem the file with my home phone number in it. This next part is tricky, when the VAX returns the "$" prompt I quickly hit the "DATA/TALK" button on my modem twice as to disconnect me quickly (my modem ports are all set "/MODEM/HANGUP") and in less than 30 seconds my VAX will disconnect me from the first modem and call me back on the second. Contact: Tom Heim Jr. Digital Equipment Corporation 2500 W. Union Hills Drive Phoenix, Arizona 85027 (602) 869-5780 Date: June 6, 1983 INPUT/OUTPUT 165 Caption: Print symbiont disappears for variable time lengths Message: I am having problems with the print symbiont for one line printer (LP11) on a long line driver (M5915) going away and then coming back. This only occurs on the one printer and it is the only printer on the system on a long line. The boards have been replaced multiple times. Stopping and restarting the queue does not help. It is not possible to delete or stop the print job that shows as current. Rebooting will bring the symbiont back but only temporarily. So far we have been running for two weeks without it going away. This is all I have been able to gather thus far. If anyone has had this problem or has an idea about it, please contact me. Contact: Cindy Letton Apex Oil Company 11221 Olive St. Rd. Creve Coeur, MO 63141 Date: June 7, 1983 13 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT INPUT/OUTPUT 166 Caption: Installation Staffing Message: We are new VAX 11/780 users. We are aware of a number of sources of information on hardware capacity planning, but need to plan our staffing requirements as we expand our system to support more users, more applications and DECnet to several other sites. If anyone could provide statistics on the "average" staffing requirements of various size VAX installations or guidelines to aid in such planning we would appreciate the help. Contact: Don Anderson Automated Sciences Group 700 Roeder Road Silver Spring, MD 20910 (301) 587-8750 Date: June 7, 1983 INPUT/OUTPUT 167 Caption: Backissues Message: Is there any way to obtain back issues of Pageswapper, specifically Volume 3, Number 1 through Volume 4, Number 2. If so, how? Contact: F. Kurt Schulte H. H. Cutler & Co. 120 Ionia SW Grand Rapids, MI 49503 (616) 459-9101 x329 Date: June 8, 1983 Editor's Response We have no such mechanism; your best bet would be to find someone in your LUG who has them, or hope that a reader of this column will make contact. 14 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT INPUT/OUTPUT 168 Caption: System 1032 (tm) Offered Message: System 1032 Data Base Management is now being offered for sale. Based on System 1022 (r) used on 36 bit DEC machines, System 1032 is aimed at providing a simple English-like command format for query, data entry and reporting, rapid access using multiple keyed fields, and a unified Host Language Interface. A 60-day on-site evaluation period is offered to VAX sites. Contact: Betsy Ziegler Software House 1105 Massachusetts Avenue Cambridge, MA 02138 (617) 661-9440 Date: June 10, 1983 INPUT/OUTPUT 169 Caption: Data Compaction Routines Message: We are looking for an efficient set of data compaction routines to compact and decompact large amounts of engineering data to be stored online for use with our 11/780 running VMS. Contact: Don Anderson Automated Sciences Group 700 Roeder Road Silver spring, MD 20910 (301) 587-8750 Date: June 14, 1983 INPUT/OUTPUT 170 Caption: Screen Editor for non-DEC terminals Message: We need a screen-oriented text editor that runs under VMS on non-DEC terminals. Does anything exist for VMS that is analogous to UC Berkeley's UNIX terminal programs that are used to tailor their visual screen editor, VI, to the type of terminal the the user is currently logged in on? 15 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT Contact: David Taylor Richter Seismological Laboratory University of California Santa Cruz, CA 95064 (408) 429-4426 Date: June 14, 1983 INPUT/OUTPUT 171 Caption: An Accounting Fix for Pages Printed Message: We need to count pages off of a letter quality printer queued through a terminal port. At present the counts are not correct and there is no obvious difference or system to compute the correct amounts. Does anyone have a command procedure or fix for the count? Contact: Barbara Kurshan Dana Hall Hollins College Date: June 14, 1983 INPUT/OUTPUT 172 Caption: Foreign terminal handler for Hewlett Packard Terminals Message: I am interested in implementing the foreign terminal package, SCRFT, for Hewlett Packard 2620 and 2640 terminals on a VAX-11/750. Anyone who has solved the problem for utilities like MONITOR, EDT and SHOW PROCESS/CONTINUOUS please contact me. Contact: Steven Joerger Armament Systems 712-F North Valley Street Anaheim, CA 92801 (714) 635-1524 16 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT INPUT/OUTPUT 173 Caption: Fortran file I/O Optimization by Tom Kent (3/82) Message: The May 1983 article on Performance Tuning referenced an article by Tom Kent that appeared in the March 1982 Pageswapper. This article discussed Fortran file I/O optimization techniques. Could someone forward me a copy of this article? Contact: Iris Wolfe 1375 Sunflower Avenue Costa Mesa, CA 92626 (717) 957-2404 Date: June 16, 1983 INPUT/OUTPUT 174 Caption: VAX-to-VAX File Transfer Message: We have a 750 and campus based 780 connected by a serial port. Both systems run VMS but DECnet is not available. Is there any software available to allow a user on one system to do file transfer with his/her account on the other system? Contact: Jim Leone Computer Science, Canisius College 2001 Main Street Buffalo, NY 14208 (716) 883-7000 Extension 415 Date: June 16, 1983 17 PAGESWAPPER - July 1983 - Volume 5 Number 1 Standalone BACKUP for 750 Simplified Standalone BACKUP for 750 Simplified by James Downward KMS Fusion, Incorporated Post Office Box 1567 Ann Arbor, MI 48106 (313)-769-8500 This is just a brief note on standalone BACKUP. It seems that both past articles have made things much harder than they need be. There is no necessity to boot the 11/750 from the TU58 tape (there is with the 11/730). If one has an 11/750, place the files generated using STABAKIT.COM on [SYS'n'.SYSEXE] (I use B for BACKUP). Then when the system is halted, type BOOT/'n'0000000 DDcu: where DDcu: is the disk you normally boot from. In our case, we normally boot from an RM03, so we type B/B0000000 DBA0: and immediately the stand alone BACKUP system starts up. How We Rounded Cape Horn A Book Review By Larry Kilgallen, Pageswapper Editor The book in question is actually titled "engineering a compiler, VAX-11 Code Generation and Optimization" and published by Digital Press, DEC's version of diversification. It reminds me though, of oh so many sailing books (the other half of my own book collection) which expound in unending detail on how a particular crew of cruising sailors made good their voyage. After several of those books one comes to realize that labels washing off the canned vegetables is a distinct problem, but only of consuming interest to those about to start their own voyage. The sailors, however, are trying to make a buck to support their expensive habit. My own gut feel for economics says that DEC is not going to make that much selling compiler books, compared, for instance, to selling compilers. The authors acknowledge in Chapter 12 that their approach to their compiler(s) is that of engineers, rather than computer scientists. Their approach to writing books has the same tone, making it bearable to me. But their approach is also not that of journalists, which can make some books downright enjoyable. 18 PAGESWAPPER - July 1983 - Volume 5 Number 1 How We Rounded Cape Horn Since this book is not written or published by an impartial observer we do not read about any "Hardy Boys" as in "Soul of a New Machine". But even though the book "The Fifth Generation" is written by semi-insiders, it manages to give details about those who dissent from the party line. When "engineering a compiler" tells how the PL/I back-end was chosen for C, PEARL, and an un-named fourth compiler still under development, it would have been nice to read about how the PL/I back-end was NOT chosen for Pascal! Since specific technical information is to be communicated, we cannot expect a masterpiece of narrative. But the presentation of technical information lacks the degree of organization, for instance, of Aho and Ullman's "Principles of Compiler Design" which IS more of a computer-science book (i.e. the kind your reviewer hates to read). Perhaps the difference is that "engineering a compiler" is too much of an internals manual. The authors were unwilling to leave out any area on which their work had touched. An organization based on various "lessons learned" would have been more suitable, since the authors intend to be sharing their experiences rather than providing a text on compiler construction. My own desire before opening the book was to learn "why is this compiler different from all other compilers". That was not provided with much more clarity than at the original DECUS symposium discussions. The authors do include a comparison between the two syntactic analysis techniques they used for PL/I and C, but the service rendered would have been much better if they had included a comparison with methods used by developers of other recent DEC compilers. 19 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A Spring 1983 VAX Advanced Q & A by Philip Stephensen-Payne, TSP Marketing Systime Computers Limited Millshaw Park, Leeds LS11 OLT, England The following is a transcript from the tape of the Spring 1983 VAX Advanced Q&A Session in Saint Louis. I have generally tried to keep the spirit of the session and the questions intact, whilst at the same time tidying up some of the rambling prose, and deleting a couple of questions that didn't seem to cover any new ground. Q. If I'm using spooled terminals how do I stop the Print Symbiont from skipping six lines when it thinks it's come to the end of the page, because I'm doing address labels, and don't want those spaces. It was broken when we went to V3.2 - you used to set the terminal as a Scope terminal and that stopped it, but it doesn't work anymore. You can't do a PRINT/NOFEED because you're going to the terminal directly (auto-spooling) and there is no NOFEED parameter to set on the queue. A. Noted. The Queue Manager and Job Control System is being redesigned for V4 and this is being considered. A. (questioner reporting earlier comment from DEC). Apparently you set the terminal to SCOPE and set the WIDTH to 255 and it will work. Q. Given that DEC are now producing intelligent controllers doing optimization and suchlike so that you are not guaranteed to get I/O's out the end in the same order that you put them in, how does BACKUP avoid this situation? A. Doesn't bother backup at all. The devices will complete I/O's out of order, but all you need to do is to ensure that the particular I/O you are depending on is complete before you do anything else. Tape devices do I/O in the order in which they are posted - it makes no sense to write tape records out of order! The way to ensure that a particular sequence of disk I/O's is completed in a particular order is to wait for each one in sequence before issuing the next one. Q. I've had this problem since VMS V2, when a product was released called the DEC Standard Editor. At previous DECi we've had lots of sessions on EDT and we keep being told that it's the be-all and end-all text editor. But it's not - it's three editors: there's a Line-Mode editor, and a Keypad Change Mode Editor and a Nokeypad Change Mode Editor and it's hard to get them to talk to each other. There's no pattern matching which even poor old spastic, compatibility-mode SOS 20 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A could do. You can't go into change mode from a macro or a command procedure because it has to have that screen to paint, so that it can do all those neat Little QIO's to sew up Unibuses and stuff. I noticed that there weren't any EDT sessions this time and was wondering if there was a reason for that - if you've decided to drop the product - and it would really be nice if the text editor you give us and tell us is the most wonderful thing since sliced bread could in fact do a lot of things the only unsupported product distributed with VMS - being TECO (cheers from audience) - can do. I have to go find my TECO freaks and get them to do things for me and watch the Compatibility Mode time go wild on my machine. It's not just an Editor it's a Text Processor and we've got to do things with it here. A. We must have a zillion different editors internally and externally. The reason that you don't see EDT sessions this time is that we have done EDT sessions on just about every darned thing there is in EDT Version 2. We are looking at new editing things to be done in VMS as VMS gets bigger. We're well aware of the need for pattern matching and for usable features like that and that's high on our wish-list. Q. There's a little problem in the LIB$GET_VM and LIB$FREE_VM routines. The manual says, and states very explicitly, that they can be called from AST's, but if you do it it blows up. When is it going to be fixed? A. V4. For the next release the VM routines do have their own sort of AST-Level counter which allows them to operate truly at AST Level, so a fix for this is coming in V4. (Questioner responds with shout of "Write the F...ing Manual".) Q. VMS has some neat things in it like directories and file protections, and when I look around through the microfiche, you guys have some neat routines like LIB$SET_FILE_PROT and LIB$CREATE_DIRECTORY. How come they are not in the Runtime Library so that I can do it, because I like directories and I like file protections? A. Primarily a question of bureaucracy. Obviously we wrote those routines for ourselves and there's a certain amount of work in terms of documentation and getting some of the interfaces cleaned up - making these routines follow the Runtime Library standards - which explains why they have not been officially included in the Runtime Library yet. Request noted. Q. How do you change the cluster size on a system that has one disk and one tape? 21 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A A. You cannot do it at all. Normally, to change the cluster size on a disk, you back it up, you initialize it with the required cluster size and any other parameters that you want, and then you restore the backup onto the disk with a /NOINITIALIZE switch. The CATCH-22 here is that you have no way of initializing that disk in a one disk system. (Cry from the audience of "How about Standalone INITIALIZE?") A. (DEC Software Services). We have a workaround for that. What you do is take another system somewhere, INIT a disk with the volume name of your system disk and the required cluster size, and back it up to tape with nothing on it. Backup your real system disk onto tape on your system with Standalone Backup. Restore the empty disk onto your system disk with /INIT specified, then restore your system disk with /NOINIT specified, and (bingo) your cluster size has changed. Q. My question has to do with the DCL Parse area. If you do a BACKUP and you have a long list of files that you want to backup it runs out of space too early for my taste. Are there any forthcoming changes to make that an awful lot longer? A. Noted. Q. We've got a VAX with two memory controllers, 8 Mb of memory and two battery backups. Occasionally in the past we've had problems with a power failure that wouldn't recover. So we'd call the serviceman in and he'd be unable to fix it. So we swapped batteries in and out a couple of times. We noticed that after a power failure the machine tended to hang while reloading off the floppy. Eventually we set the key to LOCAL ENABLE, and now the machine restarts perfectly. Is this the way it's supposed to be, and where is it documented? A. (Field Service). Doesn't sound right to me - see me offline. (A show of hands indicated that this was a common 'problem'.) Q. The Command Definition Language is real nice, but I look in the fiche and there's all these neat things you do in yours that are not documented anywhere like $TYPE and $FILE and $DATE. What are they and can you document them someplace? A. A lot of them are placeholders for when we get the support into DCL. Some of them work, and most of them don't and all the ones that don't work are equated to one that does work. In V4 a lot more will be supported and documented. Q. In reading the VMS V3 Release Notes, I got excited by the WSEXTENT, so I put a large WSEXTENT on my batch queues, and I have a fairly typical mix of interactive and Batch jobs. On a normal day the Interactive users go home, and the Batch jobs start getting some time, and on a good day the Batch 22 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A jobs all finish before the next morning. On a bad day they don't finish, and at about 11 am the Batch jobs aren't getting any computer time because of their priority but they have all the memory - the interactive jobs are getting all the computer time but all they're doing with it is paging and swapping. Is it true that your Memory Management will allow a Priority 4 Interactive job to be swapped out while a Priority 1 Batch job enjoys a ration of memory greater than its quota up to the limit of its extent? A. Yes it is true. A known problem that we are looking at. A. (from audience) At the start of batch command files do a SET WORK /NOADJUST. This will return it to the state of VMS V2, whereby the batch jobs will use all the memory during the evening when there's nobody there, but will not hog it during the day time. (Warning from DEC: There is a bug in which a job will get trimmed by the swapper to SWPOUTPGCNT and if you have disabled working set adjustment it will not grow, and so will fault it's life away.) A. (from audience) Suspending the batch jobs will cause them to swap out very rapidly. Q. When I do a DEASSIGN/GROUP/ALL to deassign my group logical names they stay there. A. It is a know problem. We're not sure when the patch will be out. Q. Using standard Network Management protocols how can I ask a remote node in a DECnet network what O/S and version it is running? A. Interactively you can use NCP to set Executor to the node. If you want to do it from a program there are undocumented NML functions that might get supported one day, and this might be one of them. One way to work it out would be to set NML$LOG to 000000FF to dump all NML logging and then use NCP to do the function and watch to see what NICE message NCP uses and copy it. Q. Many of us enjoyed the level of detail in the first Internal Manual, and are eagerly awaiting the alleged new version in the Fall. Now that DECnet is a bundled part of VMS when can we expect the same level of detail in the internals manual on DECnet? A. There will be a V3 Internals Manual set published by Digital Publications group in August 1983, but there are no plans to publish any internals information for Network products. The V3 Internals Manual will cost under $50. 23 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A Q. Are you going to support any form of ROLLOUT utility with Restart capability? We need the ability to specify it externally rather than internally via Checkpoint. A. Noted. No current plans but we have had several requests for it. Q. I had a problem when we went to Rev 7. We have two memory controllers and when Field Service installed Rev 7 they put in a new boot floppy which did not have interleaved memory turned on. It worked for a while, but as soon as the system became busy VMS started consuming vast quantities of memory, and the last time I did a SHOW MEMORY, VMS was permanently holding 7900 pages! It would be useful if the boot process did some sort of verification to see if anything was wrong. A. Noted. We will take this up with the memory management developers, none of whom are here at the moment. Q. I log into two terminals, say A & B. On A I do a SHOW PROCESS/CONT with the ID of B's process. If I log off of terminal B and then log back on and do a SHOW SYSTEM terminal A is in an MWAIT state. Why, and can it be fixed? A. There is some suggestion that maybe process B disappeared while holding a special kernel AST for process A that was never completed. but we're not even sure that is possible. It is more likely that you have a resource problem such that process A is working at the limit of one of its quotas (such as BYTLM). To analyze it further go into SDA. Look at process A and examine the first longword of its local event flag wait mask - in an MWAIT that contains the resource number it is waiting for: a small integer that can be referenced to the RSN... symbols in the system map to work out which resource it is. Q. We have some graphics workstations on our VAX which are intelligent enough to support other devices, such as plotters and printers, hanging off them. These devices are hooked up to the VAX via a CSMA/CD type of link, which also has other VAX's attached, and the workstations can tell the VAX when they are 'leaving' it to attach to another VAX. On disconnect we have a VMS job that tries to delete any processes that owned the workstation or its assocated devices in case they did not finish up tidily. However, the printers attached to the workstations are associated with VMS print queues and hence are owned by the Job Controller. So I thought I would modify the job to work out if the device was associated with a queue and stop the queue instead, but I couldn't work out how to find out that information about a device. 24 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A A. If you name the queues according to the physical devices then you could look at the queue file which is mapped as a global section, although we don't document the format and it will be changing in V4. We are also rewriting the Job Controller for V4 and the problem should be alleviated to some degree then, as the symbionts will no longer be subprocesses of the Job Controller, but will be distinct detached processes. Q. I have a strange network with a bunch of researchers who have their own budgets and hence buy their own terminals based on what they think will serve their needs best. This causes several problems, in particular with EDT. Any comment? A. The problem with EDT is that it is not our own, and it is written to run on several other DEC operating systems as well, and the folks that write it are very unwilling to use routines that are not available on all the systems they write for. Q. I have a problem with not being able to distribute object files to people. I have a plain vanilla FORTRAN program. I compiled it on a VMS 3.2 system and made some object libraries and tried to build them downstairs on the 750 to give a demo. It linked OK, began to run OK, but as soon as it reached a run-time FORMAT statement it crashed with INFINITE LOOP IN RUNTIME FORMAT. I rebuilt it from source and it worked perfectly. Why can't I distribute my objects? A. We'll look at it on the machine tomorrow. Editor's Note A subsequent report from the VMS developer who looked at the problem said that different procedures were being used for the link from the object library and the link from freshly compiled object files, and that as soon as that discrepancy was resolved the problem went away. Q. I heard a rumour that EDT had been rewritten so that it wouldn't be such a hog, and was in Beta test. Is that true? A. We do use the newer version and noticed that it is faster, but we don't have any benchmark figures. 25 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A Q. When I look at my load maps from the Linker it tells me everything I want to know about where it found my files. FORTRAN tells me at the top of every page which source file it is compiling, and refuses to tell me where it found the INCLUDE files. Why is that? I use logical names for my INCLUDE files and would like to know which file spec it references. A. Noted. Q. On occasion, to come out with major new releases, I accept that it is necessary to 'break' some existing user code. Before V3, you could submit a multi-file Batch job with various parameters, specifying the /PARAM qualifier, and they would all go to each file in the Batch job. With V3 the subsequent files get nothing at all. I could understand it if it were replaced with something useful, but it just seems to have gone away. A. I don't know why that was done, we will investigate it. Q. You have all these SYSGEN parameters to control paging, but there is nothing about swapping, so that we have no control over who is going to swap or why. A. The Memory Management material in the System Manager's guide does describe where to swap. V3 does have better control than V2. Q. I had a recent problem where the FPA seemed to be off by about 10%. Can you recommend any software-type things to check the FPA periodically to ensure it is on target? A. We have discussed internally the possibility of supplying such a beast, but people are already complaining about the number of process slots VMS takes up, so we are reluctant to do so. Q. Back to the CLD qualifiers. I was told this afternoon that one of them would allow me to pass double quotes without them being eaten up. Is this true and, if it is, can we assume that it is safe to use, and is there anything else like this that we should know about? A. Look at the CLD for the TYPE or PRINT command. It's something like $INFILE, and although undocumented does work, and will be there (and documented and supported) in V4. Q. I'm having some problems with PRO350's on VMS. Theoretically there is a mode where they will emulate a VT125, but there are problems in DATATRIEVE with split screen and shading. These problems do not seem to be solvable in the PROP emulator, so will it be possible to interface the PRO to VMS as itself, and will DATATRIEVE use it correctly? 26 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A A. V4 for PRO support. Ask the Datatrieve developers about Datatrieve. Q. Is there any plan to allow Floating Point keys in Indexed Files? A. No plans at this time. V4 will support Quadword keys, though. Q. I'm responsible for a 780 that runs several projects, some of which have their own machines ranging from 11/23's to 750's. One of my users had written some software to interface to a special piece of hardware on a 750 running V3.1, and he wanted to put it on the 780, running V3.2, for a while. He assured me that it would do no harm as he was using the Connect-to-Interrupt driver, but it failed at the first QIO with bad parameters. When I chased it up I found that the Connect-to-Interrupt driver had been completely replaced when I installed V3.2, although the Release Notes do not mention it. So I replaced it with the one from V3.1 and it worked, but I was surprised nobody mentioned it. A. Q. I need to connect devices other than terminals to serial lines and I need to find out what the specifications for input through the terminal driver are. I've heard rumours that if my devices talk to the VAX through the terminal driver at 9600 baud I'm going to run into problems and I'm wondering quite what is practical and supported, both with and without typeahead. A. There are notations in the SPD giving total character throughput. We have never tried to run 9600 baud flatout, but in a burst it might work, and you will want to use PASSALL mode. A. (From audience) Several people had done it with no problems. Q. I have a system coming soon with RM80's. Can we upgrade the RM80's to RA80's? I am told they have the same disk head mechanism. A. No. Q. I need to be able to transfer files from one user to another and instead of forcing my users to set all their files to allow WORLD Read Access, I thought I'd write a nifty little program, that could be installed with Privilege and would set my UIC, copy the file, and do all sorts of neat things. It worked fine from SYSTEM, but when I installed it with privilege and ran it from a non-privileged account I had problems because when you spawn you lose the privileges that your image is installed with - when you spawn the privileges 27 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A are taken from the user running the program not from the image that did the spawn. Is there any way around this? A. (from audience) If you install the image with SETPRV privilege and use Create Process rather than Spawn you can give the subprocess any privileges you want. A. If the first line in your spawned command file is a SET PROCESS/PRIVILEGE then your subprocess will gain those privileges even though the user does not have them, as long as the image is installed with those privileges. Q. In DCL you can check your verification level with F$VERIFY, but if you set your message level there is no way of working out what it was set to before so that you can set it back. A. In V4 there is a new lexical function that will let you get virtually anything we can think of such as the current protection setting, the current message setting, etc. Q. On our system we have four Versatec V80's, and we're also running DISPLAY, and every time we dump a plot the Print Symbiont, at Priority 8, stops everything! Is there any reason the Print Symbiont is started at Priority 8, and if not is there any way we can get it to start up with some normal priority. A. We think it starts up at that priority because we believed it would not adversely affect the system and it would allow the printers to run flat out. A workaround would be to set the priority of the Print Symbiont from an account with ALTPRI privilege. There is certainly no problem if you set the priority lower. Q. I recently acquired a 730 system and I need to communicate with RSTS. The machines are not DECnetted together and I need to haul RSTS RL02's to the 730 and read them there. There's a program on the RSTS kit, in the Unsupported section, in VAX BASIC. I put it on my 730 in three different flavours - from the RSTS 7.2 distribution, from Patchkit C for V7.2, and from RSTS V8 distribution. The first and third ones don't work at all, and the second one only works after some effort (such as specifying /DEBUG). Can't we have some direct support in FLX or some other utility to read RSTS disks? The only other option is to PIP my files to an RT11 disk and then read them on the VAX with a compatibility mode program, which involves four Operating Systems and seems slightly strange. A. The RSTS people did not ask us if they could ship VMS programs with RSTS. With VMS V4 we have a new native-mode utility, EXCHANGE, to replace FLX, but it will still not talk to RSTS disks, although the requirement will be noted. We have gone out of our way to remove unsupported things from 28 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A our kit because by virtue of the fact that it is on the kit it is perceived as supported, and we will not ship unsupported software. Q. What about TECO? A. The TECO situation has been fixed in the positive direction, and it will be supported in the next release of VMS. (Cheers from audience.) Q. We occasionally like to link our programs in a particular order in memory, and we do a LINK/MAP to check the order. So that we do not waste disk space with extra images we do a LINK/MAP/NOEXE with the result that routines in user-libraries do not get referenced. A. Fixed in V4. Until then you can do /EXE=NL:, which is what the Linker will do for you in V4. Q. There are coding standards for users on what characters are allowed in names and suchlike, and there's a dollar-sign that DEC reserves for its own use. Every time a new release comes out we look at the new features and try to find the coding standards that apply. Our problem started with the release of the Lock Manager when we asked how we could name our system-wide resource locks so as not to conflict with your naming standards, and the answer was "well, we don't have any naming standards, but we're working on them" so the question again is, Where are the naming standards we can follow for naming system-wide locks? A. Our belief is that all DEC products use dollar-signs. If you know of something that doesn't please let us know. Q. I have a user that likes to prowl the system armed with DUMP and one day he was dumping the SYSUAF file while another user logged on, who found himself with no password protection and full privilege. I found this was actually documented as a feature, but would like to know if we are also vulnerable in that mode when running AUTHORIZE? A. No - AUTHORIZE opens the SYSUAF as shareable. To avoid the other problem, your UAF should not have World Read access. The feature that he's mentioning allows you to log in on the console (only) if the UAF is unavailable, i.e. has been corrupted. Q. I'm having problems understanding VMS Concealed Devices - I don't know what's concealed. I get two underscores, and it doesn't seem to behave all that different from the old method. Could you clarify it for me? 29 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A A. The whole business of concealed devices and rooted directories is groundwork for V4 to allow us to do a couple of things - firstly it has the potential of allowing us to replace a device transparently under a running program provided that the running program is not aware of the physical device name; secondly it allows us to conceal the root directory so that we can have multiple root directories without affecting existing software. Q. We received a new VAX which was a Rev Level 7. On Warm Restarts it appeared that the VAX would always hang. We'd wait a while for it to restart, and when it didn't do anything we'd reboot it. We called in Field Service and we noticed while they were working on it sometimes it would go ahead and warm restart. We found out that with Rev Level 7 there was an ECO to cause a 3 second delay before Warm Restart went through, to handle RP07 disks which take a long time to spin up. This was on V3.2. Can you comment on this? A. In V3.0 there was a bug that caused the system to hang for three minutes after Warm Restart, caused by bad passing of arguments to the Unit Initialization routines. We fixed that in V3.1, which is why we are puzzled. A. (from audience) There is a problem with Warm Restart on systems with SI disks attached, even if they're not the system disk. It seems to be a problem with getting the wrong 'chip-set' for controlling the formatting of the SI drives. Q. My question relates to LIB$DO_COMMAND and LIB$SPAWN. LIB$DO_COMMAND allows you to execute a command and leave the program, but LIB$SPAWN is a pain to use. Is there any plan in V4 or later to have a LIB$EXE_COMMAND which returns to the program? A. If you spawn a lot from the same program it is probably worth creating a subprocess to do the commands and communicating with it via mailboxes. If you create the mailbox, then do a $GETDVI on the channel to get the device name and pass that device name to LIB$SPAWN ans both input and output device you can get away with using only one mailbox. Q. There has been a problem in VMS for several years whereby a process is logged out if carrier is lost on a dialup line. Is this fixed in V4. A. Yes. Q. I have a problem in V3 with my shutdown process going into an MWAIT. I found out after a while that this is because of a problem with Broadcast. Will SHOW PROCESS in V4 give any more information about what sort of MWAIT a process is in? 30 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A A. In V4 the Broadcast code has been rewritten so that this problem should not recur. Most of the utilities that display process state will break them down a little more finely than just MWAIT, so that from a SHOW SYSTEM you could see that someone was in an AST wait or whatever. Q. What, if any, details can you give us about ODS-3, or whatever the next version is? A. What we are doing for V4 is not ODS-3; it is upwards-compatible extensions to ODS-2. There are a lot of new security features such as Access Control Lists. Filenames have been lengthened to 39 characters (for compatibility with TOPS-20). Q. What is the DEC-supported way of avoiding the LOGINOUT image execution for non-terminals attached via serial lines? A. Set the terminal line to /NOTYPEAHEAD/PERMANENT. Alternatively you can just allocate the terminal to a process. Q. In V3.3 will the /INTERCHANGE switch be supported in BACKUP? A. Yes - the /INTERCHANGE qualifier is documented in the release notes. It writes a backup save set that is suitable for duplication, i.e. for software distributions. On tape it limits the block size to 8192 (maximum size ANSI tape standard allows) and inhibits the rewrite error recovery which makes BACKUP leave bad blocks on the tape so that you end up with a normally-written perfect tape. It will also be extended in the future to do other things consistent with software distribution - e.g. for V4 BACKUP it will suppress the saving of Access Control Lists, which are not meaningful for interchange situations. Q. Do RP07's do auto badblock detection in software and if not, why not? A. Yes, as for all Massbus disks. Certain versions of the RP07 microcode/HDA and things like that tend to produce Operation Incomplete errors on blocks that are ostensibly bad. I'm assured by the RP07 people that this is an RP07 problem and they are fixing it, so VMS is not planning to take any action about it. Other errors should be trapped by the bad block detection algorithm. Q. We have RT11 running on a number of PDP11's, and VMS on a couple of VAX's which need to communication with each other. If we have a disk used as a standard Structure Level 2 VAX disk and then at some later time we zero it with FLX (/ZE/RT) then it appears to be a nice clean disk you can write on, but in reality it still remembers its former owners and if you accidentally mount it non-foreign VMS will try to repair the 31 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A "damage" and to undo what's been done by the data acquisition on the 11's. Will EXCHANGE fix this? A. EXCHANGE will try to find all the places that FILES-11 stores home blocks and zero them all. EXCHANGE has no relation to FLX in terms of source code at all, and so has probably fixed many other FLX problems as well. Q. When is DEC's ADA going to come out? A. ADA is a very important language to us, but we cannot talk about schedules yet. Q. What happens if I put an RP04 on a VAX? A. It should work. Q. I have a VAX 11/780 running VMS talking to some 11/70's running a semi-supported Operating System called IAS via DECnet on a semi-supported device called the PCL-11B. DECnet works fine most of the time, but if I am on a VAX system and run EDT to edit a file on the IAS system everything seems to work fine until I look for the file afterwards and find that all I have is a journal file. Anybody know what is wrong? A. Noted. Q. We have a VAX 11/780 and two generic print queues. It seems that when we do a STOP/QUEUE/NEXT on one of the queues, the other equeue is stopped as well until the job on the stopped queue finishes. A. Will be fixed in V4. Q. When defining root directories the syntax is precisely Device:[Dir.] - are there any plans to extend it so that I can root sub-directories? A. We are extending this mechanism and I think this is one of the things we are considering for V4. Q. We have bought a very fast tape drive for doing BACKUP's to, and yet on a machine with nothing else happening I cannot drive the tape drives flat out, and yet the Null job is still getting about 60% of the CPU. Where is the bottleneck? A. The bottleneck is in your disks. For every file BACKUP has to go off and read the file header and then reposition the disk head to read the file and that takes time. A TU78 at 6250 BPI has a raw transfer rate of 750 Kbytes/sec, while the raw transfer rate of an RP07 running in interleave mdoe is only 1.2 Mbytes/second and the head repositioning cuts down that rate drastically. There is another limiting factor in BACKUP, even when backing up a large contiguous file so that 32 PAGESWAPPER - July 1983 - Volume 5 Number 1 Spring 1983 VAX Advanced Q & A there are no head motion problems, as BACKUP needs CPU resources to compute the XOR blocks and the software CRC, all of which adds up to about 5 microseconds per byte transferred, which gives a limiting data rate of 200 Kbytes/sec. If you turn off the CRC and use very large blocks (10 - 20 Kbytes) you can drive the TU78 much faster, subject to the fragmentation of the disk. 33 PAGESWAPPER - July 1983 - Volume 5 Number 1 VAX CI Software Systems VAX CI Software Systems Paul Beck Digital Equipment Corporation Nashua, NH Dave Murphy, Session Chairperson Massachusetts General Hospital Boston, MA Reported by Frank Fluke, DECUS Scribe Service The VAXclusters are VAX/VMS computer systems consisting of multiple, loosely-coupled, cooperating VAX/VMS CPUs operating with global, shared, intelligent, mass storage subsystems interconnected by a high-speed bus. It is important to note the distinction between the cluster system and Local Area Networks (LANs). The processors in clusters cooperate much more closely than nodes in LANs so that individual processors in a VAXcluster can share data at high speeds without special programming or procedures in the application. Furthermore, each node has its own operating system but is covered by a tighter environment than a network is. This article is a discussion of present and planned VAXcluster software functionality. The first question to be asked is what capabilities do we want to provide? The following characteristics were listed: 1. Incremental system expansion 2. Higher total system availability 3. Increased data availability 4. Transparent data sharing among nodes 5. Broader cost/performance range 6. Benefits from economies of scale A VAXcluster is a loosely coupled system falling somewhere between a single system and a Local Area Network, and has the following attributes: 1. "Single Machine Room" environment 34 PAGESWAPPER - July 1983 - Volume 5 Number 1 VAX CI Software Systems 2. Managed as a unit 3. Multiple autonomous processors (each with its own operating system) 4. Based on high speed packet-oriented interconnect 5. Processors fail independently, not together 6. Shared file system 7. Assumes generous memory per node One of the characteristics of the VAXcluster is the total connectivity of VAX/VMS nodes. The interconnect used is a high-speed, fault-tolerant, dual-path bus. Each node in the cluster not only can see, but must see every other node in the cluster thereby maintaining the integrity of the shared file system and the shared manager, and preventing a single failure from interrupting the cluster operation. Another characteristic of the cluster is the total accessibility of the mass storage devices. The Distributed Lock Manager is a tool for synchronizing access to shared resources. It provides cluster-wide locks which survive node failures and is the basis for many distributed capabilities including: cluster-wide file system, cluster-wide journals, distributed job controller, and user-written applications. Because the lock manager is the low-level facility on which other facilities are stacked, it may be considered the glue of the system. The lock manager also implements deadlock detection preventing a circular list of processes from waiting for each other. The Shared File System performs like local disks. It is an all single-system file and records operations available across the cluster. Other attributes include: 1. File and record control distributed over VAX CPUs with no central bottleneck, and no single point of failure. 2. File ACP and RMS use Distributed Lock Manager. 3. Disk server emulator for local disks (preserves hardware investments). 4. High availability features supported. 5. RMS locking to record level across cluster. 35 PAGESWAPPER - July 1983 - Volume 5 Number 1 VAX CI Software Systems Two of the high availability features which will not be discussed are dual-ported disk fallover on HSC50 and shadowed disk volumes on HSC50. The high availability features which will be reviewed include: common journaling facility, recovery units, RMS journaling, and checkpoint/restart facility. The Common Journaling Facility protects your files from corruption in a number of ways. You may bring your backup and roll it forward to the state just before the corruption, or you may similarly take the current state and roll it back to the previous state to remove faulty input. The facility also allows the recording of independent events and provides support for atomic operations. Two other characteristics are support for user-written recoverable facilities and RMS support. A Recovery Unit equals the atomic operations within a single process. Take a bank account for example. If you are transferring money from your savings to your checking account, you don't want the system to crash after it has withdrawn from the savings but before it has deposited in the checking or you will have lost your money. If both operations are declared a recovery unit, then, if either operation is not successful, the unit is rolled back to the previous state. Thus, unless it is a complete success, the roll back is automatic when a node fails. RMS Journaling has the following basic characteristics: 1. Automatic for marked files 2. Transparent to the User 3. Support for all journal types (After Image, Before Image, Audit Trail, and Recovery Unit) 4. Support for all RMS file organizations The Checkpoint/Restart Facility has these attributes: 1. Save state of application at user-defined points 2. Application can be restarted after system failure or process failure by either automatically on same node or with manual intervention of another identical node 3. Preserves investment in computation 4. Based on Recovery Unit Facility With the Distributed Job Controller, Job and Print queues are cluster-wide. A Batch or Print queue can supply jobs to multiple nodes allowing for the efficient use of resources. 36 PAGESWAPPER - July 1983 - Volume 5 Number 1 VAX CI Software Systems The Cluster is a single protection domain managed as a single entity. Each node joins the cluster when booted and all VAX nodes run the same VMS version/update. Adding equipment has a minimal impact on users and applications. Cluster growth now moves from the VAX-11/750 to the basic cluster consisting of two 11/750s linked by a Computer Interconnect. You may increase capacity by simply adding one 11/780 and HSC50 support, and the number of further combinations is endless. Question and Answer Period: Q. Can I get an HSC50, several drives on it, and create a disk server island and have several VAXs connected with a CI able to transparently share data? A. At present the major restriction with using the clusters in 3.3 and 3.4 is that when you have two processors getting two files in the same system on the same disk, only one of those processors can write to it and the other processors will be read only. It's not until the next major release that we will have the full file sharing. ----- Q. At this point you're saying that if you have three disks on an HSC50, and no local disks, that some of the disks have to be specifically mounted to read only to some of the systems? A. With 3.3 and 3.4 I believe that each disk can only be mounted writeable from one system and read only from all of the other systems. ----- Q. Does declaration of a recovery unit necessarily apply simultaneously to all RMS files being journaled you have open, or can you select which of your currently journaled files this recovery unit applies to? A. Any RMS file which is marked for RU journaling will be covered by that recovery unit. ----- Q. So the recovery unit is a per process declaration? A. A recovery unit and a checkpoint both are limited to the confines of an image and will go away on image exit. ----- Q. We cannot from one unit have two asynchronous recovery streams going using the RMS? A. There is one recovery unit per process at a time. ----- Q. On the checkpoint/restart facility, since most users aren't 37 PAGESWAPPER - July 1983 - Volume 5 Number 1 VAX CI Software Systems bright enough to checkpoint themselves, is it going to be possible to force a checkpoint down someone's throat? A. No the checkpoints are issued from within the process at user-defined points. ----- Q. If you have three VAXs, a CI, HSC50, and multiple disks and you want to share data among all three systems in a read-only fashion like copies of corporate data bases, under what release of VMS is that possible? A. 3.3. ----- Q. If you want one of the systems to be able to write and the others only to read? A. That's also 3.3. ----- Q. But sharing at less than a volume level in a write mode? A. 4.0. ----- Q. What is the overhead in terms of CPU time, disk access for the distributed lock manager, checkpointing, and the emulator? A. For the distributed lock manager there is no disk access overhead so I can't give you any numbers as far as that goes. For checkpointing it would depend on how frequently you take checkpoints. ----- Q. Are you writing to disk on a checkpoint? A. Yes, anytime you write to a recovery unit, you write to a disk. As far as CPU overhead, I don't have any numbers so it's hard to go into detail for that at this point. ----- Q. When if ever will it be possible to share a common SYSUAF so that a user can log into any CPU and basically his authorization record and his default directory will be the same so that it doesn't matter? A. Version 4. 38 PAGESWAPPER - July 1983 - Volume 5 Number 1 UNIX Hints and Kinks UNIX Hints and Kinks Armando P. Stettner Digital Equipment Corporation Merrimack, NH Dorothy Geiger, Session Chairperson Cal Poly San Luis Obispo, CA Reported by Dorothy Geiger, DECUS Scribe Service The UNIX Hints and Kinks session was led by Armando P. Stettner and included panelists Joe Sventek, Norman Wilson, Bill Burns, and Vance Vaughn. Questions were as follows: Q. Does UNIX have support for networking? A. VMS/UUCP is being examined, as is DECnet for VAX/UNIX. Software is available to support "tar" under VMS and TCP/IP under UNIX. ----- Q. Does 4.2Bsd have new networking software? A. 4.2Bsd provides rich support for interprocess communications between machines, including named sockets for servers, TCP/IP, and Ethernet. In addition, processes do not require the same parent for interprocess communications. ----- Q. What is the Software Tools Group's Virtual Operating System? A. VOS is NOT UNIX, it is an entire program development environment which is based on the Software Tools book by Kernighan and Plauger. It provides a variety of shells, tools and utilities, and has been implemented for various operating systems such as RSX and VMS. The VOS software is available on the DECUS SIG tapes. ----- Q. Will DIGITAL's release of UNIX have system performance and monitoring tools? A. DIGITAL's current release includes "vmstat", "iostat" and kernel profiling. There are no firm current plans for more. ----- 39 PAGESWAPPER - July 1983 - Volume 5 Number 1 UNIX Hints and Kinks Q. Can files be accessed across the Network? A. 4.2Bsd provides for symbolic "links" across network nodes. In addition, work has been done elsewhere to allow file systems to be "mounted" across nodes. 4.3Bsd has transparent file access as a design goal. ----- Q. Will DIGITAL's VAX/UNIX release have the same basic "goodies" as 4.2Bsd? A. No decision has been made on 4.1Bsd vs 4.2Bsd. Because 4.2Bsd is a new release, assessments of suitability and reliability are premature at this time. ----- Q. How does one get 4.xBsd? A. 4.xBsd may only be supplied to holders of Bell UNIX source licenses. ----- Q. Is DIGITAL shipping UNIX source licenses? A. No, pending resolution of legal questions with Bell. ----- Q. Does DIGITAL support UNIX in the VAX cluster world? A. Not at the present time. ----- Q. It would be highly desirable for DIGITAL to provide support in this area. A. Noted. (APS) ----- Q. Are there bugs in the 4.2Bsd DMF-32 driver? A. None are known to the panel. However, be aware that the DMF-32 driver only uses the serial asynchronous portion of the DMF-32, and that the DIGITAL board have modem control on only two lines of the eight on the board. ----- Q. Will DIGITAL continue to provide free source to device drivers as has been done in the past? A. Hopefully. ----- Q. What is the state of Berkeley INGRESS? 40 PAGESWAPPER - July 1983 - Volume 5 Number 1 UNIX Hints and Kinks A. Good. The VAX architecture necessitates fewer processes per user than the PDP-11 architecture. This greatly enhances performance. INGRESS is available on the current 4.xBsd tape. ----- Q. Are there any changes in 4.2Bsd from 4.1Bsd? A. Many. For example, Bsd4.2 has a new file system which is much better for high I/O bandwidth applications such as VLSI design graphics. ----- Q. What about the new F77 compiler on 4.2Bsd? A. The old compiler produced code which executed about half as fast as the VMS Fortran compiler. With the new compiler, execution speed is about the same as VMS Fortran, but compile times are SLOW. ----- Q. Are there incompatibilities between 4.2Bsd and 4.1Bsd? A. Executable images will run on both systems. Problems arise when munging on directories, since directory formats have changed. ----- Q. Does 2.9Bsd include job process control via C shell? A. Yes. ----- Q. How does System V differ from 4.xBsd? A. System V supports a two UNIBUS configuration on the VAX-11/780 with distinct restrictions on device placement. 4.2Bsd supports a four UNIBUS configuration with no restrictions on device placement. In addition, VAX-11/750 support is new on System V but not on 4.2Bsd. 41 PAGESWAPPER - July 1983 - Volume 5 Number 1 Writing C Code for VAX/VMS and UNIX Systems Writing C Code for VAX/VMS and UNIX Systems David Moore Digital Equipment Corporation Nashua, NH Jim Livingston, Session Chairperson Measurex Corporation Cupertino, CA Reported by Todd Spangler, DECUS Scribe Service There are several problems when writing portable C code. There is no formal standard for C. Confusion exists between boundaries of C and UNIX and there is a general lack of awareness of the problem. David Moore of Digital Equipment Corporation described the differences between UNIX C and VAX/VMS C. VAX/VMS does not use UNIX-style file specifications, fork and exec sort utilities on standard I/O, command line parsing (piping and file redirection), and other routines not in the emulation set. On VMS, RMS provides an RTL stream which is compatible with that of the UNIX stream system. Exec-family RTL routines only provide sharable reads between parent/child directories and only initialization with "=" is supported. Also supported are the = and *= operators. To help maintain the portability of C code, it is advisable to keep track of size of data items. On the PDP-11, a long variable is equivalent to a short variable as far as memory is concerned. On VAX, a short variable is half the size of a long variable. The memory order and continuity must also be kept track of as VMS can have variables that exist but that do not exist in storage. When refering to this variable with a pointer, there will be an error message created since the variable cannot be found. On VMS, the layout of program address space is important, especially uninitialized pointers, end, edata, and etext. In UNIX, there is a zero pointer, but on VMS this pointer is protected. Unlike UNIX, characters are signed on VMS. Pointer/integer exchanges are possible, but not portable due to size conflicts. External identifiers on VMS are 31 bits long and on UNIX are 6 to 8 bits long. Unlike VMS, it is possible to have holes in the structure alignment. The order of operation on VMS is only forced when using the COMMA, logical AND,and bitwise or logical OR. To be safe, one should not rely on character set dependencies (VMS uses 7 bit ASCII). VMS does not have an ASM program. 42 PAGESWAPPER - July 1983 - Volume 5 Number 1 Writing C Code for VAX/VMS and UNIX Systems Basically, rules to follow in making C code as portable as possible are to use DEFINE as much as possible and stay away from manifest constants. Make use of common header (.h) files in order to keep some record of system dependent constructs, use SIZE OF to keep track of data, and use common routines with all C libraries. When writing non portable C, one should choose the right support environment. Making use of all the compiler capabilities is important. Using symbols like 'vax', 'vms', and 'vaxlic' will identify the environment to the compiler, while using include text libraries will improve optimization. Allowing the compiler to manage temporary files and assign registers will increase portability. Using constant folding, DEFINE, MODULE, INCLUDE libraries, passing constants by reference, compiler listings including symbol tables, cross references, and preprocessor substitutions will also be helpful. The proper use of PERROR RTL routines to diagnose errors is desirable. With these things in mind, one can consider the system, use the suggestions, and be on the road to writing portable and non-portable C code. 43 PAGESWAPPER - July 1983 - Volume 5 Number 1 Q-Bus/UNIBUS Hardware Hints and Kinks Q-Bus/UNIBUS Hardware Hints and Kinks Wayne Kesling, Session Chairperson Monsanto Research Corporation Miamisburg, OH Reported by Marc Caffee, DECUS Scribe Service During this session different people from the audience presented problems they had with their system and if there was one the resolution to the problem. One of the more frequently appearing problems occurs on VAX UNIBUSs. Sometimes failures occur on machines having more than one high speed DMA device. The problem can be fixed if the DMA device can be slowed down. The details on the timing on the DMA bus transfers was not available at the session however they are available and hopefully will be appearing in a newsletter shortly. Similar problems apparently occur on at least on VAX-11/780 using a combination of foreign DMA devices running UNIX. There were a variety of complaints and suggestions dealing with RK06s and RM05s. In particular on the RK06 the supports holding the drive motor give way and the motor falls on a transformer located underneath it. However, before it does this it does make a somewhat loud noise. Don't just assume it needs some lubricant, rather check and make sure the supports are not wearing out. Also on the RM05 there are rubber parts that the door closes against. If these are loose, pull them off before they fall into the disk. If they're not loose they probably won't cause a problem. The moral of the story is that they should be checked frequently. Also, caution should be used if you have to move a RK06 bus cable. If it is not handled gently there is a possibility of damaging it. Finally, there is a diagnostic available on the RK06 called operator intervention diagnostic. This diagnostic checks out the batteries that operate the head retractor. If you are powering up a unit that has been sitting idle for a while be sure and check this out. Disk Drives: The word is now that the RA81 disk drive is no longer on engineering hold, however there was no sure word on when retrofits would be available on RA81s already in use. Hopefully it would be soon. 44 PAGESWAPPER - July 1983 - Volume 5 Number 1 Q-Bus/UNIBUS Hardware Hints and Kinks If your RL02 temporarily goes off line check the sector transducer. Also, the RL02 ready lights burn out quite frequently. Additionally, an old RL02 controller will not necessarily work on a 11/23 backplane. On an 11/60 running RSX the RA80 does not spin up after a power failure. The solution to this is to "mount" the disk. The process of accessing the disk will spin it up. This might seem strange but it's true. RX02s are not supported on the 11/23. At this time the resolution is to fix the driver in lower memory. There are at least two outside vendors that supply boards which will support floppies. In connection to the Q-Bus problem, it is possible for a parity error to occur when using the RX02s. This problem occurs on the 22-bit system because of the extra lines in the bus. The 18-bit machine apparently never detects this error. NASA has conducted a study on head wear versus relative humidity for high speed disks. The conclusion is that for best head wear the humidity should be kept between 40 and 50 percent. Interfaces: In using the DMR-11 with V3 of VMS you can go to full duplex. However, you have to remove the W17 jumper or else things will be slower than half-duplex. The DMF-32 documentation is at odds with reality. BEWARE!! A DLV11J can be used as an interprocessor link. If two or more of the lines are used heavily the DLV11J will write PSs and PCs all over memory. The problem involved here is really quite general in that the DLV11J came into being before 4 level interrupt. In other words this device doesn't check other device priorities and this will cause severe problems. The resolution to the problem is to have this device (or any other similar device) placed behind the devices which have 4 level interrupt. You might have problems if you try to connect a LP11 to a DMF-32. The LP11 has a 40 pin connector that doesn't fit into the DMF-32. A cable which might work is CSS LXY21-PA. It is possible to transfer disks from a VAX to a DEC-10, however there is one problem which must be overcome. The VAX MASSBUS computes parity on 18-bits. The DEC-10 checks parity on 16-bits. In order to make the transfer work the two high order bits must both be 0's or 1's. Other: 45 PAGESWAPPER - July 1983 - Volume 5 Number 1 Q-Bus/UNIBUS Hardware Hints and Kinks On an 11/34 after a RSX-11M 3.2 SYSGEN on RK05s the system refused to boot after several days. RT-11 and the diagnostics ran fine. only RSX wouldn't run. It turns out that the ribbon connector on the DZV11 raked across the memory card and caused the problem. Diagnostics are really scarce on the VAX-11/730. If you have some technical questions contact Tom Provost, M.I.T./Bates Linear Accelerator, Middleton, MA. 46 PAGESWAPPER - July 1983 - Volume 5 Number 1 Security Mechanisms for VAX/VMS Security Mechanisms for VAX/VMS Mark Pilant Digital Equipment Corporation Nashua, NH Douglas Brown, Session Chairperson Sandia National Labs Albuquerque, NM Reported by Omar El-Ghazzawy, DECUS Scribe Service The session on VMS security, presented by Mark Pilant of DIGITAL, discussed various security mechanisms and their implementation in future releases of the VAX/VMS operating system. The security mechanisms covered were access control list (ACL), non-discretionary controls and audit logging. The term security was applied to the need to protect information from unauthorized distribution and/or modification. The protection services required to achieve these objectives were the ability to grant or deny a request, and the ability to monitor request. Currently, the VAX/VMS (version 3.2) operating system's main security mechanism is the use of UIC to determine access. In the security environment each request is comprised of three components : 1. object - anything which is protected (i.e. file, disk) 2. agent - anything which requests access to an object (i.e. process, device) 3. identifier - a unique method of identifying the agent. In the VAX/VMS environment the identifier is externally represented by an alphanumeric name and internally by a unique 32-bit number which may be defined at several levels - system wide, group wide and per user. In the access control list (ACL) mechanism an identifier list composed of access control entries (ACE) is constructed that determines what access is to be allowed by an agent to an object. Some specifications on ACL are : 1. an ACL may contain one or more ACEs 47 PAGESWAPPER - July 1983 - Volume 5 Number 1 Security Mechanisms for VAX/VMS 2. an object may have only one ACL 3. for files, ACLs are propagated from the previous version of the file, or the parent directory. A second mechanism is the use of non-discretionary controls which provide protection by containing access to an object by an agent. In contrast to the previous mechanism the owner has no direct control over this method of protection, all changes in protection require the intervention of the system manager. Two forms were presented : 1. Integrity defined as a measure of trustworthiness. 2. Security defined as a measure of sensitivity. The integrity level, which ranges from 0 (least trustworthy) to 255 (most trustworthy) is used to control modifications to the objects. It properties are such that an agent cannot read less trustworthy information or write more trustworthy information than he is allowed access to. In addition integrity compartments can be created which are functions of 64-bit mask identifiers. The security level is identical in structure, it also ranges from 0 (least security) to 255 (most security), and is used to control access to the objects. Its properties are that an agent cannot read more sensitive information or write less sensitive information then he is privileged. Similarly, security compartments can be created which are functions of 64-bit mask identifiers. During the question and answer period the addition of group level bypass privilege to the VAX/VMS operating system, and a mechanism for the encryption of data transmitted via DECnet were mentioned by DIGITAL personnel as possible future additions. 48 PAGESWAPPER - July 1983 - Volume 5 Number 1 VAX Security Panel VAX Security Panel Stephen Tihor New York University New York, NY Ross Miller Online Systems Spokane, WA Dave Schmidt Management Science Association Pittsburgh, PA Frank Kieltyka Los Alamos National Laboratory Los Alamos, NM Almon Sorrell Westinghouse Defense and Electronics Center Baltimore, MD Andy Goldstein Digital Equipment Corporation Nashua, NH C. Douglas Brown, Session Chairperson Sandia National Labs Albuquerque, NM Reported by J. Rick Mihalevich, DECUS Scribe Service Dave Schmidt from Management Science Association discussed security from the commercial perspective. Security on large networks is an issue of great concern. Special commercial security concerns are: 1. Client data protection 2. Accounting protection 49 PAGESWAPPER - July 1983 - Volume 5 Number 1 VAX Security Panel 3. Personnel and planning data protection 4. Simple pitfalls Addressing the problem of client data protection requires consideration and monitoring of a variety of computing functions. First, a consideration of each client as a unique group must be maintained. Access across clients should be protected to prevent "world access" to a client's directory; In addition to online consideration of data protection, there is an equally pressing issue of offline data protection. Offline procedures must be monitored to prevent accidental or unauthorized access. Possible areas of security penetration are uncontrolled hardcopy output, backup tapes, communication lines, data tapes, and delivery and shipping mistakes. One goal of commercial security is to provide an easy way to use tool for controlled access to client directories by support persons. In this case, the SETUIC (fall 1979 SIG tape) was implemented. This allows for installed images, screens UIC access by a control list, and can be modified to log access if desired. In terms of accounting protection, SNDACC allows entry of transactional or value added billing entries to the system accounting file, however this is not available from DCL. SNDACC does provide a secure record that is difficult to transfer with and can be used with a command procedure and user written program to do royalty record keeping for leased software. The second approach in accounting protection is to provide a tool for support personal to easily change the VMS account field in the billing records with which they work. In order to accomplish this goal the use of ACCSET (fall 1980 SIG tape) was used as will as the installing of image. The issue of personnel and planning data protection has some very serious security protection problems. If it is essential that security be maintained with total certainty, then a microcomputer and its processes in a stand-alone environment is the only alternative. It is impossible at this time to protect data from system programmers and managers. Even the use of data encryption is not a viable alternative. A dumb terminal requires plain text data transmission that can be intercepted easily. Also, SDA and other tools allow investigations of a running process's memory space. It is possible for a system manager to become any user, allowing several methods in which to penetrate the system. 50 PAGESWAPPER - July 1983 - Volume 5 Number 1 VAX Security Panel In summary there exist some simple pitfalls. Awareness of these and the implementation measures to prevent them will do much to improve protection. The first suggestion from this presentation is: do not install AUTHORIZE or INSTALL utilities. Secondly, monitor and require management accountability of all users with highly privileged accounts. This is done knowing that most computer crime is traced to a person in this category or one with access to these accounts. Also, try to actively penetrate your own system and involve senior management in the monitoring and accountability. Thirdly, don't leave terminals logged in and unattended. Fourth, thoroughly check all details of changes to operating procedures and computer room environment. The fifth suggestion given was do not write down and file passwords in a central place for record keeping convenience. The final suggestion given was not to rely on user ignorance of VMS. There are currently computer science majors being graduated with 4 years of VMS experience and growing numbers of other persons are now familiar with VMS. Ross Miller of Online Systems, another representative from the commercial sector, was the second presenter. In regard to complex security systems, it was mentioned that one can reduce the number of combinations and permutations with simpler or more manageable security systems. Another item of concern is manager mobility within a system and their availability to a number of passwords, therefore increasing access and penetration protection. To combat the problems of system penetration the implementation of the "CHANGE" facility was suggested. This facility removes default devices and default directories. Another feature of this facility is that it monitors logon sessions and creates a record of user/time/location information. A final precaution is the use of time out screen management techniques. This requires the user to maintain interaction with the system otherwise termination of the job will occur. Steve Tihor from New York University discussed the problems of security in the academic world. In the academic computing world there are often users which differ greatly, from faculty research and records to administrative functions to student assignments and projects. The challenge of system security is to find the middle ground between too much access to system and too much security. Users are in effect, handcuffed. Efforts to discourage system penetrations by the tracking of logons are often very effective. This produces a thorough report of all sessions. 51 PAGESWAPPER - July 1983 - Volume 5 Number 1 VAX Security Panel The selection of passwords are often too predictable and sometimes result in a problem with penetration within certain system domains. Frank Keieltyka of Los Alamos National Labs who works in the classified environment addressed the network security and the use of proxy logins. In protecting the default DECnet account the best solution was not to have one. Otherwise prevent interactive use, prevent batch jobs, and prevent "task" access. The proxy login allows for selective mapping between systems' user bases. however this is an experimental, unsupported implementation in VMS V3.0. The DECnet access control provides default access. Which is non-selective, insufficient protection, and insufficient accessibility. The explicit access control, on the other hand, is cumbersome, hard to audit, and there is risk of password compromise. Current vulnerabilities are wire tapping, integrity of remote systems and node impersonation. Future directions will require greater flexibility, possibility of encryption, and the use of non-discretionary security support. Almon Sorrell of Westinghouse defense and electronics an Department of Defense contractor. This presentation addressed the problem of label marked printouts with classified levels. The implementation of a utility called "CLASSPRIN" and another utility used with great success which is available through Software Services. Three comments were made regarding general system security. the first is the use of auto logout utilities to protect unattended terminals. Secondly, protecting against field service personnel who use the same password between classified and non-classified routes, and the use of patterned passwords. The final comment is that PFT systems are not suitable for privileged files. 52 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List Fall 1983 System Improvement Request List Everyone has an opinion about what is right or wrong with VAX. Here is your chance to influence the directions of future DEC development. The VAX Systems SIG System Improvement Request program provides a major vehicle by which the VAX user community expresses its concerns and desires to Digital. Attached you will find the current collection of System Improvement Requests and a ballot form on which to record your preferences. Please take the time to review the enclosed SIR's and assess their effect on your use of VAX's. Rank your favorites by assigning them a point value. Please return your ballot AS SOON AS POSSIBLE. Given the early date for the Fall DECUS Symposium, ballots received after September 26 cannot be counted. The results of the ballotting and DEC's responses will be given at the VAX SYSTEM SIG SYSTEM IMPROVEMENT REQUESTS session of the Fall DECUS Symposium. Instructions You have a total of 100 points to allocate among the SIR's on the ballot. The more points you allocate to a particular SIR, the more strongly you wish to see this feature included in the VAX or VMS. You may assign your points in either a positive sense (to encourage the change) or a negative sense (to discourage the change). In order to assure a wide range of improvements, we have limited the number of points that may be allocated to any SIR to 10. To allocate points to a SIR simply record the number of the SIR in the column labeled SIR NUMBER, and the number of points to allocate to it in the column labeled POINTS. Remember only 100 points total (absolute value). Please note that any ballot not following the points assignment rules, or not specifying a DECUS membership number will not be counted. Only one ballot per member will be accepted. 53 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List SIR: F83-001 Abstract: Provide a terminal characteristic which supports 8 bit characters as well as flow control. Description: When controlling special devices through a terminal line, it is sometimes necessary to use PASSALL mode to transmit 8 bit characters. However, it is then impossible for the device to use the XON/XOFF characters to provide flow control. There should be a terminal characteristic (independent of PASSALL and TTSYNC) which allows this type of processing. SIR: F83-002 Abstract: Make the traceback facility callable by a user program. Description: It would be useful to be able to access the traceback facility of the VMS default condition handler via a call from a user program. SIR: F83-003 Abstract: The DEBUG command EXAMINE should have a /DESCRIPTOR qualifier. Description: It should be possible to use the DEBUG EXAMINE command to display a data item by specifying the address of its descriptor. This function should be independent of the programming language in use, and would be useful for dynamically allocated data items, particularly character strings. SIR: F83-004 Abstract: Compilers should have an option to produce narrow listings. Description: Many inexpensive serial printers only handle 8.5 inch wide paper. It would be useful to have the VAX compilers produce listings with a user specified width. This would allow using 80 columns on 10 character/inch printers; 96 columns on 12 character/inch printers, etc. This capability was available in FORTRAN 4-PLUS on PDP-11's. SIR: F83-005 Abstract: Provide an alternate interpretation for the TAB 54 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List character in EDT Description: It would be useful in the EDT editor to be able to set tab stops at specific columns. When the TAB key is pressed, an appropriate number of spaces should be inserted to position the cursor at the next tab column. This provides compatibility with programs and printers which do not recognize the TAB character. It also simplifies preparing input for programs which are column oriented. SIR: F83-006 Abstract: The $GETJPI service should return the image count. Description: All of the information written to the VMS accounting file is also available during program execution, except for the image count. This value should also be accessible. SIR: F83-007 Abstract: Provide a macro capability within Digital Standard Runoff. Description: DSR should be expanded to include a "DEFINE COMMAND" command. The user could custom tailor DSR for individual needs. The defined command would consist of a block of text and other DSR commands to be inserted. Arguments to defined commands should also be supported. The arguments would be substituted for predefined strings within the body of the command definition. SIR: F83-008 Abstract: Improve the accuracy of VMS CPU time counting. Description: CPU usage for a process should be kept to the microsecond by recording the interval timer when a process gives up the CPU. This would eliminate inaccuracies in timing proceses which become synched to the VMS timer interrupt and execute in short (< 10 msec.) bursts. SIR: F83-009 Abstract: Add a "speed limit" for autobaud terminal lines. Description: There should be a way to set the maximum baud rate allowed on an autobaud line. Attempts to login at a higher baud rate would be rejected. This limit should be selectable for each line via SET TERMINAL. 55 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List SIR: F83-010 Abstract: Provide the ability to add user-written lexical functions to DCL. Description: A need exists to incorporate user written lexical functions in DCL. These could be used in command files and return values to be used in assignment or conditional statements. SIR: F83-011 Abstract: Provide for time stamping of DCL statments. Description: DCL should have a VERIFY option which causes each DCL statement to be time stamped when it is echoed. This would be especially useful to trace the progress of long running batch jobs. SIR: F83-012 Abstract: The VMS Run-Time Library should provide access to the branch on bit interlocked instructions. Description: The RTL currently supports high language interfaces to several useful VAX instructions. The ability to perform interlocked tests from all high order languages would be desirable. SIR: F83-013 Abstract: LIB$SPAWN should have a mechanism for specifying the privileges to be passed to the subprocess. Description: It is currently extremely difficult to pass extended privileges to a subprocess created via LIB$SPAWN. It would be useful to provide a means by which the subprocess privilege mask could be specified. It should be possible to pass privileges which are available in the calling image as a result of it being INSTALL'ed. SIR: F83-014 Abstract: Flag unprintable files in the file header. Description: There should be a file header flag for files which should not usually be printed (i.e. .OBJ or .EXE files) which would be set by compilers, linkers, etc. and which would cause a TYPE or PRINT command to return an error (unless overridden). 56 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List SIR: F83-015 Abstract: LIB$TPARSE should be expanded to include all common VMS objects. Description: There should be LIB$TPARSE codes to specify parsing of all common VMS objects. Such objects include floating point numbers, times, delta times, file specs, node names, usernames, etc. If the syntax of such an object were to change in a future release, the use of a standard parse code would ease the transition to new syntax rules. SIR: F83-016 Abstract: Expand the current logical name capabilities for specifying directories. Description: Multi-level rooted directories should be supported, e.g. DBA0:[TOP.A.B.]. Also, explicit directory names in a file spec should be appended to those in a logical name. For example, if LOG2 was DRA0:[ALPHA.BETA], LOG2:[GAMMA] would access DRA0:[ALPHA.BETA.GAMMA]. SIR: F83-017 Abstract: Allow specifying disk parameters with stand alone BACKUP. Description: Stand alone BACKUP should be able to initialize disks with different parameters than specified on the save set tape. Possibly the INITIALIZE command could be added to the stand alone BACKUP environment. Without this capability, there is no way to change the cluster size on a single disk system. SIR: F83-018 Abstract: Make CPU time a deductable and pooled quota. Description: This change would allow a subprocess to share the total CPU time quota with its parent. The subprocess could give back what it didn't use. SIR: F83-019 Abstract: Provide a memory for the last n default directories. Description: It would be useful to have VMS remember the previous n default directories. This would make it easy to 57 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List move back to the previous default directory, or back two directories, etc. SIR: F83-020 Abstract: Modify date should always be set when files are copied. Description: When a file is created using the COPY command, the copy should always receive a new modify date. Otherwise, it becomes difficult to rationally perform cleanup of old files on a disk. A "old" file may in fact have been recently created by copying from another disk. SIR: F83-021 Abstract: Compilers should allow certain symbols to be suppressed from listing. Description: Many languages allow using "include" files or macros to define system symbols. For any language which allows this, it should be possible to suppress unreferenced symbols of this type from appearing in symbol tables and cross reference listings. SIR: F83-022 Abstract: Interactive utilities should allow for script creation. Description: Allow script creation from interactive utilities, especially SYSGEN. These scripts could then be edited and reused as input to the utility. SIR: F83-023 Abstract: Add a $SETJPI service. Description: There should be a $SETJPI service, similar to $GETJPI, which allows non-privileged portions of the process data structures to be set. Of particular interest is the ability to set the ACCOUNT field. SIR: F83-024 Abstract: Provide a virtual terminal driver. Description: There should be a virtual terminal driver for VMS. Such a pseudo-device would allow a program to easily 58 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List communicate with a process, where the process was running software which expected to perform device-dependent I/O on terminals. This would be useful for terminal journaling, foreign networks, etc. SIR: F83-025 Abstract: Expand the length and syntax of some VMS names. Description: All VMS names such as node names and usernames should be of consistent length. They should allow up to 31 characters and certain special characters such as (. $ - % etc.) SIR: F83-026 Abstract: Provide a UNIX(tm) emulator. Description: Provide a UNIX(tm) emulator (like EUNICE) at the system service and CLI level, with file syntax translation and image activation. Less than 100% compatability would still be very desirable. SIR: F83-027 Abstract: Provide a mechanism to disallow DECnet access control strings. Description: There should be a mechanism that allows a site to disallow incoming or outgoing access control strings in DECnet VAX (on a node by node basis). This would require the use of proxy logins by those nodes for which access control strings are disallowed. This could prevent network users from getting into certain (perhaps privileged) accounts for which no proxy entry exists. SIR: F83-028 Abstract: Provide a prompting mechanism for DECnet passwords. Description: There should be an automatic mechanism by which password prompting would occur when a DECnet access control string is specified with only a username. At the minimum, the COPY command should provide this prompting capability. SIR: F83-029 Abstract: Provide a dial-back facility on VMS. 59 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List Description: VMS should provide a dial-back facility with a dial-back phone number field stored in the UAF. If that field is set when the user makes a dial-back request, it should be used to call the user back. If the phone number is not set for a given user, then the user should be allowed to specify a phone number on the dial-back request. SIR: F83-030 Abstract: SET HOST should be controllable on a node-by-node basis. Description: DECnet VAX should have a mechanism for disabling incoming network remote logins on a node-by-node basis. This would allow the system manager to prevent SET HOST from nodes that are not highly trusted. SIR: F83-031 Abstract: A minimum password length should be enforceable. Description: There should be a SYSGEN parameter that allows the system manager to set a minimum password length. No shorter passwords would be allowed. SIR: F83-032 Abstract: A system rollout should be provided to save the current state of all batch jobs in the system and allow them to be resumed on a subsequent reboot. Description: This rollout feature is desired so that the system manager or operator can shut down the system without stopping existing batch jobs for purposes of preventive maintenance or system testing. It should also be especially useful when changing temporarily to a more (or less) secure mode of operation. This is different than a user-callable checkpoint in that the state of all jobs would be saved without having to modify them. The environment of the proposed checkpoint would require that all disks and other hardware resources would be preserved intact until the system was restarted. SIR: F83-033 Abstract: There should be support for customer-definable privileges. Description: VMS should provide customer-definable privilege bits, perhaps via an additional privilege mask, so that a 60 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List system manager may define additional privileges for a site. Ths would provide a quick and easy means of insuring that users have privilege to execute shared images and programs which must have world execute access. SIR: F83-034 Abstract: Add more comprehensive traceback for login failures. Description: Add an option to turn on recording of login failures in more detail, including terminal name, username, attempted password, and time, even if the username is invalid. This would help to pin down noisy lines and attempted penetrations. The system manager would be responsible for protecting the account log to prevent disclosure of passwords. SIR: F83-035 Abstract: Provide for password expiration. Description: There should be a password expiration period and a password expiration date for each account in the UAF. The SET PASSWORD command should update the expiration date every time a user changes passwords. If a user logs in when a password has expired, the system should force the user to select a new password before completing the login. SIR: F83-036 Abstract: It should be possible to install command procedures with privilege. Description: It should be possible for the system manager to install a command procedure with privileges, so that users could perform operations via the command procedure that they would not ordinarily have privilege to perform. SIR: F83-037 Abstract: Provide an account expiration date. Description: There should be an expiration date for each account in the UAF. The user should not be allowed to log in after that date, until the expiration date is updated by the system manager. This allows temporary accounts to be created without having to constantly monitor them for expiration. 61 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List SIR: F83-038 Abstract: Enhance VMS's ability to be used as a host for developing programs for both RSX and the PRO systems. Description: At this time, it is difficult to use VMS to develop programs for RSX and the Professional systems. As soon as possible, all major PDP-11 languages available for RSX should be certified and licensed for use on VAX. An option for transferring existing licenses from RSX to VAX should be provided. The major VAX compilers which have RSX counterparts could be modified to have a /RSX switch to generate RSX compatible object code. This option could be bundled with the product, or licensed at extra cost. SIR: F83-039 Abstract: The LINK/RSX option should be converted to run in native mode. Description: The ability to do RSX program development on VMS is hampered by the slow running of RSX TKB. A native mode linker capable of generating RSX and Professional task images would speed up development work. SIR: F83-040 Abstract: The VMS RSX AME should support a full set of RSX program development utilities. Description: For VMS to be used effectively as a development system, a full set of RSX utilities must be supported. This includes supporting BRU under the AME so that files may be transferred via tape from RSX to VMS systems. SIR: F83-041 Abstract: The AME must support RMS V2.0. Description: RMS V2.0 is being supplied with RSX V4.1 It is essential that the VMS AME support RMS V2.0 (including DECnet support) to that the AME can be used for developing RSX/RMS tasks. SIR: F83-042 Abstract: Current users of the VMS AME should not have to pay for an AME license. Description: The VMS RSX AME is becoming a separate layered 62 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List product supported by the RSX group. Current users of the AME should not have to pay for an additional AME license. SIR: F83-043 Abstract: RSX SYSGEN's and NETGEN's must be better supported. Description: At present, RSX SYSGEN's are barely possible on VMS and NETGEN's are impossible. Both should be supported fully, as well as the installation of RSX layered products. SIR: F83-044 Abstract: An RSX SYSGEN/NETGEN should be possible using an ODS-1 format virtual disk on a ODS-2 format volume. Description: At this time, a requirement for performing an RSX SYSGEN or NETGEN is that one has a dedicated ODS-1 format disk. This is an unreasonable requirement given that VMS systems are increasingly using large capacity Winchester disks. For performing SYSGEN on a large ODS-2 disk, a virtual ODS-1 disk should be supplied as part of the AME. SIR: F83-045 Abstract: VMS should support an RSX Executive Directive subroutine mapping library. Description: One way to ease migration of RSX tasks written in a high level language to VMS is to provide a collection of native mode VMS subroutines which emulate the functions of the RSX Executive Subroutines. Such a mapping library of subroutines should either be supplied as part of VMS, the AME, or as an additional layered product. SIR: F83-046 Abstract: Allow CMS (Code Management System) to handle multi-line remarks. Description: Currently, CMS only allows a single line for remarks on library operations. Although the line can be long, there is still no way to format multi-line remarks nicely. For major changes to software, multi-line remarks are necessary to maintain adequate documentation. Perhaps CMS could prompt for remarks until an empty line is encountered. SIR: F83-047 63 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List Abstract: Allow VAX compilers to access elements in a CMS library. Description: The VAX compilers cannot access "include" modules which are part of a CMS library. This would be particularly useful for the VAX-11 C compiler which is (along with CMS) part of the "VNX" family of tools. SIR: F83-048 Abstract: SYSGEN should provide a DISCONNECT command. Description: When configuring devices with the CONNECT command, it would be useful to be able to reconfigure the device without rebooting the system. This would allow correcting the specification of CSR address or vector due to an error in the CONNECT or in the device switch settings. SIR: F83-049 Abstract: Additional peripherals are needed for laboratory applications. Description: There should be additional DEC supported peripherals for use in laboratory applications, particularly with the advent of inexpensive VAX's such as the 730. The current offering, the LPA-11K is too expensive and complex for many applications. SIR: F83-050 Abstract: Allow multiple key values for a single key of a given RMS record. Description: It should be possible to create several instances of a value for a single key of a given RMS record. Consider a field which represents the title of a book. It should be possible to generate several values for a given (primary or secondary) key to this record, one value for each word in the title. If available from RMS, this would facilitate layered products (e.g. DATATRIEVE) in performing lookups of a record "containing" a given value. SIR: F83-051 Abstract: Implement a flag to disable buffering of mailbox messages. Description: There should be a flag on mailbox creation which disallows queueing of data. The mailbox would contain only 64 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List a single data area which would be rewritten upon each write operation. This would provide a very easy mechanism for simulating real-time devices and would allow gross program errors to be discovered very cheaply. SIR: F83-052 Abstract: Provide additional flexibility in specifying disk quotas. Description: It should be possible to assign disk quotas on other than a user basis. It is sometimes necessary to limit the disk resources used by a project (group). Also, it is sometimes desireable to provide a limit on the number of blocks created in a given directory tree. This is useful when a [SCRATCH] directory must be restricted from consuming all of the free blocks on the disk. SIR: F83-053 Abstract: Allow VAX MAIL to notify the sender if mail was actually read. Description: There are times when it would be useful to know if mail sent to another user was actually read. Currently there is no way to verify receipt of a message. SIR: F83-054 Abstract: The VAX MAIL utility should incorporate store-and-forward capability. Description: As networks become larger and more complex, there is increasing chance that a remote node will be inaccessible when a mail message is to be sent there. VAX MAIL should provide a store-and-forward mechanism to insure that messages are eventually delivered. Although there is a DEC layered product with this capability (DECmail), it is not justifiable in a large, diverse (e.g. multiple university) network with primarily technical users. SIR: F83-055 Abstract: Allow BACKUP to be used to be used over DECnet. Description: It should be possible to use BACKUP over a network, so that files could be backed up to a tape drive on a remote node. This would be particularly useful for 11/730 systems where the cost of a tape drive is not justifiable. 65 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List SIR: F83-056 Abstract: Enhance the reporting capabilities of the ACCOUNTING utility. Description: The VMS ACCOUNTING utility should have simple report generator capabilities to allow titles and better formatting. It should also be possible to perform various levels of totaling, including resources used, total number of logins, etc., along with subtotals based on such characteristics as username, group, etc. Ideally, there should be a hook for specifying a charging algorithm so the report could directly specify dollar charges for the resources used. SIR: F83-057 Abstract: There should be additional support for account number fields. Description: The VMS account number field associated with a process should be expanded to 32 characters. There should be an account number field associated with a file, since the file ownership or access may still not determine who actually has to pay for the storage. There should be DCL commands and system calls for setting the account number of a file or process. The ANALYZE/DISK/USAGE should be able to report file account numbers. Process account numbers should also be propagated to subprocesses and batch jobs. SIR: F83-058 Abstract: Provide supported access to the DCL tables in source form. Description: System managers need access to the DCL command tables to change command defaults or to remove commands. This is particularly important in a hostile environment when DCL symbols are easily changed or bypassed. However, there is currently no supported way to obtain a current command definition (except from the microfiche). The DCL tables should be provided as a text library, or a utility to extract these definitions should be supported. SIR: F83-059 Abstract: Expand UAF entries to include default information for print jobs. Description: It would be useful to be able to specify default print attributes on a user by user basis. This would 66 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List include such things as default print queue, header page attributes, maximum page count, etc. SIR: F83-060 Abstract: Provide for selective image accounting. Description: It is often useful to perform accounting on an image basis for a few selected images, for example to surcharge for the use of a particular program. VMS should allow an image to be "installed" for image accounting, avoiding the overhead of performing accounting for all images. SIR: F83-061 Abstract: Provide the ability to control the percentage of CPU devoted to batch. Description: VMS should have the ability to control the CPU consumption for a class of jobs (interactive, batch, network, etc) to a particular percentage of the CPU time over a selectable interval. This would provide for fairer response to batch jobs in a mixed workload. SIR: F83-062 Abstract: Allow images to be installed with elevated priority. Description: It is sometimes desirable to allow images to execute at higher priority when responsiveness must be emphasized. It is preferable to have an external mechanism for specifying this priority to facilitate tuning it, rather than having the image alter its priority. SIR: F83-063 Abstract: Provide additional support for operator tape handling. Description: It should be possible to request mounting of a tape without specifying a device name. The system should select the tape drive (with operator override). There should be operator support for INITIALIZE. All relevant information about the mount, including selected density, write/nowrite status, and visual-id's should be passed to the operator on the initial mount request, and on all requests for continuation volumes in a multi-volume set. 67 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request List SIR: F83-064 Abstract: DCL should provide more commands for foreign tape handling. Description: There should be DCL commands to support foreign tape handling. It should be possible to skip forward and backward by file (and even by block). It would be useful to be able to specify the blocksize without dismounting the tape. It should also be possible to write a tape mark on the tape from DCL. SIR: F83-065 Abstract: Provide access to files on labeled tapes by file number. Description: It should be possible to specify the file on a standard labeled tape by using the ordinal number of the file. This would be particularly useful in the case where multiple files with the same name exist on the tape. 68 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request Ballot Fall 1983 System Improvement Request Ballot Questionnaire DECUS membership number __________________ Installation Delegate? YES NO Your experience level (circle one): 1. Wizard 2. Expert 3. Knowlegable 4. Normal 5. Novice ------------------------------------------------------------------ Configuration: Indicate how many of each CPU type you have: 1. 11/782 ______ 2. 11/780 ______ 3. 11/750 ______ 4. 11/730 ______ For your primary system indicate: Memory size (Mb.)_________ Average user load _________ Total disk space (Mb.) ________ VMS Version No._______ Operating System (if not VMS) or Operating System migrating from _______________________ ----------------------------------------------------------------- Application Area: (Circle all that apply) 1. End User 2. H'ware OEM 3. S'ware OEM 4. Sys Integrator 5. Engineering 6. Research 7. S'ware house 8. Service Bureau 9. Scientific 10. Business EDP 11. Government 12. Utility Co. 13. Turnkey System 14. Manufacturing 15. Education Principle Programming Language _________________________ 69 PAGESWAPPER - July 1983 - Volume 5 Number 1 Fall 1983 System Improvement Request Ballot VAX SYSTEMS SIG FALL 1983 SIR BALLOT Tally Reminder: The total number of points (absolute value) which you allocate on this ballot may not exceed 100 points. No more than 10 points may be given to any single SIR. SIR Number: Points: ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ ___________________ __________________ Mail to: Mr. Gary Grebus Battelle Columbus Laboratories Room 11-6011 505 King Avenue Columbus, Ohio 43201 70 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT Submission Form INPUT/OUTPUT Submission Form A SIG Information Interchange Please reprint in the next issue of the Pageswapper Caption: ______________________________________________________ Message: ______________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ Contact: Name _______________________________________________________ Address ____________________________________________________ ____________________________________________________________ Telephone ____________________________ If this is a reply to a previous I/O, which number? ________ Signature _____________________________ Date ________________ Mail this form to: PAGESWAPPER Editor, DECUS, MRO2-1/C11, One Iron Way, Marlborough, MA 01752, USA 71 PAGESWAPPER - July 1983 - Volume 5 Number 1 INPUT/OUTPUT Submission Form Tear out to submit an I/O item PAGESWAPPER Editor DECUS, MRO2-1/C11 One Iron Way Marlborough, MA 01752 USA 72 PAGESWAPPER - July 1983 - Volume 5 Number 1 System Improvement Request Submission Form System Improvement Request Submission Form SIG ref no. _________ Page 1 of _____ ________________________________________________________________ Submittor: Firm: Address: Phone: ________________________________________________________________ Circle application area(s) most closely related to yours (OEMs circle end use): Transaction Processing Business EDP (accounting) Program Development Systems Development General Timesharing Student Timesharing Shared Small Applications Shared Large Applications Process Control Word Processing Large Simulation ________________________________________________________________ System Configuration: CPU Model: System Disk: Memory Size: Average User Load: Operating System: Version: ________________________________________________________________ Abstract (Please limit to four lines): ________________________________________________________________ Description (include justification and expected usefulness): Use additional pages if required Completed SIR should be returned to: Gary L. Grebus, Battelle Columbus Laboratories, 505 King Avenue, Columbus, Ohio 43201, USA 73 PAGESWAPPER - July 1983 - Volume 5 Number 1 System Improvement Request Submission Form Tear out to submit an SIR Gary L. Grebus Battelle Columbus Laboratories 505 King Avenue Columbus, Ohio 43201 USA 74