The Summer seems to have disappeared and here we are ready to face the storm after the lull. As usual unpredictable seasonable variations have distorted the use of the facilities. In particular the IBM system has delivered anything from 160 to 350 CPU hours per week during the period.
Machine performance has been generally good across all facilities with no really bad situation? arising.
The need to change the image of FORUM is highlighted in this issue with an article summarising the views of User Support Group on how this should be achieved. I would like to make a personal plea to any of our readers who wish to contribute to send their comments and views to me.
Readers will also discover in this issue that one can still apparently get something for nothing. The zero-cost upgrade of the IBM 3081D will probably be as baffling to you as it was to us. However it almost certainly will happen and we eagerly await the outcome. Unfortunately we are not aware of any such possibilities involving DEC, GEC or PRIME.
The move to Rev 19.1 of the PRIME Operating System has not been as smooth as we would have liked. The present situation is summarised in this issue, and to the best of our knowledge we have overcome the worst of the difficulties. Any users who are still in trouble should contact RAL.
| 17 Oct | Central Computing Committee meeting |
| 19-20 Oct | ELECTRIC/CMS Conversion Course |
| 24 - 27 Oct | IBM New Users Course |
To understand how you can get something for nothing, consider the following parable:
Once upon a time, a company sold threshing machines to farmers all round the world. As it was a large company, it only updated its price of threshing machines at intervals. Consequently, if there were wild fluctuations of exchange rate between countries, it was possible for the price of threshing machines to vary significantly in different countries. Normally the company managed to review the prices in time, but it was possible to buy a threshing machine in one country and sell it in another making a modest profit. However, the average farmer was unable to understand the complexities of high finance sufficiently to take advantage of such variations, particularly as he had no experience of selling threshing machines abroad.
The situation was further complicated by the company's benevolent attitude to agricultural colleges. It gave such establishments a discounted price. Some people said this was because the company wanted to make sure that new farmers were aware of the possibilities opened up by the use of threshing machines, while others rather maliciously said it was to make sure that all young farmers coming out of college had only used that company's threshing machines and so would continue to buy them in the future.
Be that as it may, agricultural colleges could buy-threshing machines cheaper than young farmers could. To stop colleges selling the machines to farmers, the company insisted that a college paid back the discount if it resold it, taking into consideration the age of the threshing machine.
Of course, if the amount for which you could sell the threshing machine to a farmer (less the discount that you paid back) was the same as the discounted cost of a new threshing machine, the agricultural college could sell its old machine and buy a new one at no cost.
I assume you have followed me so far. Now the real situation was nowhere near as simple at that! Depending on the time of year, the number of threshing machines in stock and how philanthropic the company was feeling, the discount given to agricultural colleges varied. For example, you could buy a machine one day at one price and pay more or less a year later.
Upon this scene came a veritable financial wizard who knew farmers in many countries. By buying second-hand threshing machines from agricultural colleges in one country and selling them to farmers in another it was possible for the wizard to give the college more money than it needed to replace the threshing machine it had and still make a profit for the wizard. As threshing machines were not always easy to come by, the wizard found this relatively easy to do.
You may ask who was the loser in this parable. The answer is nobody. The wizard made a profit. The college got a new machine. The company sold another threshing machine. It is a pity that all life is not like this!
In October, the current 16 Mbyte 3081D is being replaced by a 32 Mbyte 3081D at no cost to SERC.
Computing and Administration Divisions are jointly establishing a six-month pilot project in office automation embracing a limited set of office workers on the RAL site. The system under evaluation in the pilot is IBM's PROFS package; PROFS being an acronym for PRofessional OFfice System. It is one of IBM's strategic products which grew out of a joint venture between IBM and AMOCO Labs in the USA. It is widely used both within IBM and by its customer base, and is a mature development having been through about ten major releases.
PROFS consists of a set of VM/CMS - based application programs designed to support a wide range of office activities via full screen terminals. In machine terms it requires a configuration equivalent to an IBM 4331 or larger, a minimum of 1 MByte of main storage, and discs for which there is VM/SP support. Networking to terminals, printers etc. uses VNET. As originally conceived, PROFS had a command driven interface but this has been replaced by a menu driven system which makes extensive use of the PF keys available on 327x terminals. Its origins do, however, mean that PROFS can be used on line-mode devices, though it is less comfortable to work in this environment.
A major feature of PROFS is its relationship to other well-known IBM products. Since it uses the standard CMS filestore for storage of information, many housekeeping functions and CMS utilities are available to the PROFS user. For text production, it uses IBM's Document Composition Facility (DCF) which grew out of SCRIPT. Among the features available are alignment, pagination, multi-column layout, various styles of headings and lists, indexing, several macros and spelling verification. PROFS overall design is sufficiently flexible to cater for a broad spectrum of potential users. At one extreme someone with little or no knowledge or computer systems can use it very effectively and be completely unaware of the substructure on which the system depends. However, an expert user is free to exploit his knowledge of CMS and DCF; for instance, the full power of the CMS Editor is available to him as well as any DCF control words he may care to use for text preparation. The office activities on which PROFS will be brought tc bear during the RAL experiment include:
The pilot project is nearly ready to start. The installation of wiring and terminals is now well advanced and the PROFS package has been running for some time on the 3081/Atlas 10 complex. About 70 users are participating in the experiment drawn mainly from General Administration, Finance and Personnel Sections, plus selected Heads of Division, secretaries and typists. Although it is possible to use a range of terminals with PROFS from standard VDUs to Displaywriters (IBM Word Processors), the pilot will concentrate on using IBM 3178 Model C2 full screen terminals. Approximately 30 are being installed together with matrix and letter-quality printers to give an overall terminal printer ratio of about 4:1. The IBM Personal Computer will be evaluated as a managerial workstation and some standard VDUs will be tried. As the pilot progresses, further decisions will be taken on its possible expansion.
It has been said in the past that FORUM is too:
A project is now underway in User Support Group to update FORUM and attempt to redress some of these criticisms.
We have spent some time analysing the problems with FORUM, and have a series of proposals which we intend to implement. We have already made some of the smaller changes; May's issue included the Table of Contents on the front page, and the first issue with an Editorial column was in June. Also, in an attempt to cover a wider area of interest, we have tried to include more news items without a bias towards the mainframe computers. Major changes will be introduced in the first issue of 1984.
First, we tried to define the function of FORUM. As the RAL Computing Division's Newsletter, its main function should be to provide an information service to all our computer users.
It should be interesting as well as Informative, and should include articles from a variety of sources over a wide range of topics.
Improve FORUM's appearance. This includes:
As from January 1984, FORUM's name will be modified annually to include the year. Hence next year, FORUM will be called FORUM 84.
The front page will be redesigned, and will contain only the Table of Contents. The inside front cover will include the Editorial, the names of those on the editorial board, dates for the deadlines of future editions, and a preview of special forthcoming articles.
We will investigate how to make the headings more prominent and the possibility of using different size characters, and try to introduce more diagrams and photographs to illustrate the text.
However, in improving the layout, we are restricted by the facilities available; we want the process to be as automatic as possible, to keep "pasting-up" time to a minimum, for example.
Include more user contributions.
This will be done in two ways; articles from users on their experiences in the computing environment will be encouraged and we will also have a "Letters to the Editor" section. To start this off, if you have any specific suggestions for changes to FORUM, or feel strongly (one way or another) about any proposal in this article, please mail your comments to FORUM@RAL. If you do not have access to SERCNET, then letters may be sent to The Editor of FORUM, R27, RAL, Chilton, Didcot, Oxon, 0X11 OQX.
Each December issue will include an index of all relevant articles over the past year.
We will choose a "theme" for two or three FORUMs during each year; these will be publicised, and articles relating to that theme will be requested. Suggestions for themes include Graphics, a specific computer range (eg GECs), HELP systems, and the Network.
We would also like to have a regular section, covering all the different machine ranges, where a small number of topical user queries and answers are printed.
Users may also find a list of "who to contact at RAL for what, and how to contact them" useful. Regular inclusion of this list is being considered. If you have any comments on any of the proposed changes to FORUM, please mail them to: FORUM@RAL
Revision 19 of Primes and its utilities have now been installed on all but three SERC Primes (which are scheduled for September). In general, the conversion has gone well, particularly at the University sites, as problems with the conversion procedure were ironed out at Rutherford. I am sure that most Prime users welcome the new facilities and security provided in Revision 19.
RLPE was the first service machine to acquire Revision 19, but after several days of service a spate of machine halts was experienced, and with the help of a magnetic tape crash dump a bug was quickly identified and cured. A problem relating to the new Prime quota system was also fixed. Please note that a problem still exists whereby batch jobs are not allowed to exceed a disk quota, and a solution to this is being sought. Apologies are due to owners of one or two UFDs mistakenly ACL'd, due to a bug in the conversion procedure which was resolved before subsequent conversion.
Prior to the conversion to Revision 19, a disaster occurred on RLPF as the backup of one of the disks was performed the wrong way, overwriting the previous backup. Apart from this, all the conversions went smoothly, although the following problems were experienced.
The Salford FTN77 compiler was found not to be able to compile some of the latest Prime supplied INSERT files, so specially converted ones had to be installed at all sites. Also a Prime F77 bug at revisions 18.4 and 19.1 of the compiler was discovered, although only one user on RLPH was affected. F77 18.3 is therefore still running on RLPF. On RLPE, what appears to be a bug in IRSTATUS (used by the Library database) was unearthed by a new version of LOGIN. A temporary solution to this has beer, implemented, and users are no longer affected.
The new Prime disk quota system is causing problems for some users, as it enforces disk controls more rigidly than the old SERC system, which unfortunately was easily abused. Users who get into disk space difficulties should contact Resource Management at RAL.
The new file protection mechanism, ACLs, has also caused several problems for users, who are again urged to read all the NEWS files concerning Revision 19 and to check their back copies of Prime Support News which document the new features of Revision 19. The main difficulty is that the default Access Control Lists set up often prevented users from accessing other users' programs and it has been necessary to adjust the ACLs to allow such access.
It is still possible that difficulties may be encountered with little used commands or utilities, and users are requested to report such difficulties immediately to User Support.
The formative years of the Interactive Computing Facility were closely linked with the development of the SERC network. Encouraged by the team at RAL, and much aided by rapidly improving network communications, a valuable community spirit soon appeared amongst the staff of the GEC sites. File transfer and remote login facilities meant that it was easy to share software, and the idea of a Program Library was formed to reduce unnecessary duplication of effort.
We decided at an early stage not to impose rules or standards: instead, we would keep a list of sharable items, and the list would indicate the likely level of support and documentation for each item. The validity of this approach is shown by the development of PREDICT, designed by a spendthrift user to give warning if he was likely to exhaust his allocation of AUs before the end of the accounting period. Its first form was a command file which invoked a system accounts utility, whose output it then edited and processed to provide meaningful figures. Today, PREDICT is an efficient single program written at another site, but the user interface is unchanged from the original user's design.
Other 'convenience' items represent greater effort by the original author, and are still in use largely unchanged. In the field of graphics, for instance, there are utilities for the production of A4-size pages of histograms, block diagrams, and elegantly lettered overhead projector transparencies, all without demanding that the user understands the subroutine libraries used or compiles any code. One enterprising user, tired of producing test copies of plots destined for his thesis, wrote a substantial program to read the plotfiles and display on a screen exactly what the plot would look like. GVIEW is now a valued utility, providing extensive previewing and conversion facilities for many graphics devices.
The design of GVIEW was aided by experience of a similar utility on the local Computing Service machine. In other cases, complete packages have been moved to the GEC systems, where they can now reach a far larger user population than before. Wider use means more suggestions for enhancements -and more rapid discovery of bugs! CLADP, originally the 'workbench' of a single Control Engineering research group, is now used by groups at several other Universities. Disciplines other than Engineering have installed compatible GEC machines: the library includes items inspired by High Energy Physics facilities at CERN, and items from the fields of Organic Chemistry and Crystallography. Some such items have found use outside their originating disciplines.
As ideas are shared and developed, the library begins to offer the kind of in-depth software coverage that was previously the province of certain imported computers. There are, for instance computer-aided learning packages, desk-calculator programs that operate on vectors and matrices, program development tools such as preprocessors and cross-referencers, microprocessor development aids, and a growing list of language compilers.
Many listed items are freely available to be used or modified at will; others are 'for academic use only', and still others are licensed products for which payment is required. All are accepted for inclusion in the lists (there is no 'editorial policy') and wherever possible the availability is indicated.
Every GEC Site Manager has access to the lists, can list existing programs, and can make contributions or circulate requests for items not (yet) listed.
A new version of the Harwell library has been installed on MVT and CMS.
In MVT the new library is held in SYS1.HARLIB. The old library has been renamed as SYS1.CHARLIB and will be deleted in two months time if there are no problems with the new library.
In CMS the library is held in HARLIB1 TXTLIB R and HARLIB2 TXTLIB R and can be accessed by the GLOBAL command:
GLOBAL TXTLIB FORTLIB HARLIB1 HARLIB2
The routines OB01A, OB02A, OB11A, OB12A, OB13A, OB14A, ZA04AS and ZA16AS are not included in the new MVT library, but this should not affect any users. No other routines have been withdrawn from the library.
The CMS library contains all the routines available in MVT except for ZA02AS, ZA06AS, ZA12AS, ZA16AS, ZA17AS, ZA18AS, ZR02AS, ZR03AS and ZV01AD.
There is no online HELP information for individual routines in the Harwell library.
The RAL User Liaison Committee approved the setting up of such a group at its last meeting. The chairman will be Dr Gareth Hughes (Physics Department, Lancaster), the Deputy Chairman will be Dr Richard Ansorge (Cavendish Laboratory, Cambridge) and Mrs Ros Hallowell (Computing Division, RAL) will be the Secretary.
Representatives have been nominated from all SERC provided VAX(VMS) systems and are as follows
| University of CAMBRIDGE , High Energy Physics | Dr R E Ansorge |
| IMPERIAL COLLEGE , Blackett Laboratory | Mr Paul Baker |
| IMPERIAL COLLEGE , Atmospheric Physics Group | Mrs Lesley Grove |
| >University of LANCASTER , Department of Physics | Dr Gareth Hughes |
| University of LEEDS, Dept of Mechanical Engineering | Systems Manager , (to be appointed) |
| University of LONDON , Queen Elizabeth College | Mr Nigel Arnot |
| University of LONDON , Queen Mary College Department of Chemistry | Dr Nigel Walker |
| University of LONDON , Queen Mary College | Mr Paul Kyberg |
| University of MANCHESTER, Dept of Mechanical Engineering | Mr Vale |
| University NEWCASTLE UPON TYNE , Dept of Electrical & Electronic Eng | Mr G F Mole |
| University of OXFORD , Chemical Crystallography Lab | Dr K Prout |
| University of OXFORD HEP | Dr J B Macallister |
| University of OXFORD , Nuclear Physics Laboratory | Dr D Sinclair |
| University of OXFORD | Dr J Cox |
| SERC | |
| CERN | Dr C N P Gee |
| Daresbury Laboratory, SRS | Miss Frances Rake |
| HEP | Mr M Waters |
| Starlink | Dr D L Terrett |
The initial meeting is scheduled for Monday 19 September 1983. Users wishing to raise any issues at future meetings are invited to contact their local representative or any of the people mentioned above.
A short report of these meetings will be published in FORUM.
I have been asked to write a paper for the User Requirements Committee, stating the requirements of HELP systems and documentation. The following is a resume of what I see as the users' requirements - have I got it right? Please mail me a message (J.J.C.Hutchinson@RAL) if you disagree (or agree) with any point, or feel that I have omitted anything important.
THESE MUST BE UP-TO-DATE.
This is the most important feature of any HELP system. Manuals may get out of date, but HELP should always contain the current information. HELP systems should be easy to use, and should be self-explanatory; no-one should need to LEARN how to use HELP.
They should be keyword driven, with a certain amount of fuzzy-matching to cater for spelling mistakes; this would also introduce a certain amount of that essential element called 'user-friendliness' .
There must be a large number of synonyms included; Delete, Discard, Destroy, Erase, Expunge, Purge and Remove (there may be more) are all possible keywords to delete files, mail messages, text in an editor, jobs in batch, files in spool queues or files awaiting transfer. The default system should assume the user is new to the system, and should give a menu choice to the user when there is any possibility of confusion (as when one of the keywords above is typed). However, a user who knows what he is looking for and knows the required keyword should be able to override this option easily. This should be possible when calling HELP and it should be easy to switch from 'novice' mode once in HELP.
There is a problem of what HELP systems should assume at certain points. For example, in a hierarchical HELP system, if a user has just been browsing in the MAIL sub-system of HELP, and then types REMOVE, should the system assume that the user wants information to remove a mail message, or should it give the user a menu choice? Menu systems get very tedious when they are used to excess (and when the user knows what he wants).
HELP should be a memory jogger, it should be easy to produce brief information on a wide variety of topics in manageable quantities. It. should not contain screeds of information, although controlled access to manuals can be useful. (Well organised access to manuals on-line could reduce, but will never remove, the need for paper manuals; users will always want to refer to manuals when not at terminals.) It is also easier to read from paper than from a screen.
HELP should be easily accessed from sub-systems, and on leaving HELP should return to the sub-system .
Each document should be produced for a clearly defined audience, whether it is a User Guide for those familiar with computing but relatively new to a specific system, a Reference Manual for those familiar with the system, or a user note to inform of changes. It should also be made clear in each document if (and how) the document is to be updated, and if the document is not to be updated, where new versions are advertised and whether they are available on request.
Text should be easy to read, and descriptions should be clear. Jargon and long tedious descriptions should be avoided, users do not enjoy reading manuals, and information should be provided with as little (reading) effort as possible. All manuals should contain examples. Manuals should also be as professional as possible; a well produced manual instils confidence in the user and is worth the effort.
A documentation standard is being defined at RAL; when this is produced it is hoped that all manuals produced at RAL will reach the same high standard.
At present, we produce Primers or Introduction manuals for novice users, and a new user is told to read the first 'n' chapters before using the system. This does not happen; users get bored reading, log in, and then fumble round the system, learning by trial and error. All systems should provide a tutorial session to new users; they should be given a sheet of paper giving their username, how to log in, and how to access the tutorial. This should then take the user through basic information on local editing controls, the command syntax, the filestore and basic editing instructions. The user should then be in a position to use the User Guide.
Each system should have syntax cards; these have proved very popular and useful on systems where they are provided. They should serve a similar 'memory-jogger' service to HELP, with the advantage that they can be carried round and used when not at a terminal.
The Finite Element Library was conceived by Dr. Ian Smith of the University of Manchester. It is designed as a development tool for those people wishing to experiment with the practicalities of solving differential equations. It enables them to solve both steady state and time dependent equations with up to 3 spatial dimensions. It addresses problems (such as those arising in fluid flow) not covered by the large standard packages and provide a valuable teaching tool for those interested in the mathematics of the finite element method.
Use of the library is by no means restricted to the solution of small problems. Problems involving many thousands of unknowns may be tackled despite the lack of out of core linear algebra routines and methods such as nested dissection.
London University NAG Panel has arranged for Dr. Ian Smith to conduct a half-day workshop on the use of the library for teaching and research purposes on 19 October 1983 from 2.00pm until 4.15pm at
Imperial College, Lecture Theatre Room 220,Mechanical Engineering Building, Exhibition Road, London SW6
If you wish to attend the workshop please contact
Mr S Budd,
Imperial College Computer Centre,
London SW7 2BX
or telephone (01-)589-511 ext 1197
Readers may be aware that CMS is considerably more productive when used from full-screen terminals. Unfortunately, until now IBM 3270-type terminals have only been available to a few local users at RAL. The reasons for this are that 3270-type terminals are expensive, cannot be used over SERCNET and can only be used on the IBM system.
However, an upgrade to the standard Cifer terminal, which has been developed in collaboration with Cifer Systems Ltd, will solve all of these problems.
The upgrade allows the terminal to be used as a 3270-type terminal whilst retaining the ability to be used as a standard ASCII terminal or graphics terminal (if the graphics option is fitted). The upgrade is cheap (£290 for a model 2605, £125 for a model 2634, while a new 2634 with the 3270 modification costs £915, which is little more than the cost of a basic 2634). Most importantly, the terminal can be used in full-screen mode over SERCNET.
Cifer models 2605 and 2634 can be upgraded to full-screen terminals, though a model 2605 terminal has to be first converted to a model 2634 (this is part of the upgrade). The upgrade involves changes to the internal electronics and modifications to the keyboard.
We make no pretensions that the upgraded terminal is as good as a real 3270, but it does support all of the important 3270 functions and is a vast improvement on a dumb terminal.
A field trial of the terminal is now nearing completion and the upgrade will be available shortly. A User Note giving details of the use of terminal is in preparation and will be available on request.
The removal of the CDC 1700 at Imperial College at the end of August resulted in the termination of the HASP connection to SERCNET from the Imperial College Computer Centre. Any user experiencing difficulties with access to SERCNET as a consequence of this should contact the PAC (ext 6111) or me (ext 5408).
The degree to which CMS can be considered a user-friendly system has frequently been discussed by users. Discussions at the Central Computer Site Users Meeting culminated in a paper by J C Hart giving a user's point of view. A reply to this paper was prepared by D M Asbury. We would like to widen the discussion to involve all CMS users. Hard copies of the papers can be obtained by using the commands:
NEWS FRIENDLY LISTING (VPRINT NEWS REPLY LISTING (VPRINT
If, having read these documents, you want to contribute to the discussion, send your views to me either as a note or a file. My ID on CMS is RM.
