Jump Over Left Menu
Informatics Annual Report 1985-6
- 1. INTRODUCTION
- 2. SOFTWARE ENGINEERING
- 3. IKBS
- 4. MMI
- 4.1 Introduction
- 4.2 Window Management
- 4.3 WW - Graphical Toolkit
- 4.4 PostScript Interpreter
- 4.5 Coordination
- 4.6 Common Base Programme
- 4.7 SPP in the Foundations of the Design of Interactive Systems
- 4.8 Research Proposals
- 4.9 Other Developments
- 4.10 Future Developments
- 4.11 Publications
- 4.12 Conferences
- 5. DISTRIBUTED INTERACTIVE COMPUTING - ALVEY INFRASTRUCTURE
- 6. DISTRIBUTED INTERACTIVE COMPUTING - ENGINEERING COMPUTING FACILITIES
- 7. DIVISIONAL INFRASTRUCTURE
The Annual Report covers the year ending September 1986. Its main purpose is for internal use in the Division indicating what has been achieved and by whom. Hopefully it gives new members of the Division some idea of the work programme and its objectives.
Informatics Division was formed in July 1984 when the single Computing Division was split into two parts with Central Computing Division being responsible for the mainframe computer services and wide area networking. The responsibility of the new Informatics Division was to support SERC's Engineering Board's activities in the information systems area, namely:
- Interactive Computing Facility: to continue the provision of interactive facilities for the Engineering Board users via PRIME and GEC multi user minis. The facility started to provide a service in 1976 and is the most mature of the Division's projects.
Single User System Common Base Programme: to continue the provision of a common
hardware and software base for Engineering Board users initially with PERQ1 systems but more
recently with PERQ2, SUN2 and SUN3 systems. This project started to provide a service in 1982
although work on single user systems started in 1979 as part of the ICF project above and in
support of the Distributed Computing Systems programme which was being coordinated at that time.
Alvey IKBS, SE and MMI Support: to provide coordination, support and some research and development activities in these three areas of the Alvey programme.
The Alvey Programme started in 1983 and was aimed at providing a significant impetus to cooperative pre-competitive research in the enabling and underlying technologies of Information Technology. Funding is provided by DTI, MoD and SERC via the Engineering Board.
- Alvey Infrastructure: to provide and develop the infrastructure, both hardware and software, for the IKBS, SE and MMI areas. This started in 1984 providing multi-user support via a distributed set of GEC Series 63 and SYSTIME VAX systems. More recently support has also been provided for SUN2 and SUN3 systems.
The major change in funding this year is in the non-Alvey area where the decision has been made by the Engineering Board to merge the ICF and SUS programmes into a single programme called the Engineering Computing Facility (ECF). The motivation for this is to recognise the trend towards single user systems and the need to provide relevant server facilities for clusters of single user systems. It was felt that this could best be done by a single project having less rigid constraints between the two activities.
The internal organisation of the Division has remained static during the year, comprising three Groups:
- Software Engineering R W Witty (deputy D A Duce)
- IKBS/MMI C J Pavelin (deputy K Robinson)
- Distributed Interactive Computing K F Hartley (deputy M R Jane)
Ken Hartley joined the Division near the start of the year as head of DIC Group. Appendix A gives the internal Divisional Structure with people in post in September 1985 and September 1986. Rob Witty was Deputy Director of the Alvey Software Engineering programme for most of the year and took over as Director in August. Both functions have entailed considerable interaction and time spent at Alvey Headquarters in Millbank, London.
The format of the report corresponds quite closely with the main project structure and the associated Group structure.
For those unaware of the prevailing jargon in the Division:
|ACARD||Advisory Council for Applied Research and Development|
|ALFEX||Alvey Financial Expert Systems Club|
|Alvey||Alvey Programme funded by DTI, MoD and SERC|
|ASPECT||Alvey-sponsored 2nd generation IPSE|
|CFC||Computing Facilities Committee of EB|
|CO||SERC Central Office at Swindon|
|CSI||Client Server Interface|
|CSSC||Computer Science Sub Committee of IEC|
|DCS||Distributed Computing Systems SPP|
|DTI||Department of Trade and Industry|
|EB||Engineering Board of SERC|
|ECF||Engineering Computing Facility (funded by CFC)|
|ECLIPSE||Alvey-sponsored 2nd generation IPSE|
|ERCC||Edinburgh Regional Computer Centre|
|FLAGSHIP||Functional, Logic And General Software-Hardware Implementation Project. Large Alvey project including ICL, P1essey, University of Manchester and Imperial College, London.|
|GRE||Government Research Establishments|
|ICF||Interactive Computing Facility (part of ECF)|
|IEC||Information Engineering Committee of EB|
|IKBS||Intelligent Knowledge Based Systems|
|IPR||Intellectual Property Rights (the rights to market the results of a particular piece of research)|
|IPSE||Integrated Project Support Environment|
|IPSE 2.5||Alvey-sponsored 2nd generation IPSE|
|Millbank||Location of Alvey Directorate in Central London|
|MMI||Man Machine Interface|
|MoD||Ministry of Defence|
|NFS||SUN's Network File System|
|NRS||Name Registration Scheme|
|RFS||AT&T's Remote File System|
|SERC||Science and Engineering Research Council|
|SIGAI||Special Interest Group for Artificial Intelligence (originally an ICF group but now part of Alvey IKBS)|
|SPP||Specially Promoted Programme of SERC|
|STI||Software Technology Initiative - SPP that became part of the Alvey SE programme|
|SUS||Single User System|
|SVF||Senior Visiting Fellow|
|Swindon||Another name for Central Office|
|Uncle||An individual who oversees a non-collaborative academic research project in Alvey to ensure awareness of industrial relevance both into and out of the project.|
|VLSI||Very Large Scale Integration|
2. SOFTWARE ENGINEERING
2.1 The Role of the Software Engineering Group
SEG's role is four-fold:
- Alvey SE Programme management, ie the Millbank function, with special responsibility for the academic SE Component, SERC/RAL finance and liaison with SERC Central Office at Swindon.
- Technical Support to the Alvey SE Programme. This involves advice to the Millbank staff on technical issues (eg Wadsworth's briefing to Oakley on Japan), compilation and distribution of the SE Mailshots, refereeing project applications, information gathering (eg cellular radio, 3270 emulation).
- Development work on behalf of the Alvey SE programme. This can be on behalf of specific projects (eg BSD 4.2 Ring driver for Newcastle/Aspect) or for the Alvey SE infrastructure generally (eg ML/LCF on UTS, Newcastle Connection). Some Development work of the specific project variety will be on behalf of SEG' s own research projects (see below).
- SE research, in line with Alvey SE strategy, IEC/CSC policy and in collaboration with industry, GREs and Universities.
Management, Technical Support and Development work are funded directly from Millbank to SERC/RAL. They attract 75% DTI funding because these activities are services on behalf of both academic and industrial communities.
The research work is funded differently. This must either follow normal Alvey rules, ie a complete, collaborative project must be constructed, approved and awarded or go through the peer review system of IEC/CSC. At any given instant SEG aims to have more than one project under way with successors in the pipeline. SEG aims to have both Alvey and CSC funding simultaneously because some work will be more suited to CSC.
2.2 Alvey SE Programme Management
2.2.1 SEG's Contribution to Alvey Management
The software engineering component of the Alvey Programme is a significant part of the total, amounting to about Â£65M, of which about Â£16M is for the academic part of the SE programme. An existing portfolio of academic research grants in SE (the SERC's Software Technology Initiative) has been included in the Alvey Programme to ensure continuity and involvement of all relevant SE research in the UK. The Alvey SE programme currently has 87 projects; 32 are fully collaborative and 55 are smaller uncle projects and SVFs etc.
Support for the Alvey Programme consists of interaction with academic groups to assist in the preparation of grant applications and coordination of the SERC/Alvey Directorate interface.
Research supported by Alvey differs from that usually supported by SERC in that the Alvey Programme is a directed programme. This involves much closer involvement in both the preparation of the grant applications and especially during the execution stage of the research. For this purpose all projects are monitored, both technically and financially. This involves preparation of a database on all projects, collation of reports, generation of summaries and spend profiles. Regular reports are required from Monitoring Officers for collaborative projects and from industrial Uncles for academic only projects. Any exceptions raised by this monitoring process are actioned quickly.
Support is provided also for the Alvey SE Office in Millbank by providing assistance in meetings, workshops, panels and presentations. Much effort is expended in maintaining the files and reporting procedures for the programme. To assist in maintaining continuity of research, a mailshot service is operated. An increasingly important aspect of the overall support as the programme develops is the analysis of the budget and preparation of data for forward planning. For this purpose a special procedure has been developed with the objective of minimising difficulties associated with over commitments of the budget.
The RAL Software Engineering Group has continued to provide management support to the Alvey Programme. Since the major part of the programme is now in place attention has concentrated on monitoring the individual projects and installing a procedure for disseminating the information generated by the programme.
General support is provided for coordination of industrial and academic proposals in the area of software engineering. In addition to processing new proposals to the directorate the directed nature of the programme results in necessary changes to existing Alvey projects. These changes result from the dynamic nature of the collaborative projects, reflecting changing industrial positions and the evolving nature of the research to ensure that the products continue to have industrial relevance.
The monitoring programme now is well established and is generating vital information on the progress of all SE projects. Especially important is the rapidity with which problems are identified in particular projects. Remedial action then can be taken or planned to minimise disruption to the main R&D thrust of those projects. The reports produced by the monitoring officers, and Uncles for non-collaborative academic projects, are used regularly for overall programme reviews. These reports form the basis for detailed assessments by the directorate staff in the various speciality topics within SE.
An essential part of the Alvey programme is the dissemination of information generated by the many research projects. These results are presented in the form of discrete deliverables for each project. To meet this part of the Alvey programme a procedure for disseminating information about the SE deliverables has been established. As the deliverables become available abstracts are provided, edited and collated for publication as a supplement to the bi-monthly Alvey News, thereby reaching 7000 potentially interested parties. Further information about the deliverables can be obtained from the relevant project manager or owner of the IPR.
Other general support activity includes continuation of the SE Mailshot service and generation of numerous reference data relating to the SE programme. Lists of participants in the programme giving their electronic mail addresses and current interests are maintained. Also, detailed analyses of the financial position are provided to permit optimisation of commitment of the residual funds in the SE division.
R W Witty, F M Russell and J Cheney, supported by Lilian Valentine, undertake the above work.
Alvey Conference, Sussex, Left to Right: Tony Dignan (Ferranti), Mike Falla (Software Sciences), Hui Chiu and Mike Russell (RAL) who were part of the Software Engineering team run by David Talbot and Rob Witty. In the white suit behind is Geoffrey Pattie Minister for IT who took over from Kenneth Baker.
As Deputy Director of the Alvey SE Programme RWW had responsibility for the overall strategy of the Programme, its infrastructure policy and the general academic contribution. On 1/8/86 RWW succeeded Mr D E Talbot as Director of the Alvey SE Programme.
2.2.2 ACARD WG on SE
ACARD (the Advisory Council for Applied Research and Development) is a high level body which reports to the Cabinet on long term issues.
An ACARD working group, chaired by John Coplin, the Design Director of Rolls Royce, spent the period Jan 85-March 86 studying the role of software in the manufacturing and service industries, the role and future of the UK software industry and the future impact of software on the UK economy. RWW was a member of this working group through his position in the Alvey Directorate.
The working group produced a report called Software: a vital key to UK competitiveness. RWW, John Coplin and Sir Francis Tombs (ACARD Chairman) launched the report at a Press Conference in June 86. The report has caused some continuing public debate. The official government response is due towards the end of 1986.
2.3 Alvey SE Technical Support and Development Work
The section has now completed its work for the DCS, STI and Common Base programmes and has spent the year concentrating on aspects of specialist infrastructure for the Alvey SE programme such as the conversion of ML/LCF theorem proving system to use the new version of Standard ML.
This section has also handled the relocation of the Group's computing facilities from the Atlas building to R1 in Dec 85.
2.3.2 Standard ML/LCF
ML is a functional programming language which was designed and first implemented at Edinburgh as the metalanguage for the interactive theorem proving system LCF. ML soon attracted interest quite independently of LCF with the result that a number of dialects appeared.
Standard ML is a consolidation of tried-and-trusted developments which has been agreed by the ML community. The final definition of the new language was completed by Edinburgh in the summer of 1985 with Alvey backing. The principal additions for Standard ML are: the data constructors and pattern matching facilities of HOPE, and exception mechanism which enables values to be passed with exceptions, a richer and more systematic set of definition constructs, a module facility based on the latter, and I/O handling.
Mikael Hedlund started work on converting the old ML/LCF to Standard ML in March 1985. The project has two aims:
- to ensure that LCF, as the largest application of ML to date, is fully available for Standard ML, and
- to widen the group's expertise in the implementation of functional languages.
The new implementation follows the same structure as the original, although much of the detailed code has had to be completely rewritten since the new language has a completely new syntax and a richer semantics. The new compiler for Standard ML was completed and tested in June 1986, and the 20 or so LCF source files which were written in old ML have now been transliterated to Standard ML.
The implementation is written in FranzLisp and has been developed on a VAX under UNIX BSD 4.2. FranzLisp is also available on several other machines at RAL (SUN and Pyramid under BSD 4.2, and Atlas 10 under UTS) and the new system will be moved to some or all of these when final integration and testing is completed on the VAX. The system will then be made available for distribution to the Alvey community.
2.3.3 Theorem Provers
Theorem provers are software tools to aid in the process of proving theorems. In the software engineering context we are primarily concerned with theorems about programs, for example that a particular program has a particular property.
Beside the LCF system mentioned above, the group has continued its involvement with two other current systems, the Boyer-Moore theorem prover from the University of Texas and the IOTA system built at Kyoto University in Japan. We have not been working with the Boyer-Moore system ourselves but have continued to act as a UK distribution point for the Alvey SE infrastructure.
Investigations into how best to make IOTA available in the UK are at last nearing a successful conclusion. An attempt to port the original Standard Lisp implementation of IOTA encountered greater difficulties than had been anticipated (differences between early and current versions of Standard Lisp) and was eventually abandoned when a simpler option became available. A version of IOTA ported to Kyoto Common Lisp (KCL) was received from Kyoto in April, together with distributions of KCL for VAX and for SUN. Installation and testing have been undertaken by a research student at Manchester, with assistance from Chris Wadsworth, and a more or less complete system is now running in Manchester on a VAX. The same system also loads successfully for KCL on the SUN but fails when attempting to save the IOTA core image. At the time of writing this is unresolved but it is thought to be due to minor differences between SUN2 and SUN3.
It is hoped to make the KCL version of IOTA available for distribution for the VAX shortly and for the SUN when the problem with saving core images is resolved.
2.4 SEG Research
The overall SEG research theme is Quality Certification of Software Products. Within this theme, the main research interests of the group are in formal specification, the development and application of machine-assisted proof systems and the role of formal reasoning in software development.
2.4.2 Formal Specification
David Duce (DAD) holds two research grants, one from the Alvey Software Engineering Programme entitled Specification of the Graphical Kernel System (GKS), the second is from SERC's Computing Science Committee entitled Theoretical Studies of Emerging Computer Graphics Standards. The latter project was due to start in October 1985, but staff shortages have meant that the technical work has been slow in starting.
Pictures in GKS are described in terms of basic building blocks termed output primitives. GKS allows primitives defined in different coordinate systems to be composed and for different regions of the resulting picture to be viewed on different workstation display surfaces. GKS also provides elegant mechanisms for controlling the appearances of primitives in such a way that the capabilities of different workstations may be exploited to the full, for example a polyline primitive might be displayed as a solid co1oured polyline on one workstation and as a dotted polyline on a monochrome workstation. The technical work in 1984/85 was mainly concerned with specifying the appearance control mechanism in GKS. During this last year, the problem of specifying the geometry of all the GKS output primitives has been tackled and resulted in a paper which was accepted for the Eurographics 86 conference. In essence primitives are represented by the set of points in the real plane which they cover. Primitives as displayed on a workstation are represented as relations between points and colour values.
The 3D extension of GKS, GKS-3D uses the same primitives as GKS itself, and the definitions given are not limited to 2D. This is an essential step in finding a structure for the GKS specification that will encompass GKS-3D also.
The main thrust of the technical work now is to produce a reasonably complete specification of the output side of GKS. Essentially this involves integrating the specifications of the output primitives with the control mechanisms, though in reality this is far from straightforward because without due care the specification rapidly becomes cluttered and unreadable. For this reason the Z specification language (developed by Programming Research Group, Oxford) is being explored. One of the major strengths of Z is the facilities it provides for structuring specifications. One paper has already been written which describes, in Z, a simple model GKS system consisting of a single output primitive and describes the geometry and transformation control for this primitive. The structure of specification closely follows the GKS viewing pipeline model. This has formed the basis for a number of seminar presentations. A more substantial paper including appearance control is in preparation.
The project suffered a major setback with the departure of Elizabeth Fielding at the end of December 1985. At the time of writing, her post had not been filled.
The second project is concerned with two standards currently under development, GKS - 3D and PHIGS. The latter is a standard aimed at the needs of highly interactive applications where the ability to manipulate highly structured pictures is important. PHIGS essentially defines a graphics database which can be edited by the application program. Traversal of the database generates output primitives for display. One of the aims of the project is to define a specification structure that will accommodate GKS, GKS-3D and PHIGS and bring out the conceptual integrity of these systems. So far one paper has been produced, which is an annex to a UK position paper on PHIGS for the ISO Graphics Working Group meeting in September 1986. This paper gives an OBJ specification of a model of the PHIGS structure store, traversal of which generates a sequence of GKS functions which are defined in an OBJ specification of a model GKS system. At this time, PHIGS cannot be defined on top of GKS, the paper explores some ideas for extending GKS and reinforces the UK view that certain extensions would then enable PHIGS to be defined on top of GKS.
Seminars have been given at the Centre for Mathematics and Informatics (CWI) , Amsterdam; the University of Queensland and the Australian National Defence Academy. A paper was presented at a one-day seminar on OBJ, held at Imperial College.
2.4.3 Graphics Standards
David Duce participated in the ISO computer graphics working group meeting in February 1986. He is secretary of the BSI computer graphics panel. Bob Hopgood and David Duce gave tutorials on graphics standards at CERN in December 1985 and at AUSGRAPH 86 in Sydney and at Perth in July 1986.
David Duce and Liz Fielding at the Graphics Standards Meeting at INRIA, Sophia Antipolis
2.4.4 Equational Reasoning
Jeremy Dick's research continues in the field of Automated Equational Reasoning, and has been centred around the theoretical and practical development of his Prolog implementation, ERIL (Equational Reasoning - an Interactive Laboratory). Based on a polymorphic extension of the Knuth-Bendix algorithm which allows a rich form of sub-sorting, ERIL is a useful tool for experimenting with the specification and prototyping of abstract data-types, and can be configured for various models of computation, including functional rewriting, resolution of horn-clauses with defined functions, and narrowing.
During the last twelve months, ERIL has been enhanced in the following ways:
- by compiling rewrite-rules into a shared structure, the efficiency of matching and unification has been increased by an order of magnitude;
- the flexibility of the syntax has been considerably improved by allowing mixfix operators, which is a generalisation of prefix/infix/postfix allowing, for instance, single operators of the form if_then_else_;
- translators between OBJ and ERIL allow ERIL to be used directly with OBJ specifications (but not yet with associative/commutative operators).
Jeremy Dick has been active in promoting the use of ERIL in other departments. It was used as an integral part of an algebraic specification course at the university of Stirling, where it is also being evaluated as the basis of a simple Pascal program verifier for 1st year undergraduate students. It is also in use at two sites connected with the Alvey-sponsored FOREST project, at AERE Harwell and IST, Cambridge.
2.4.5 IPSE 2.5
IPSE 2.5 is a major Alvey project to research and develop an Integrated Project Support Environment based on advanced distributed systems and man-machine interfaces, and incorporating support for both the practice and organisation of design tasks. A major objective is the development of an integrated framework for supporting the use of formal methods in software development, including formal specification and theorem-proving techniques.
The project began in October 1985 with three initial collaborators (ICL, STC and Manchester University) and had a successful first review with the Alvey Software Engineering Directorate in July 1986. RAL applied to join the project from April 1986 and, after some delay, received its grant allocation at the beginning of July 1986. Three additional industrial partners have also recently joined the collaboration.
The project as a whole emphasises the generic nature required of IPSE 2.5. This introduces a key distinction between a generator for an application system and a generated application. This is essentially a conceptual distinction, rather than a rigid division, which is important in understanding the tools to be provided. A generic IPSE may be viewed as a collection of specific generators and other tools together with an integrating framework.
The project thus consists of two main tasks:
- to specify and implement a generic prototype on which specific supports environments can be constructed;
- to construct some such specific support environments for demonstrations that stress formal reasoning and the integration of the practice of design with the organisation of design.
The latter process may be understood as one of instantiating the generic system through the application of relevant generators.
Work to date has mostly concerned the detailed requirements and initial work on the system architecture. Concept studies have been completed for the six main areas of the project. Three of these (design support, management support, and formal reasoning) concern the specific requirements in these areas, while the other three (databases, man-machine interface, and languages) concerns the infrastructure needed to meet the requirements.
Chris Wadsworth has participated in these activities since March 1986 and jointly authored the study of requirements and concepts in the area of formal reasoning. He also represents RAL in the regular project meetings and has contributed to the technical reviews, both internal to the project and with the Alvey SE Directorate.
2.4.6 Transformation and Verification of Occam
Tony Cox (ADBC) has spent almost all his time working on the Alvey (VLSI/CAD) funded research project The Transformation and Verification of Occam Programs, which is a collaboration between Inmos Ltd and Oxford University; at Oxford the project has consisted of Dr Bill Roscoe, supervising, and two research officers, in addition to ADBC. ADBC is one of the SERC facilities requested by Oxford on their RG2!
Occam, a language for concurrent systems, was derived from the language Communicating Sequential Processes, and inherits its clear semantics; in Roscoe, A Denotational Semantics For Occam, a model is given which accounts for both the purely parallel and the internal and store-based aspects of the language. This semantics readily gives rise to a set of laws between Occam programs (Roscoe and Hoare Laws of Occam programming) which can be used as the basis of an automated transformation system. A semantics preserving transformation system offers various possibilities: deciding the equivalence of programs (through transforming to a normal form); improving efficiency; transforming to restricted syntax for VLSI implementation.
A prototype system has been written in Edinburgh Standard ML which is able to parse Occam Programs, apply any of the transformation laws and, most recently, to convert a program to normal form, including infinitary programs to a specified number of communication steps. ADBC has done almost all the programming on the project, while other members have contributed ideas on concurrency theory and the user-interface.
Future work on the project will aim to extend the system to full Occam Syntax, develop useful transformation strategies in collaboration with Inmos and improve the user interface.
ADBC has remained a member of the BSI working group producing a Prolog Standard, taking an active interest in the Semantics sub-group.
2.4.7 Dimensional Design Editor
SEG has been honoured to have Prof Miquel Bertran-Salvans from UPC Barcelona as a visiting researcher for the year beginning March 86. MBS has been interested in Dimensional Design, a graphical technique for presenting software designs originated by RWW and DAD for use in the FR80 Driver and Roots toolset.
A collaborative project between members of SEG (MBS, Duncan R Gibson (DRG), RWW) and DEC(UK) (Tom Povey) was established to build a syntax-driven compiler/compiler/editor system to be called the Dimensional Design Editor (DDE).
A requirement specification, user interface prototype and initial design work had been completed by July 86 when DEC posted Tom Povey to the USA and thus had to withdraw from the collaboration.
MBS has been working on the design of a syntax-driven editor for Dimensional Designs. A document collecting some ideas about the user interface is available.
Basing the approach on the re-use of the same type of display structure, namely the Dimensional Design (DD) itself, the document discusses the interface objects and proposes concrete layout forms for them. The following are examples: Global event sequences or menus, Library Dimensional Design Grammars (DDG's), and the edited DD's. The display of all these objects as DD's is proposed.
The reuse strategy has implications on the internal design since the same data type, namely the DD, is used to represent most of the objects.
Special attention is given to DDG's. They correspond to a possible extension of grammar notation (BNF) to allow the definition of dimensional structures (DD's). An example of their use in the definition of a Simple Dimensional Pascal (SDDP) language is included. As another example the structure of the library is defined with a DDG as well.
DRG did some investigative work on the 4-labelled cuboid model of Dimensional Designs. This involved taking some existing design work by RWW, and producing a translator using the TreeMeta compiler-compiler system. This translator took a simple linear representation of a DD as input, and output the necessary commands to draw it, then a simple pascal program took these commands, interpreted them, and drew the result on a PERQ screen.
Since then DRG has been experimenting with a Pascal program which is the prototype for the user interface of the Dimensional Design editor. This prototype tests ideas and the algorithm used to manipulate trees, both internally to the program and what is presented on the screen to the user. Experience gained here will be invaluable when it comes to building a more complete system.
Other odd jobs have been: sorting out Cambridge Ring equipment (still ongoing!) and distributing copies of the Boyer-Moore theorem prover. DRG has also been responsible for the installation of software on several PERQs. DRG now represents the SE Group at the Unix Service meetings.
SEG welcomed Brian Ritchie to the Research Section in August 1986, and Jill Cheney to the Alvey SE Management section in July 1986.
Elizabeth Fielding left us in December 1985 to go to Holland, and Donald Findley left in March 1986 to return to social work in London.
The staff structure at the end of September 1986 is given in the organogram in Appendix A.
2.6. Future Programme
2.6.1 Specification Research ( D A Duce)
The two current projects will continue their comparison of the efficacy of specification notations and theories as part of the overall search for viable techniques to specify the forthcoming standards such as GKS-3D and PHIGS.
2.6.2 Theorem-Proving Research (A J J Dick)
Future work will revolve around a new SERC-funded project entitled An Assessment of Knuth-Bendix Techniques in Theorem-Proving. The basis of the research is to compare the scope and performance of a polymorphic predicate calculus theorem-prover based on the Knuth-Bendix technique with existing resolution/paramodu1ation-based methods, and assess the long-term potential of this new approach to theorem-proving.
The three-year research programme will be in three phases:
- The extension of the existing ERIL system to treat the full Predicate Calculus.
Two major tasks are involved in such an extension:
- The implementation of a unification algorithm capable of treating associative-commutative functions.
- The implementation of the Recursive Decomposition Ordering with Status for proving the termination of set of rewrite rules.
Given a working model of the extended ERIL system, the second stage is to apply the theorem-prover
to test cases to assess the capabilities of the Knuth-Bendix approach. This is the major part of
the project; its main aims are as follows:
- To compare results with those reported in the literature for resolution-based methods.
- To gain experience in applying the prover to a range of problems in theoretical computer science (eg the processing of algebraic program specifications, the solution of domain equations, etc), with a view to identifying design criteria for a tool, especially use/tool interaction.
- To gain a deeper understanding of the strengths and weaknesses of the Knuth-Bendix approach.
- The final stage will be to produce a feasibility report and design study for a fully-fledged theorem-prover tool providing the full power of the Knuth-Bendix techniques for use in a formal methods environment.
2.6.3 IPSE 2.5 Research (C P Wadsworth)
RAL is funded for the duration of the IPSE 2.5 project to September 1989. The grant application to RAL provides support for Chris Wadsworth as project leader and two other posts. Brian Ritchie joined RAL in August 1986 to fill one of the posts. An offer has been made to fill the other post from October 1986.
The project at RAL will continue the work that has begun on the formal reasoning aspects of IPSE 2.5. The long-term aim is to enable humans and machines to work together effectively in the accomplishment of formal reasoning tasks. Particular objectives within the context and time scale of the overall IPSE 2.5 project are:
- to develop the theorem proving capability for IPSE 2.5,
- to extend the scope, practice, and acceptability of theorem proving techniques in the production of verified software,
- to understand the requirements for integration with other components both in the generic IPSE 2.5 system and in specific support environments,
- to investigate new approaches to proof construction designed for advanced interactive use on large, high-resolution displays equipped with windows and pointing devices,
- to consider evolution to the IPSE 2.5 distributed delivery environment in which the computational power and storage capacity of mainframes is linked with the interactive convenience and flexibility of advanced terminals or single-user workstations, and
- to achieve a design which is readily portable between alternative delivery environments.
Much of the detailed work will be undertaken jointly with project staff at Manchester University, with whom we already have good working links.
2.7.1 Publications - External
- D A Duce and E V C Fielding, Towards a Formal Specification of the GKS Output Primitives, Proceedings of Eurographics '86, A Requicha (ed), North Holland 1986.
- F R A Hopgood, D A Duce, E V C Fielding, K Robinson and A S Williams (eds), Methodology of Window Management, Springer Verlag, December 1985.
- D A Duce and E V C Fielding, Formal Specification - A Comparison of Two Techniques, accepted for publication in the Computer Journal (revision of RAL-85-051).
- F R A Hopgood, R J Hubbold and D A Duce (eds), Advances in Computer Graphics II, Springer Verlag, August 1986.
- F R A Hopgood and D A Duce, Graphics Standards Introductory Tutorial, Australasian Computer Graphics Association, 1986.
- F R A Hopgood and D A Duce, Computer Graphics Programming - Professional Seminar, 1986. Australasian, Computer Graphics Association,
- A J J Dick, ERIL-Equational Reasoning: an Interactive Laboratory, RAL-86-0l0.
- A J J Dick and R J Cunningham, Using narrowing to do isolation in symbolic equation solving - an experiment in automated reasoning, in Proceedings of 8th Int Conf on Automated Deduction, Oxford, July 1986, LNCS Vol 230, pp 272-280.
- Theorem Proving Concepts Paper, IPSE 2.5 project paper, June 1986.
- F R A Hopgood and D A Duce, Graphics Standards The Current State, RAL-86-08l.
2.7.2 Publications - SEG Notes
|95||Alvey SE Visit to USA Oct 85||R W Witty||04.11.85|
|97||Progress Report on the Implementation of Standard ML||P M Hedlund||10.02.86|
|101||SE Research at RAL||D A Duce||13.02.86|
|116||DD - A Layout Algorithm for the Labelled Cuboid Model Draft 2||D R Gibson and R W Witty||14.04.86|
|117||Notes on a visit to CADCAM8 9.4.86||D A Duce||17.04.86|
|119||SEG Development Software on RLVC SML, KCL, IOTA||C P Wadsworth||23.04.86|
|121||Notes on Lancaster Conf on SE||M Bertran||06.05.86|
|122||Conf on Text Processing and Document Manipulation||D A Duce and N Calton||12.05.86|
|124||Progress Report on the Implementation of Standard ML||P M Hedlund|
|130||Chinese Univ Development Project SCIT April-May 1986||R W Witty||10.06.86|
|137||Trip Report||A J J Dick||23.06.86|
|140||Trip Report Australia 86||D A Duce||21.07.86|
|142||Dimensional Design: Requirements Specification for Prototype Draft 3||D R Gibson||20.08.86|
Bob Hopgood and David Duce on the right with Lena and Wendy Olive at Noosa. Bob and David visited Queensland University on way to Sydney
2.8 Conferences, Visits, Seminars
RWW conferences, seminars
- Ada Conference, York Univ, January 86
- Cellular Radio, London, February 86
- Lecture to Sussex branch, lEE, February 86
- USA, Various sites, October 85
- China, SCIT, April-May 86.
Conferences attended by CPW
- Conference on Software Engineering Environments, Lancaster University, April 1986 (Formal Methods Panel member).
- Course on Domain Theory, Laboratory for Foundations in Computer Science, Edinburgh Univ, May 1986 (joint course tutor).
- CADE-8 Conference on Automated Deduction, Oxford, July 1986.
ADBC visits include:
- IFIP International Conference on Functional Languages and Computer Architecture, Nancy, September 1985.
- Occam User Group, University of Kent, September 1985.
Conference attended by DRG:
- Software Engineering Environments, Lancaster, April 86.
MBS has attended the following conferences:
- Programming Support Environments, Univ of Lancaster, April 1986.
- OBJ Tutorial, Imperial College, April 1986.
- CSP Tutorial, Oxford Univ, April 1986.
- Domains, Edinburgh Univ, May 1986.
- International Conference on System Arrays, Oxford Univ, June 1986.
- International Congress, Automated Deduction, Oxford Univ, July 1986.
AJJD made the following visits:
- Stirling University Computer Science department to install ERIL, demonstrate its use, and give a seminar on Equational Reasoning.
- Two day tutorial on Domain Theory at Edinburgh in May 86.
- One day workshop on OBJ, Imperial College, June 86.
- Seminars on Artificial Intelligence in Software Engineering at Frieberg, Germany in June 86.
- Eighth International Conference on Automated Deduction (CADE-8), Oxford, July 86, to present paper on narrowing.
Conferences and Courses attended by PMH
- Functional Programming and Computer Architecture, Nancy, France, September 85.
- Domain Theory Workshop Edinburgh Univ, May 86.
- Technical Writing Course, Harwell, June 86.
In April-May 86 RWW spent four weeks visiting the South China Institute of Technology in the People's Republic on behalf of the World Bank. RWW acted as an advisor to the World Bank's Chinese Universities Development Project which is investing some millions of dollars to improve China's academic computer science teaching and computing services. Whilst at SCIT, RWW gave a course of lectures on software engineering which are to form the basis of a new undergraduate course.
RWW currently serves on the Computing and Control Division Board of the IEE and IEE's Software Engineering Action Group.
3.1.1 Staff List
Current staff are as follows. They are referred to in text by initials.
- C J Pavelin (CJP)
- M B Dunn (MBD) (seconded from Central Office 3/86)
- M K Jackman (MKJ) (joined 11/85)
- C Y L Kwong (CYLK)
- A J Lucas (AJL)
- D F Mac Randal (DMR)
- G A Ringland (GAR) (joined from HEP 11/85)
- J W T Smith (JWS)
This section has three objectives.
Promotion and support of SERC's research programme in AI/IKBS.
This is principally funded through the Alvey programme at present, so most of the work is connected with the coordination, management and technical support of Alvey/IKBS. We work very closely with the IKBS Directorate based at Millbank. As post-Alvey plans become clearer we would expect non-Alvey work to increase, and already some work has begun for committees of SERC's Engineering Board.
- Investigate the application of IKBS techniques to other parts of the RAL programme. This is considered essential for the long-term needs of the Laboratory's scientific and administrative programmes.
- Build up a small funded collaborative IKBS research programme. This is required in order to maintain the effectiveness of the technical support and coordination role. The general theme is knowledge representation.
3.1.3 Progress Summary
The support effort is going very well and the work has adopted rapidly to the changing nature of the requirements with the monitoring of the programme. The new staff appointments during the year have had a very significant impact on the effectiveness of Alvey support.
The R&D effort has suffered because of shortage of effort and the large learning requirements. However the group is now almost up to complement and the technical expertise has rapidly increased, we are in a good state to make progress over the next year.
3.2 Program Management and Support (CJP, MBD)
The Alvey programme has moved, during the period, from a phase of project selection to one where projects are running. Thus the RAL activity in helping coordination of proposals has moved to one of monitoring projects and setting up and running clubs. But the section takes on a vast number of support activities, only the more significant of which can be mentioned in this report.
The last major IKBS round was at the beginning of the year; CJP prepared the cases for the Alvey Board. Since then the group has set up all the IKBS-only clubs, and monitoring is almost totally under way; the arrival of MBD has proved crucial in this. SIGAI, organised by AJL and CJP has become very influential in setting standards for IKBS software, and in initiating new support initiatives. The mailshot under JWS's very active editing is regarded, by the research community, as a questionnaire showed, as a very valuable medium. To sum up, the Alvey IKBS support operation from RAL has been very proactive during the year: we believe this is appropriate for the highly directed Alvey IKBS programme.
In the IKBS/architecture area there is a problem in coordinating work at three different sites (RAL, Swindon and Millbank) on many different aspects of the programme. To combat some of the problems, monthly 'IKBS Management meetings' have been held involving all the key people in the programme. MBD has acted as Secretary to these meetings, which have served as a useful focus for providing early warning of problems and for ensuring a fairly rational distribution of tasks.
Among the plethora of tasks that flow from the IKBS Management meeting, it is worth mentioning that some work has been done on planning a revision of the original IKBS Architecture Study (the blueprint for the Alvey IKBS programme). The Laboratory took the lead in putting together the original study and it is likely that the revised plan will play a significant role in determining whether or not there will be life after Alvey.
3.2.1 Research Clubs
(1) Architecture Club (MBD)
A significant amount of time has been spent getting the Alvey Systems Architecture Club under way. The Club comprises those 23 projects funded as part of the (IKBS) Systems Architecture programme together with others (eg ANSA, Design to Product Demonstrator) with a major architectural component. Projects range from the huge (eg FLAGSHIP, the most expensive Alvey project) to the relatively modest. Project descriptions were obtained from the project managers and bound to form the background to the first meeting of the Club held on 3 June 1986 at the Institution of Civil Engineers in London. Some 45 people attended. A Club Steering Committee has now been set up of which MBD the Secretary and Colin Haley (Director of External Technical Relations, ICL) is the Chairman, to plan future Club events. The Committee proposed that the next full Club meeting (probably in November) should focus on benchmarking new architectures as a theme, and plans are being made to set up various Special Interest Groups and workshops. Additionally, links are being established with the KBS, Logic Programming and High Performance Silicon Structures Clubs, which all suggests a fairly busy programme during the coming months.
(2) Knowledge Based Systems Club (JWS)
JWS has set up, and is Secretary of the KBS Club.
This has involved close liaison with the Alvey IKBS Directorate at Millbank and the Chairman of the Steering Committee.
Since the publication of information about the Alvey clubs in Alvey News there has been an increase in general liaison with the public and the press.
Four KBS Club Steering Committee meetings have been organised since last October, three at Millbank and one at Sussex University. A two day meeting of the full KBS Club was organised in January at Milton Hill House, near Abingdon. There were nearly sixty delegates from twenty-one projects. Eighteen papers were delivered plus other discussion sessions. A limited edition of the proceedings was also produced.
Preliminary arrangements have been made for another two-day Club meeting in January 1987.
Two open sessions and one closed session were organised for the Alvey Conference at the University of Sussex in June.
(3) Logic Programming Club (CJP)
The Logic Programming Club, comprising the Alvey projects in the Logic Programming Initiative and others interested, was set up in the middle of the year. Robert Worden (Logica) is Chairman. An initial business meeting was held in July; it is planned to hold the first technical meeting in late September at Imperial College. CJP is secretary of this club during the set-up phase.
3.2.2 Special Interest Groups (MBD)
As indicated under the Research Clubs heading, an important feature of the Club activities (arguably one of the most valuable aspects of Alvey!), is the formation of Special Interest Groups and the holding of technical Workshops in developing research areas. MBD has been responsible for ensuring proposals for SIGs/Workshops are technically respectable and serve a useful purpose, and for arranging their funding, dissemination of their results, etc.
3.2.3 Monitoring Alvey IKBS Projects
Now that the Alvey IKBS programme is essentially in place, the emphasis has switched to monitoring its success and the effectiveness of individual projects. MBD is responsible for completing the task of appointing Monitoring Officers (usually external technical specialists, in many cases from industry). The Monitoring Officers are contracted to ensure that projects involving substantial DTI funding (often projects costing a few Mpounds and involving several industrial and academic partners) are properly managed and achieve technical targets. With a few exceptions (eg where difficulties remain in starting the projects themselves) the appointment of Monitoring officer is now complete. What remains is the job of monitoring the Monitoring Officers. This is no trivial task and we shall need to ensure that the Alvey Directorate establishes appropriate systems to achieve value for money from its Monitoring Officers. In projects (mainly longer term ones) where no DTI funds are involved, industrial 'uncles' have been appointed to steer the project in the direction of industrial relevance and to act as an interface with the Alvey Directorate. During the period covered by this report, contact has been made with all the uncles and the role expected of them, particularly in monitoring, has been more clearly spelled out.
In-House Monitoring Effort
In some cases the section supplies monitoring effort directly. AJL monitors one of the large Vision projects; CJP has just become Monitoring Officer for a Logic Programming Environment project, and monitors all the SIGAI contracts.
Related to monitoring is the Alvey representation required for the IKBS Community Clubs. MKJ provides this function for the RICS and WIESC Clubs this has involved attending Committee meetings and discussion with project staff.
3.2.4 SIGAI (AJL, CJP)
The Special Interest Group AI, supported and run from this section, has continued to meet over the year to advise the Alvey IKBS Director and SERC Committees on the Infrastructure requirements of the AI Community. A significant piece of work commissioned by SIGAI has been a contract with IST to investigate existing Prolog Benchmarking tasks to discover exactly what aspect of Prolog they are able to measure, and to propose a definitive Prolog Benchmark suite that SIGAI could endorse.
Arising out of SIGAI are a number of support and development Contracts which the section normally initiates, administers and monitors (CJP, AJL). The current set is:
- Prolog support at Edinburgh
- Lisp support at Edinburgh
- POPLOG developments (2 contracts)
- Prolog benchmarks (complete)
- Lisp developments at Bath (Computing Facilities Committee)
- KRSTL at Edinburgh
The latter is a Knowledge Based Toolkit Trials Laboratory, jointly funded with other facilities by DTI and SERC. The initiative came from SIGAI; CJP is on the DTI/SERC Steering Committee for the combined facility.
3.2.5 NIP (CYLK, AJL)
NIP, the New Implementation of Prolog from Edinburgh, is now complete. It has been distributed by RAL to 29 research groups for use on a large range of UNIX based systems. A range of utilities and tools developed on the DEC 10 at Edinburgh are distributed with NIP and are available independently over the networks for users of other Prologs.
Release 1.1 of NIP was received from AIAI at the end of January. This was unloaded onto the following machines:
- SUN2: fileserver, richard and harold
- VAX: rlvd
- Pyramid: rlya
The installation of NIP required a few alterations to a Makefile which determined where the NIP executable files were to be found after installation. Then they were compiled and linked (all with the supplied makefile) . The compiled binaries were then copied to the destination directories for general access. All but the Pyramid were relatively straight forward; the Pyramid was most frustrating due to compiler problems. (Similar problems have prevented a Whitechapel installation).
There have been two new releases of NIP. Release 1.2 arrived at the beginning of February and was successfully installed on the SUNs, VAX and Atlas-10.
Release 1.3 arrived at the end of April (24th) and has been installed on the SUNs, Orion and VAX.
Release 1.3 manuals were received from AIAI and 10 copies of these were made and set to various Prolog users within the Division.
3.2.6 POPLOG (CYLK, AJL)
The development of Poplog, the AI development environment, has continued with the improvement of the Lisp sub-system to the standard of Common Lisp (itself the de facto international standard). It is now commercially supported by System Designers on a wide range of computers. This commercial support is being made available to academics working on Alvey projects through special agreements with the University of Sussex, who are developing Poplog, and a contract between the SERC and System Designers.
POPLOG is run here; we upgraded from V9.2 to V11 during the period. It was discovered that V11 did not work under Release 3.0 of SUNs operating system, after all the Lab 11 machines had been upgraded to run Release 3.0. A SUN 2/50 (edna) which had not been upgraded was set-up in Lab 11 to enable POPLOG to run.
CYLK has incorporated additional modifications from Sussex, to enable POPLOG to be used from Cifer terminals.
POPLOG Stand at the Alvey Conference, Sussex
3.2.7 AI Software List (AJL)
The list of AI software that is available on Alvey Infrastructure computers has been updated 6 times this year to reflect the varying status of the software packages. This is published in both the paper mailshots and on the electronic IKBS Bulletin Board. Its impact on the community has been the increased demand for software especially developed for and packaged for the Alvey programme.
3.2.8 UK AI Toolkits (JWS)
Both SIGAI and the KBS Club have been concerned about the lack of any UK source of advanced IKBS toolkits seen as essential for the next generation of Expert Systems.
As a result of discussions at the January meeting of the KBS Club plus further meetings during the Alvey Conference it was decided to form a small group to consider the production of a UK AI Toolkit, both in the medium and long term.
JWS organised the first meeting at Systems Designers Ltd in August, and another is scheduled for mid-September. Although JWS attended only as an observer it was requested that he act as Secretary and Alvey/RAL liaison.
3.2.9 IKBS Mailshot (JWS)
As the result of a questionnaire sent to the subscribers the IKBS mailshot was reduced from 470 to approximately 360. However with the growth of interest in IKBS in general and Expert Systems in particular the mailing list has grown to over 400 again.
There have been four mailshots so far this year containing sixty-three items in all. Most were volunteered but some were solicited from various people including members of the Laboratory.
A new item was instigated. This is a list of new publications in the area of Expert Systems and Knowledge Based Systems. This list is based on an online search carried out by the library. This does require some additional clerical effort but it is hoped to automate the process in the near future by down-loading records from the remote database. The IEE, producers of the most important database in this area (INSPEC), have bean contacted as we need their permission to re-use the down-loaded material and we are awaiting a response.
In addition publishers have been contacted to request advance information on the latest publications in IKBS and AI. This has led to the possibility of including reviews in the mailshot as some publishers have sent us advance copies of new books.
There has been some contact with European organisations and JWS is currently negotiating with the editor of the ECCAI Newsletter to include it with the mailshot.
During the period covered the collation and mailing of the mailshot was placed with an external mailing company. This required an investigation of the company and a modification of the production procedures. Although this has caused some small delay in one case it has saved a considerable amount of clerical effort on the part of Laboratory staff.
3.2.10 IKBS Bulletin Board (JWS, AJL)
This is an electronic mailshot distributed through USENET and directly. It was set up by AJL, and JWS took over the running in June.
It has not yet achieved the impact we hoped for. JWS intends to increase both the size of the mailing list and the range of contents. This will involve more use of some of the Mailshot material including reviews and other articles. This will not lead to much duplication as there is little overlap between the readers of the Bulletin Board and the Mailshot.
3.2.11 Prolog Standards Meetings
CYLK has represented us on the BSI meeting since the beginning of September 1985.
3.2.12 Benchmarking Prolog (CYLK)
When the different machines were running NIP successfully, they were tested with a set of benchmark programs to have an indication to their relative performances. Also the other Prologs (POPLOG-Prolog, Quintus) were included in the test.
The benchmark suite used was that from Portland State University. These were written by students who were new to Prolog. The advantage of this was that there would not be any optimised Prolog code which might take advantage of certain aspects of different implementations.
The benchmarks were run on:
- NIP: Atlas-10, VAX-d, SUN2
- Quintus: SUN2
- POPLOG-Prolog: SUN2
The results of these tests can be found in IKBS Technical Note 2.
3.2.13 Alvey Conference (CJP)
CJP was actively involved in the organisation of the IKBS demonstrations in this, contacting the projects, providing equipment and giving a summary to the visiting VIPs. (Tony Rush, RAL Administration Division, was responsible for organising displays.) CJP was rapporteur for two IKBS Club sessions, these write-ups to appear in Alvey News.
Cliff Pavelin (centre) at the Alvey Conference, Sussex. IKBS Community Clubs Stand behind
Cliff Pavelin (right) at the Alvey Conference, Sussex. IKBS Community Clubs Stand behind
Alvey Conference, Sussex: Cliff Pavelin talks to Geoffrey Pattie, Rob Morland, Alvey VLSI Director looks on
Responsibility for SUN purchase etc moved into DIC with Peter Kent, although CJP has been much involved with allocation of IKBS infrastructure systems. The major central purchase has been 7 ICL Series 39 machines for some major architecture projects. This involved the usual end of financial year scramble. AJL has continued to be responsible for Whitechapel loans.
IKBS SUNs waiting to be tested, software loaded and shipped
The section (indeed the group) relies heavily on the services of DIC group to provide the basic system and infrastructure support. However certain products used by the section are looked after by section members. MKJ is responsible for looking after our IBM PC software, and has produced a folder for the IBM PC XT. NIP and POPLOG are used within the group and provided by AJL/CYLK.
3.3 Research and Development
The group has the objective of increasing its R&D capability by building up a funded research programme and by embarking on IKBS applications within the Laboratory. These activities are at an early stage at present.
3.3.1 Paralfex (GAR, CYLK)
The PARALFEX (Alvey IKBS approved proposal 077) formally started on 1 November 1985. The goals of the project are to take a real-world knowledge base (KB) - (in the first instance provided by the ALFEX community club) and
- segment the knowledge source to facilitate the modifiability and functional transformation of an expert system
- to identify different problem solving strategies in the financial sector.
Effort on the project so far has been divided into approximately two halves:
- Setting up the advanced AI-toolkit, ART, on the Symbolics and investigating its applicability to our problem. This has progressed satisfactorily.
- Investigating the ALFEX market assessor expert system. Here progress has been less satisfactory. The original plan was based on the delivery of a large working knowledge base from ALFEX in February 1986. This did not happen and we had considerable difficulties with a buggy and unstable KB. The final release of the ALFEX market assessor is now targeted for October 1986 and we intend to have results by January 1987.
We also have an expert loan advisor, based on the knowledge elicitation transcripts provided by the contractors. This was built into an expert system using the shell Savoir (by GAR) and works. We are working on this knowledge base prior to receipt of the market assessor.
GAR was appointed to the Steering Committee of the ALFEX Community Club. A project summary of PARALFEX was submitted to the first Alvey Knowledge Based Systems Club meeting, and invited talk was given at Expert Systems '85. A position paper was submitted to the Surrey workshop on Explanation and subsequently an invited position paper was given at the workshop.
Gordon Ringland (left) at the Expert Systems Community Club Stand at the Alvey Conference, Sussex
A Symbolics 3670 and the ART Knowledge Based System were installed to a functional level with the help of CYLK.
3.3.2 Intelligent Front Ends (DMcR)
This is a collaborative project with the University of Strathclyde.
It is proposed to develop an interactive front end to converse with the user in the appropriate terminology, understand what analysis they require, and drive a back-end, in this case an energy simulation program, to obtain the relevant information. The proposal was rejected by the SERC Environment Committee in the April round, but with encouragement to rework and resubmit. This is being done.
3.3.3 Travel Regulation Expert Systems (JWS, MKJ)
To gain familiarity with both shells and to compare different Expert System techniques, two attempts have been made to build a system to advise on the SERC Travel Regulations.
MKJ wrote an Expert System with about 220 rules on a section of the Conditions of Employment Memorandum (claiming travelling expenses). This used the Expert System shell Xi. A full report will be written on this.
JWS is taking another approach using the ESP Advisor shell - the full version of this was obtained after using it in the Alvey starter pack. As the full text of the Travel CEMs is available on the IBM3081 a novel full text database plus expert advisor system is being developed. Although originally envisaged as a comparison with the application of the Xi shell to the same problem this approach (provisionally called Interactive Annotation) is so different that it cannot be directly compared. In this approach the user is shown the relevant text from the CEM and then asked if a dialogue is required to clarify it. This dialogue takes the form of a series of questions and a final decision on the applicability of the rules contained in the text to the users current problem.
3.3.4 Scheduling EBL Facility (MBD, CJP, GAR)
Another trial exercise arose from discussion with Technology Division, and was used particularly to give CJP and MBD more Prolog familiarity. It was a planning task concerned with the scheduling of the EBL machine. A 'core program' was written by CJP using a very simple algorithm for comparing jobs in a job queue; it was written in (Edinburgh) PROLOG, which proved its worth as a tool for writing elegantly brief logical conditions and its disadvantages in trying to understand control procedures. MBD extended the program to cover the characteristics of real job queues on the EBL and, in the process, have had valuable experience of the SUN workstations and PROLOG. The problem is interesting in that a simple algorithm seems to work quite adequately whereas a 'true' AI approach leads rapidly to a combinatorial explosion and the need for heuristics elicited from the expert human schedulers. Sadly this has become a solution in search of a problem because the introduction of a new Electron Beam Lithography facility has eradicated job scheduling problems!
Some work has been done on identifying fruitful areas for the application of expert systems in administration in the Laboratory and these will be pursued further.
3.3.5 Internal 'Knowledge Representation' Club
GAR formed and coordinated this research club internally, attracting interest outside the section. It consists of a series of meetings with talks from participants on a certain KR theme.
GAR persuaded most participants in the Knowledge Representation Club to agree to write up their contributions for inclusion in a book. Currently he is negotiating with a publisher to produce an introductory text on Knowledge Representation from the IKBS/MMI Group.
Talks so far have been:
- Introduction: GAR
- Object Oriented Programming: ASW
- Semantic Networks: DMcR
- Rule Based Systems: ASW
- Frame Based systems: GAR
- Logic in KR: CJP
- Dreyfus on the Limits of AI: JWS
- Lehnert - Understanding Narratives: MKJ
- Representation of Time: CYLK
- Computational Tractability: ASW
These have been most valuable in improving the technical awareness of the group.
GAR and DMcR are organising a workshop/topical research meeting to look at the field of knowledge representation. This will take place in December. The topics for consideration include logics of knowledge and belief, temporal logic, qualitative reasoning and deep versus shallow knowledge.
3.3.6 Internal Natural Language Group (MKJ, DMcR)
MKJ has been responsible for initiating a natural language group. Part of this has involved getting trainees and sandwich course students in order to initiate the building of some systems. MKJ has supervised Arif Hussain (a trainee) who has now implemented three programs in common lisp illustrating various aspects of natural language understanding. MKJ also supervises Owen Benson who is implementing a more complex natural language understander. This uses Sowa' s conceptual graphs as the basis of the approach.
DMcR supervises two students who are examining Prolog databases, with the intention of devising a query language based on conceptual graphs. The intention is to use the students to lay the foundation for a serious research proposal sometime next year.
The work conducted so far has allowed us to come to grips with some of the fundamental problems of language processing. As a result we know have some ideas for a project (based on the idea of parallel processing systems) . RSRE have shown some interest in these ideas and MKJ is currently pursuing the possibility of funding from RSRE. Also, MKJ has established some contact with the psychology department of Reading University who are interested in submitting (for external funding) a joint proposal on natural language.
3.4 Other Work
3.4.1 IPSE for Energy Modelling Community (DMcR)
DMcR has been involved with the SERC Building Committee's initiative to develop an IPSE for the Energy Modelling Community. This involved supplying expertise for the proposers, and a trip to the USA to discuss the American equivalent with the people at LBL. It is hoped that this will lead to a significant research project at RAL.
3.4.2 Environment Committee AI Facility (CJP)
The Engineering Board Environment Committee has asked RAL to set up and run for it a small facility of AI systems to be made available to its research community for short term trials. It is to consist of (a) an IBM PC plus expert system shells, (b) a standard SUN with LISP, Prolog etc, (c) a machine to support the advanced toolkit KEE. JWS is responsible for (a) and is setting this up. CJP has sent out an operational requirement to workstation (mainly LISP Machine) suppliers for (c), and has asked Edinburgh to assist in the evaluation.
3.4.3 AISB Quarterly - Reviews Editorship (JWS)
After attending the AGM of the AISB, JWS has been appointed Reviews Editor for the AISB Quarterly. This should help raise the profile of RAL as a centre for AI research.
3.4.4 Demonstrations - CHAT, PRESS (AJL, CYLK)
There are now demonstrations of AI techniques available which can be run by any member of the group. They have been used on visiting parties of school children and VIP's.
They involved getting CHAT and PRESS to run on POPLOG which has a few minor syntactic differences from C-Prolog. Both the demos were originally for C-Prolog. Also they were modified to run under NIP. The demos were therefore targeted to be run under C-Prolog, POPLOG, NIP and Quintus prolog. This was achieved but there were still some problems.
In CHAT, not all the questions in the supplied demo files were satisfied. This was closely examined and with a little help from MKJ a couple of bugs were identified. further work by CYLK subsequently solved all the bugs and got all the supplied demo questions answered. Also, one or two additions have been included to correct inconsistencies. CHAT knew about American, African, Asian, etc but not Australasian. This has been corrected.
In Press, there is a lot of sample input to the system and not all of them are answered. It is not known if the building and design of PRESS is good enough to handle ALL the A-level questions that are in the demo files.
The state of both these demos is that now anybody can log into a machine in Lab 11 using the id - ikbs with password - demo, and run them by typing chat or press.
3.5 Staff Changes
Peter Kent moved to DIC group and Andrew Smith left to do a PhD at the very beginning of the period. Since then there has been steady growth (see 3.1.1).
In addition new recruits joining at the end of the period are:
- S C Lambert
- B Bainbridge
- M D Wilson
GAR wrote the advert and took part in a successful recruiting exercise which has led to three more recruits for the section joining shortly.
3.6 Future Programme
Much of this work will continue as now, with rather more emphasis on research clubs and monitoring. A major new initiative is expected to be the revision of the IKBS strategy. The After Alvey plans are likely to become much clearer over the next year, and it is hoped that the IKBS section can have some involvement in them. The environment committee AI facility should be under way.
Research & Development
It is hoped that the funded research effort increases. The current firm plans are:
- Follow up to PARALFEX (expert systems architectures)
- Intelligent front-end proposal
- Collaboration on expert systems with ICRF
It is planned to explore the area of knowledge representation further with an implementation of (part of) Sowa's conceptual studies; this could unify some of the above project and the natural language interest of the group.
In-house applications are still at the purely exploratory stage.
A large amount of paper is produced in connection with the support role, eg SIGAI papers (CJP, AJL) , Alvey Board papers (CJP), IKBS Management meeting (CJP, MBD) , Mailshot (JWS etc), Alvey Infrastructure Committee (CJP) and four series of IKBS notes within the group.
Those with some external impact are as follows:
- Prolog Benchmarks (CYLK): SIGAI paper
- The Application of IKBS in Building Design (DMR): SERC grant proposal
- PARALFEX: Research into Expert Systems in the Financial Sector (GAR). Proceedings of Alvey IKBS Workshop on Explanation Surrey, March 1986.
- UK IKBS research programme (CJP). Proceedings of SEAS Spring meeting Heidelberg, April 1986.
Conferences and Visits
- Workshop on Expertech Xi, (MKJ)
- Alvey KBS Club meeting, January 1986, (MKJ,JWS, CJP, GAR)
- ECAI Brighton, July 86 (MKJ, DMR, GAR, JWS)
- Explanation Workshop, March 1986, (GAR)
- AI in Engineering, (DMR)
- Trip to California to Discuss Energy Modelling, (DMR)
- Symbolics User Group Meeting, Washington, (GAR)
- Hanover Fair, March 1986, (GAR)
- Hanover for discussions at University, January 1986, (CJP)
- SEAS meeting, Heidelburg, April 1986, (CJP)
- Alvey Conference, Brighton, July 1986, (CJP, MBD, GAR, JWS)
- Alvey Computer Vision and Interpretation meeting, September 1985 (CJP)
Overall the section aims to contribute to the general area of Human-Computer Interaction, both by direct research and development activities, and by support for relevant research work elsewhere. These aims are supported by the following means:
- Support of the Alvey MMI (primarily the Human Interface) community. The relationship between the section and the MMI Directorate is not nearly so close as for the SE and IKBS activities, however.
- Research and development, primarily in the window management/user interface design areas, again on behalf of the MMI Directorate.
- Investigation of the more advanced techniques of possible use to the SERC Common Base programme, funded by the Engineering Board's Computing Facilities Committee.
- Direct research grants from Engineering Board committees via the usual (not the) RG2/peer review route.
The section membership has remained relatively stable over the year, with the exception of the two Janets (Haswell and Malone) who transferred to the SUS Applications section, leaving an all-male (but not chauvinist) MMI section. Tony Conway, a psychologist from Bristol University, joined in January. Helen Jenkins joined the group in January and acts as secretary to the section. C M Crampton ably managed the SO to HSO transition through the Group C promotion mechanism at the first attempt the system does seem to work sometimes!
Staff currently in post:
- Ken Robinson (KR)
- Tony Williams (ASW)
- Tony Conway (TC)
- Damian Mac Randal (shared with IKBS) (DMR)
- Mark Martin (MMM)
- Keith Dancey (KGD)
- Chris Crampton (CMC)
- Arthur Seaton (AJS)
- Crispin Goswell (CAAG)
Tony Williams and a PERQ2
4.2 Window Management
The book of the Workshop on Window Management held at Coseners in April/May 1985, was finally published in January 1986. Partly as a consequence of the Workshop, but also after much chivvying of the MMI Directorate, approval was obtained from the Alvey MMI Directorate to proceed with this project. The objective of the project is to specify the software interface between graphical applications and window managers. and to implement the interface on selected workstations.
Graphical applications which use the interface will then become portable to all workstations which provide it. This will accelerate production of applications with good quality user interfaces, and increase the availability of such software.
Following several rounds of refinement of the specification, and comments by the participating manufacturers, a version has been distributed to some potential users for their comments, prior to distribution throughout the community. Work has commenced on the design of window systems providing the CSI, to be implemented on the ICL Perq and SUN workstations. The team is experimenting with software providing terminal emulation and layout management services, which will ultimately be provided along with the CSI. A comparison of the CSI with the X window manager developed by MIT's Athena project is under way. X is beginning to emerge as a de facto standard for Unix workstations in the USA, despite its limitations.
ASW, CAAG, and CMC are carrying out this project. FRAH and KR have been involved in the discussions also, with further comments from most of the section. An eye has also been kept on the budding standards activity in the USA.
4.3 WW - Graphical Toolkit
The graphical toolkit, WW, designed to facilitate the development of highly interactive software, has been substantially enhanced by the addition of new facilities at several levels of abstraction, mainly by MMM. These include, at the lowest level, primitives for circles, bitmaps and cursors; at an intermediate level, support for icons, text manipulation, and external files representation; and at the highest level, a full terminal emulator, interactive filename and directory selection, etc. The complete package has been ported onto most of the workstations available in the Division. The screen editor SPY and the file transfer utility PUFTP have been rewritten to make use of the new facilities.
A utility, MON, for monitoring system activity was written last year for the Perq by CMC using WW. MON has since been ported to the Whitechapel and SUN. An early version was given to ICL, and a more recent version was distributed on the RAL floppy with the PNXS/SR release, as was TED (written by AJS), a graphical time editor also built on top of WW. It displays an analogue clock face and a calendar, so that the user can set the time and date using the mouse. It is intended to genera1ise these and incorporate them into WW.
In order to allow conventional engineering applications built using a mainframe/plotter philosophy (GKS level 0a) to be easily converted to a single user workstation, it was decided to develop a minimal subset of GKS which would reside on top of WW. Since the intention is to aid conversion, not to offer an easy porting path, adherence to the defined GKS levels was not considered essential. However, as far as practicable, the level 0a subset of GKS will be provided. An initial specification has been produced and KGD is doing the coding.
As a full-scale test for the graphical toolkit, WW, it was decided to build a form fill utility. This will handle dynamic, multi-mode forms, allowing the user to input, in any order, to the defined fields. These fields may be text, numeric (integer or real), buttons (eg on/off toggle), popup menus, special (eg date, filename). Data verification is carried out by the form utility, to application - supplied criteria. The form will also be dynamic, that is, the input of a value to one field may cause other fields to (dis)appear, or change their contents. Currently, coding is at an advanced stage. Although the design received comments generally from the Section, most of the design is the work of DMR, AJS and KGD; AJS is doing the implementation.
4.4 PostScript Interpreter
PostScript is a general-purpose programming language, incorporating a powerful set of graphics functions. It is used for description of printed documents, and for communication between document formatting systems and printing systems such as laser printers and typesetters. PostScript was developed by Adobe Systems Inc in the USA, and while the language definition is publicly available, implementations are not.
This project has produced an interpreter for PostScript, which runs on several workstations, and produces images of the document on the display. The interpreter implements the full language, and almost all of the graphics functions. A version has been sent to several user sites for testing and evaluation.
PostScript printers can only produce bilevel output, and use half-tone techniques to simulate colours or grey shades. This interpreter has successfully been used on a greyscale display. Additionally, an experimental distributed processor version permits the language interpreter to run on a different processor from that driving the display.
CAAG wrote the interpreter. He has written a paper about it for the BCS conference on Workstations and Publications Systems, but acceptance has not yet been notified.
The Alvey MMI Mailshot, which has CAAG as the coordinator, is now well established with nearly 2000 subscribers. Three issues have been generated (on a quarterly basis). The contributions to the mailshot come from the four Alvey MMI areas (Human Interface, Image Processing, Speech Processing, and Displays). Each of these areas is intended to have an editor who stimulates, collects and vets contributions; however the HI area as yet lacks such a person and the work is undertaken by RAL.
AJS has been working on the establishment of a Who's Who in the MMI areas. After initial experiments with shell scripts, it was considered necessary to use a real database to handle the volume and complexity of the data required. Accordingly, a system built round the INGRES relational database package was designed and implemented. A simple front-end has been written for end-users. The data collection form was redesigned and the data input partially automated. KGD is responsible for arranging the final form layout (following comments from a group of guinea pigs), and for distribution and data entry.
The Videotape library, managed by CMC, is established, but mainly consists of the SIGGRAPH tapes very few videotapes seem to be available from the community at large. Obviously encouragement is needed to spur investigators into recording real usage of their newly-designed interfaces. KGD has also acted as secretary to the Human Interface Club Committee. As well as organising meetings of HICC (sic) and the writing of minutes and so on, articles have been prepared for Alvey News (one of which received the black hole treatment at Millbank). A one-day seminar on Evaluation was organised at the LSE, attended by 57 people from 31 organisations. Latterly the HICC has been a somewhat moribund organisation following a number of resignations (including the chairman!).
4.6 Common Base Programme
MMM has spent some time implementing SPY and WW on the (never-announced) ICL Perq3. Following ICL's withdrawal from the SUS manufacturing scene, the results of his labours are unlikely to see the light of day (in a commercial sense at least).
CMC was involved with the PNX-5 field trial which finished late in 1985, and ten days were spent at ICL Kidsgrove validating PNX-SR during October. Some liaison with the PBC at ICL has been necessary since. A paper was produced describing the success of the validation.
The official AT&T release of C++ has been brought up on a Vax, and ported to the Perqs, the Pyramid and a GEC-63 at Edinburgh (C M Crampton). Bjarne Stroustrup, the designer and implementer of C++ at AT&T, has been advised of all changes made in order that later versions can be made more portable. Bjarne Stroustrup visited the Laboratory in August, and gave a seminar on What is Object-Oriented Programming?.
Several Perqs have been united with an ethernet and the Perq Newcastle Connection software tried. CMC has also been working with Robert Stroud from Newcastle on adding and testing his UDP driver for the Perq. This is now working sufficiently well that two way communications (via ethernet) are possible between Perqs and VAX and SUN machines (CMC and CAAG).
CMC has also established the ICL tape streamer on the ethernet, and configured the Perq fileserver so as to be useful (since ICL didn't!).
KR has attended meetings of the Single User System Interim Management Panel after handing over the CBP Management to JRG in November, as well as other meetings where a good memory for early events in the Programme was needed!
4.7 SPP in the Foundations of the Design of Interactive Systems
At the end of October 1985 the Computer Science Sub Committee decided to invite proposals for Specially Promoted Programmes, one of which was to be in some aspect of HCI. For a variety of purely administrative reasons, the agent appointed to prepare the proposal - William Newman was not appointed until just before Christmas. He set up a small panel consisting of M A Norman (Heriot-Watt MMI Unit), M J Underwood (MMI Directorate) (later N Bevan, and later still M Tainsh), and FRAH and KR from Informatics. After soliciting contributions from the community, the panel produced a draft proposal (also supplied to the February meeting of CSSC) for consideration at a Workshop held at the Abingdon Guildhall in March. The Workshop, which ASW and TC also attended, was very successful and a good proposal was prepared. At present the proposal is working its way through the funding mechanisms. Competition is stiff and there is no guarantee that the SPP will see the light of day. Nearly all of the Panel support, including the Workshop organisation, was the responsibility of KR [although H Jenkins did most of the work:-)].
4.8 Research Proposals
A proposal to develop a document preview system on a greyscale display was put before CSSC, who so far have been unenthusiastic. The proposal involves developing and evaluating (both for legibility and fatigue aspects) anti-aliased line generation algorithms, using a PostScript implementation on the HLH Orion. Both static and dynamic effects are to be investigated. KR and M Pitteway (BruneI) have written the proposal, with the help of comments from FRAH, , TC and ASW mainly. CAAG has implemented an initial version of PostScript on the Orion in Lab 11.
Following discussions and drafts prepared in the MMI Section, a number of outline proposals for both short and long term projects were developed. Two of these are to be submitted to SERC Committees for funding one on Extensible Graphical User Interfaces, prepared principally by ASW. The other, a joint project with the Chemical Engineering Department of Leeds University, is being prepared by TC and KR, on the use of advanced graphics techniques in a complex, multi (>3) dimensional CAD problem area - chemical plant design.
4.9 Other Developments
As usual, a number of things don't fit too well elsewhere and are treated here.
4.9.1 Finite Element Post Processor
A post-processor for a finite element package is being developed to demonstrate the improvement in the user interface made possible by using single user graphical workstations. A specification of the system has been produced by DMR and a start made on the necessary infrastructure. However, progress has been rather slow due to too many other commitments.
4.9.2 Intelligent Interfaces
In this area, there are strong links to the IKBS section's Intelligent Front End proposal. Also, there is a collaborative project with IKBS (M K Jackman) and SE (A J J Dick), and involving 4 students (0 R K Benson, J Barnsley, A Hussain, K Tarling), looking at natural language understanding. Two have already produced a system for text understanding, based on Schank's conceptual dependencies, and are looking at Sowa's conceptual graph formalism. The other two are looking at Prolog databases, with the intention of devising a query language based on conceptual graphs. It is hoped that this work will lead to a serious research proposal some time next year. DMR and 'TC are involved in this area.
4.10 Future Developments
- Longer-term development work for the SUS programme;
- Continuation of the coordination and support work for the Alvey MMI Directorate;
- Further R&D developments in the Window Management area for the MMI Directorate;
- Research projects (given funding) described in the previous section.
- F R A Hopgood, D A Duce, E V C Fielding, K Robinson, A S Williams, (eds) Methodology of Window Management, Springer-Verlag, 1986.
- A S Williams, C M Crampton, C A A Goswell, Unix Window Management Systems, Client-Server Interface Specification, RAL-86-?, August 1986.
- A S Williams, An Architecture for User Interface R&D, IEEE Computer Graphics & Applications, Vol 6 No 7, July 1986 pp 39-50.
- C Prosser, K Robinson, A S Williams, An Operational Requirement for Assessing Single User Systems, RAL-86-028, April 1986.
- K Robinson, Evaluation of Single User Systems, RAL-86-050, June 1986.
- Workshop on SPP in the Foundations of the Design of Interactive Systems, The Guildhall, Abingdon (organisation by HJ, TC, ASW, FRAH, KR).
- October 1985 meeting of ANSI X3H3 committee and window management subgroup (ASW).
- IEEE Conference on Workstations, November 1985 (CMC).
- ACM Conference on the History of Personal Workstations, January 1986 (ASW).
- Denver Usenix Conference, January 1986 (ASW).
- Third meeting of the Object-Oriented Programming Society, March 1986 (CMC).
- Atlanta Usenix Conference, June 1986 (CAAG).
- Eurographics Conference, August 1986 (ASW).
- BCS Electronic Publishing Specialist Group (ASW - speaker on PostScript).
- IEE MMI Special Interest Group (ASW speaker on Window Management).
- CHI '86, Boston, April 1986 (TC).
- SIGGRAPH, Dallas, August 1986 (KR).
- Eurographics 86, Lisbon, August 1986 (ASW).
5. DISTRIBUTED INTERACTIVE COMPUTING - ALVEY INFRASTRUCTURE
The objectives of this project are
- To provide a computing infrastructure for Alvey projects in IKBS, MMI and SE Directorates. This includes managing the project, supporting users, and developing Unix-based systems and communications.
- To provide and operate an in-house Unix service both to support local users and as a test bed for software and hardware developments, particularly in the areas of distributed file systems.
5.1. 2 Overview
The project has reached a level of maturity, with people actually using the machines. The implementation phase is now complete, but development of local area communications will continue. Funding of the current machines is assured until March 1989 and discussions have started on the evolving role of the Infrastructure into the post-Alvey period.
The main components of the infrastructure are 11 GEC 63/40s and 5 Systime VAX 8750s located at:-
- Cambridge: 2 Ã— GEC
- Department of AI Edinburgh: GEC and VAX
- ERCC: 2 Ã— GEC
- Essex: GEC
- Imperial College: 2 Ã— GEC and VAX
- Newcastle: GEC
- Oxford: GEC and VAX
- RAL: VAX
- Sussex: GEC and VAX
The only change this year has been the move of a GEC from UMIST to Cambridge.
The GEC 63/40s are now running a version of AT&T Unix System V, Release 2 and the VAXes BSD 4.2 A DEC VAX running ULTRIX and an AT&T 3B2 running the reference version of AT&T System V were installed at RAL during the year.
The Infrastructure Project is also responsible for providing a Unix Service on the Atlas-10 mainframe, for supporting over 100 SUN workstations and, since August, for running the Alvey Mail machine.
It will be clear from what follows that the Infrastructure is heavily committed to Unix systems and their communications requirements. The following systems are either fully supported, or impinge on the work of the project:
|VAX||BSD 4.2 (to be upgraded to 4.3)
SYSTEM V 2.0
UTS/System V 2.0
|3B2||System V 2.1 (to be upgraded to 3.0)|
|GEC||UX63 (Sub-contracted to ERCC)|
|Pyramid||BSD 4.2/System V (for ECF)|
|Prime||PRIMIX (Benchmarking/evaluating for ECF)|
|SUN||BSD 4.2 (for Common Base Programme)|
|PERQ||PNX and PNX/SR (for ECF)|
The variety of systems in this list demonstrates how far we are away from a single, universal operating system. It also gives a hint at the problems of providing leading edge local area network facilities both on this site and all the other sites around the country.
Ken Hartley (KFH) took over as project manager on his arrival, with Eric Thomas (RET) handling all the development work and Cyril Balderson (CB) responsible for site contracts and operational matters. KFH is also responsible for the formal link with Millbank through the Alvey Infrastructure Steering Committee, of which Bob Day (RAD) is the secretary, and the Alvey/GEC Implementation Project meetings. CB attends meetings with ERCC and GEC on operational issues (GECOPS) and RET meets with ERCC to discuss GEC support. Site Managers Meetings are organised by Ines Vollmer (IV) and attended by half the Division.
As well as the systems out in the field there is a growing in-house Unix service. The service evolved as systems were acquired but during the year it became clear that a service was being provided which needed the kind of management associated with a mainframe. Consequently a Unix Support Office was established, run by Brian Alston (BAA) and Martin Loach (MJL) , and Systems Administrators appointed: IV for the VAXes and Andy Jackson (AMJ) for the Pyramid and for the LAN service; Tony Lowe (ABL) is responsible for the hardware of the ID LAN.
Towards the end of the year KFH carried out a survey of equipment being used by Alvey projects, on behalf of Keith Bartlett. Janet Smith (JRS) entered the two hundred responses into an INFO database designed by CB. Analysis of the results is not yet complete.
5.3 UNIX Systems
Francis Yeung (HKFY) and Neil Calton (NBMC) have been working on UTS, the UNIX system which runs on IBM mainframes under VM. At present, there are two versions. UTS 2.3 is UNIX Version 7 compatible, and has been running as a service for some time. UTS/V is a UNIX System V compatible version, which was received at the start of the year.
The current UTS service runs on the Atlas 10, in competition with MVS. Users have found that it does indeed provide powerful UNIX facilities, once the problems of the IBM interface have been overcome. However, it is proving difficult to come to some agreement with CCD on the method of accounting: both the actual algorithm and the parameter values. In addition, various attempts by CCD to improve the performance of MVS have had considerable adverse effects on UTS. UTS/V will provide a much better service. The local modifications have been included, and CPROLOG has been successfully tested (with the help of A J J Dick). However, the main stumbling blocks to introduction of a service are Franzlisp and the IBM 4705 controller. The old Franzlisp will not run under the new UTS, and we do not have the source. We have been trying for many months to buy a version for UTS from Franz Inc in the USA, but have run into contractual difficulties. The 4705 is the new communications front-end for the IBMs. One of the advantages of UTS/V is its ability to handle full-duplex ASCII terminals (UTS 2.3 really only works with IBM 3270s). This will allow editors such as vi or emacs to be used. New software is required in the 4705, which has only just come into service. CCD hope to mount this software later this year.
5.3.2 Text Processing
NBMC has been working on three projects in this area: implementing a 6670 backend on UTS, providing support for the 4250 on UTS and VAX, mounting and testing Transcript on VAX and Pyramid.
The tolaser program has been enhanced to allow greater control of output from the 6670 laser printer by the user. Facilities for printing manual pages and obtaining output in a variety of fonts have been provided and will be extended.
NBMC has now assumed responsibility for maintenance of the code produced by Liz Fielding for obtaining titroff output on the 4250 erosion printer. A second 4250 has been obtained and installed in R1. Following discussions with CCD, it was agreed that Informatics should be responsible for the virtual machine which drives this. NBMC has improved the user interface for the software, provided documentation and also given access to the 4250 from the SE VAX (as well as from UTS). NBMC has installed and tested the Transcript software package on the VAXes. The system transforms UNIX documents and graphics files into Postscript format. The package has also been installed on the Pyramid where its full usefulness required the porting of the titroff software. This was done, to the extent that it exhibited the same bugs on the Pyramid as it did on a VAX! Access has been provided from all these machines to two Apple LaserWriters, enabling good quality documents to be produced. Two papers have been written; the first outlines the available software and how to use it, and the second provides a font catalogue for Transcript on the LaserWriters.
There have been various upgrades to the existing VAXes during the years, mostly to the discs. In addition, the old PDP 11/70 was upgraded to a VAX, leaving only a couple of peripherals from the original. One of these, the tape deck, is proving unreliable, and is likely to be replaced. This machine is being used to study Ultrix, the DEC supported version of UNIX.
A DEC Delua ethernet controller has been ordered for the Ultrix VAX. This is a new product, which, apart from making the machine 100% DEC, will offer better performance than the Interlan controller. The major changes to the system have been the conversion of the SE VAX from BSD 4.1 to BSD 4.2, and the incorporation of a considerable number of bug fixes received from Mt Xinu. The former required the porting of the titroff suite to BSD4. 2. In addition, much effort was expended in preparing for, and solving the problems caused by, the move of the VAXes to R1. As well as the VAXes in house, the team is responsible for the support of the Alvey Infrastructure VAX network and the ECF VAX at Strathclyde. Most of the support work is done by Simon Frost (STF) , with help from Jim Aitken (JRA) and Ian Harding (IH). Bugs are usually obscure. For example, STF started this period by investigating a series of hangups that were seriously affecting the SE VAX. These were eventually traced to the terminal interface board device driver.
Another problem, solved by IH, was caused by the power hiccup of early June, which led to the loss of the RA60 disk on the Ultrix VAX for almost a week. When the power failed, spurious data were written destroying the disk's superblock. It took some time to convince the engineers that the disc needed reformatting.
As stated above, a new VAX was obtained to gain some experience of Ultrix. NBMC and IH were involved in the initial mount, and IH has been adding software from the other VAXes to bring it into line with those machines. Early indications are that it looks very much like BSD 4.2 at present, but DEC are producing modifications to incorporate System V features. A new version is expected any time now. NBMC and IH have been co-opted onto the ULTRIX SIG committee. They have been involved in the planning of a general meeting in London.
5.3.4 Fault Reporting System
Last year plans were put in place to set up a fault reporting system at RAL which would enable faults associated with the Alvey Infrastructure machines (both GEC Series 63 and VAX) to be logged and progressed. The work was started by RAD and Mike Woods (MIW), who completed the first phase. Following the reorganisation, which included handing responsibility over to Support, IH began work on the next phase, which will provide a bulletin board. The system should go live shortly.
IH has mounted the system journal software on the VAXes and Pyramid, and has been trying to sort out the accounting software designed for the Alvey Infrastructure GECs and VAXes. The journal software was originally written for the GEC 63s only. It had to be ported to the VAXes - this gave a number of problems, not least because it had to be modified to keep track of the remarkable number of ways that exist to reboot a VAX. A number of bug fixes and enhancements were also incorporated. A common version of the journal is now working on both the Infrastructure GECs and VAXes, as well as the Pyramid.
A similar problem existed with the accounting software, with the added complication that the requirements for what was actually to be measured were changing whilst the software was being developed! IH spent some time in co-operation with ERCC on sorting out the problems, which were made worse by disagreements over what constituted ' Berkeley specific code' . With the 3B2 now available, the credibility of the reference machine is established, and hence the application programs. This should make it easier for people who have to develop application programs for both environments.
Substantial time was also devoted to creating the Usage Statistics and Service Statistics accounting suites for Alvey Infrastructure computers.
HKFY has examined the AIM2 benchmark program in some detail. This consists of a number of programs which test a single aspect of a system, together with a report generator which allows the results to be combined so that they reflect different types of workload (without having to run lots of tests). It is widely quoted in the Industry. However, it is not clear that the report produced is more informative than the raw data from the original tests. Indeed, it is possible to give misleading results if the weighting factors used are not chosen with care. Although one could be very critical about the standard models (which were in fact rather arbitrarily defined), it is not clear how they could be improved.
MIW spent considerable time mounting Release 3 of the Sun O. S on the Division's Suns and dealing with the problems that arose from this. Subsequently he has been attempting to formalise the administration of the SUNs so that a service can eventually be provided. This has included the ability to mail to the outside world from the SUNs, a rationa1isation of the user fi1estore, and various shell scripts to ease adding users in a network environment and maintaining message-of-the-day across a large number of machines. He has also been examining the issue of security (or the lack of it!).
In order to provide a reference machine for future releases of System V, a couple of 3B2's were purchased. These are the port base for AT&T's version of UNIX. HKFY has mounted the latest System V release (Rel 2.1) on one of the 3B2 machines. It took about 18 hours to build the new system (the manual said it would take 24 hours!). The documentation of the new release was bad but a few phone calls to Unix Europe and Olivetti helped to solve most of the problems. Although there are other interesting features in the release, HKFY spent most of the time kernel and in particular the virtual memory part of the system. The 3B work is a part of the project to evaluate System V Release 3 (which was ordered in August).
PERQ: A Perq machine at the Atlas Centre is presently used by Spider Systems to maintain the York X25 Software fault reporting scheme. The problem is that the machine is not located near to the IBM operators and it is difficult for them to find out if the machine is down. One possible solution is to establish a link between UTS and the Perq so that if any abnormality is detected by UTS, a message will be sent to the IBM operator console to alert the shift leaders. HKFY has helped to finalise the plan, and David Hicks (DGH) is in the process of implementing it.
Cyril Balderson (CB) is responsible for the site contracts including reconciliation at the end of the financial year. A major change this year had been the move of the GEC 63/40 from UMIST to Cambridge via GEC (for re-configuration). Further laser printers were purchased and distributed so that there is now one at each site. Two projects to improve the operational side of the system have been the implementation of the Fault Reports System and the Accounts project. CB has been responsible for specifying the requirements and coordinating the implementation between developers and support staff at ERCC.
Ines Vollmer (IV) is System Manager for the Systime VAX. A major task this year has been installation and overseeing autodump. .She has also acted as Liaison Officer with the Alvey Infrastructure Site Managers being responsible for the Faults Database, fault reporting and progress chasing. User registration, equipment movement, and software distribution all fall under this general heading.
RAD put much effort at the start of the year into preparing a strategy document for Communications. This has proved invaluable in maintaining direction in an area which is constantly changing.
The three major areas of work with X25 have been: Coloured books on SUN, York 2.2 and the Pyramid Morning Star board.
RAD, STF and MIW have been involved with the port of Coloured Book software onto the SUNs. This was a beta-test, carried out to assist SUN in tuning their product for the European market. The work has been hampered by lack of documentation (since it was an early pre-release copy) and by the R1 move. In addition, there have been problems getting access to test ports on a PSE. A newer version has just been delivered, but we still do not have access to a production version. When this is obtained, the port of the Coloured Books will be made available to the SUS community.
JRA has spent a lot of time testing the new version of the York X25 software, helped initially by Phil Overy (PJO) and more recently by Ian Johnson (IJJ). PJO fixed a number of holes in the security (obviously, we cannot divulge what these were!), and a number of bugs were reported back to York and Spider Systems. When the final version arrived, IJJ produced a version suitable for distribution to the Alvey sites, during which time more bugs were found and fixed. IJJ will take over responsibility for this work eventually.
Uucp was also provided for Sussex, and a version of uk-uucp mounted under PNX5 on the Perq. Both jobs were done by JRA. The latter was a much larger task than was initially expected as code which works under any other flavour of UNIX caused the Perq to enter the micro-code debugger. ICL had made a considerable number of changes to the terminal driver, including alterations to the data structures. In addition, two versions were required to cover both PACX and PAD connection. Eventually a work-round was discovered and this is now in operation.
PJO spent some time obtaining mmdf (an amazingly complex mail delivery system) and mounting it on a VAX for evaluation. This involved a lot of trial and error, along with lengthy telephone negotiations with Imperial College (who donated the code) to get the bits that turned out to be missing from the first release! It was hoped that this could prove to be useful as a mail interface for the Alvey machines, but in the event proved too large and cumbersome.
5.4.4 Local Area Networks
Much time has been spent trying to formulate a policy for the naming of machines on the Ethernet. After many rounds of discussion, this has finally been put into shape for ABL, the Network Administrator, to use.
RAD, JRA and MIW have been involved in planning and establishing the Division's Ethernet service in RI. This has led to a much better acquaintance with various underfloor and ceiling voids around the building! Previous experience with the development Ethernet in the Atlas Centre allowed the avoidance of some (but not all) of the mistakes that can be made during installation. Two very good decisions were the early acquisition of a Time Domain Reflectometer (TDR) and a Spider Monitor to help with the diagnosis of problems. MIW produced a package to provide a central update service for /etc/hosts (a file containing the internet addresses of the machines on the ethernet). This is now running on the Pyramid and several VAXes. JRA installed a network news reader program for all the 4.2 machines in the Division providing a news service of sorts until this can be implemented using NFS. The mail delivery system was rationa1ised so that local mail can now be sent over the ethernet whilst remote mail will be sent directly or relayed to a host that knows how to deliver it to the final site.
5.4.5 Network File System (NFS)
MIW has spent a lot of time working with SUN's NFS. This has opened up a whole new area of computing possibilities, offering a radically different approach to the sharing of files between machines on the ethernet. It has also brought a new set of problems in security and administration. MIW has become the Divisional expert in these areas. As part of the work, he has set up the Yellow Pages database (used to ease the maintenance of various system files, eg /etc/passwd, across the network) for the Sun network.
RAD and MIW have also been active in making NFS available on the Divisional VAXes. A generic version of the code for BSD4.2 was purchased from the Instruction Set, and RAD arranged for IS to run a course on the internals of NFS. This was attended by MIW and STF. As a result, a preliminary version of NFS has now been mounted on the VAX, and will soon be offered as a service.
The experience gained during the year will be invaluable in evaluating NFS along with other distributed file systems.
5.4.6 Newcastle Connection (NC)
IJG has been very involved in the provision of NC as a service in the Division. This system is broadly similar in concept to NFS, but offers a different way of doing things. The first task was to mount NC on the VAXes. This turned out to be a long job, due mainly to the very poor level of documentation of the code provided by MARI, and to errors in the code distributed. However, by working with Newcastle, IJG has managed to get a version working.
Robert Stroud, of Newcastle, visited RAL during the summer and installed his udp/ip driver code into PNX5. This has meant that the Perqs can now communicate with the VAXes and SUNs using NC, although the udp/ip driver is at present very rudimentary, and not suitable for general release. A trial service is now being set up by IJG to give NC functionality between Perqs, VAXes and, in a limited form, SUNs. Work is also being undertaken in providing NC on the Pyramid.
5.4.7 Access to UTS
Work has been progressing steadily on plans to provide Ethernet access to UTS. RAD has organised a fibre-optic link from the Divisional Ethernet to the Atlas Centre. This has now been commissioned, and will be connected in R26 to an AUSCOM box, allowing UTS to appear on the Divisional network. A project is now being started with HKFY and STF to provide the TP4/LLC1 protocols within UTS.
5.4.8 Ether Pads
Two Bridge Ether-Pads have been purchased. These allow access from ordinary terminals to hosts on the Ethernet that run tcp/ip. Higher speeds are attainable than using PACX connections (9.6 or 19.2 Kbaud), and multiple sessions may be run from a single terminal.
RAD has been involved, with ABL in the setting up of a trial service to evaluate this equipment. In general this has proved to be a popular facility, although the Bridge Pads have demonstrated one or two bugs which are now under investigation with the manufacturer. It is likely that further use will be made of these Pads.
5.4.9 Conversion to ISO Protocols
There is considerable interest in running ISO protocols in place of the ARPA ones. ERCC have mounted Co1oured books over TP4/LLC1 on a VAX with a DEUNA interface. JRA and STF have spent some time converting this code to work over an Interlan interface. Adam Hamilton from ERCC has visited Rutherford on more than one occasion to help with this work. The software makes use of the York Co1oured book code, and problems were encountered when it was discovered that the latest release of the applications programs from Spider was slightly different from the changes that Adam had suggested be put in.
5.5 Single User Systems
The SUS/Applications Section services Alvey infrastructure; its contribution is described under ECF (6.6).
Earlier this year, it was decided to set up a series of demonstrations which might be used in visits by VIPs, Committees, etc. A number of people were involved with the planning. MIW and STF produced two useful demonstrations on the SUN, using the Ether. The first involves using the Pyramid as a compute server (to draw Mandelbrot sets) and the second demonstrates the Network File System. The first of these was used in earnest during a visit by the Head of IT Division at the DTI, which, after the usual last-minute problems, went very well: MIW running the demo and RET giving the presentation.
Other demonstrations include the standard set from SUN and APPLE (on Perq) and CHAT and PRESS (on SUN). Fran Childs was involved in demonstrating these to the International Physics Olympiad finalists when they visited the Laboratory.
5.6.2 ERCC GEC Support Contract
RET is responsible for the GEC Series 63 Support Contract at ERCC, which is proving most successful. ERCC intend to mount NFS on the GECs, and the Steering Committee have also agreed to the purchase of TCP/IP boards. One of the GECs, at UMIST, was underused and has been relocated at Cambridge. These machines are now running a useful service.
However, it is necessary to plan for the future. A paper on the provision of Servers for the community was prepared by RET, RAD, KFH and David Mercer, and submitted to the Steering Committee by KFH (RAD is Secretary). It was favourably received, and further work will be undertaken.
5.6.3 Alvey Mail Machine
During the early part of the year RAL was approached by the Alvey directorate to see if RAL would agree to the running of the Alvey Mail system. RAL agreed to support the Alvey mail system. Shirley Wood (SAW) was responsible for implementing the Alvey software onto the RAL OS4000 operating system. The Alvey users were transferred to a machine at RAL running the Alvey mail system during August 1986.
5.7 Future Work
The main priority for the future must be to ensure that the Infrastructure evolves to meet the changing requirements of its users. This will involve the GEC 63/40s being used as servers for SUN workstations once NFS is available over tcp/ip. At the same time the role of the Infrastructure must be defined, and accepted, beyond the current Alvey programme.
It is recognised that in the long term all Unix systems must evolve towards a common system based on System V, whilst at the same time not jeopardising the current BSD 4.2 systems and their applications. A careful balance will therefore have to be achieved between upgrades to BSD 4.3, particularly on the VAXes, and development of System V. Ultrix may well prove to be a valuable compromise.
UTS must be upgraded to System V, put onto the Divisional LAN and improvements made in its relationships with the host IBM operating systems. We must be clear what this service is for whilst investigating the feasibility of providing a service on the Gray. This could involve mounting Unix on the Cray itself or providing access via UTS or some other Unix machine.
Wide area communications have reached a stable state but evolution of JANET to ISO OS1 will upset this once it gets under way. Gateway facilities between LAN and WAN are likely to be important so that users have access to machines and services irrespective of whether they are connected to LAN or WAN.
Consolidation of current developments of ISO protocols for Ethernet will take place, with the aim of making the transition as transparent as possible to the users. Investigation of distributed file systems will continue and attention should also be given to providing Remote Procedure Calls, so that cpu resources can be distributed in much the same way as files are.
5.8 Statistical Information
5.8.1 Staff Changes
5.8.2 Conferences and Seminars
RET gave a paper entitled UNIX versus PICK at the Oxford BCS in January. MIW assisted with the demonstrations. A report on the meeting has been published in the BCS Computer Newsletter.
KFH attended a meeting on Communications Activities in the Alvey Programme and gave a talk on our communications work (written by RAD and RET).
RET attended a two-day seminar on the proposed ISDN system (digital telephones).
RET, RAD and JRA attended Networkshop 14 in Lancaster, held in April. This turned out to be a disappointing affair as far as Informatics Division comms. issues were concerned, with most of the emphasis on X.25 and uncoupled systems. The conference was also notable for providing at the conference dinner some of the worst wine ever tasted.
RAD and RET attended EUUG in Florence, also held in April. This conference was marred somewhat by a number of the major American speakers staying away due to worries over the Libyans, but nevertheless managed to produce some interesting papers, particularly in the area of distributed file systems.
List of Other Conferences:
|Data Protection Act Guidance||Sept||MEC||London|
|IUCC LAN Conference||Sept||MHR||Reading|
|UNIX Networking Steering group||Oct||JRA||London|
|Inmos Workstation User Group||Dec||IJJ||Middx|
|1986 Winter USENIX||Jan||IH||Denver|
|PERUG Meeting on Networking||Feb||MHR||London|
|Intl Conference on Text Processing||April||NBMC||Nottingham|
|Pyramid US User Group||April||RAD||California|
|ULTRIX SIG Meeting||May||NBMC||RAL|
|GEC Users Convention||May||SAW, ND||London|
|Alvey Conference||June||KFH, RET||Sussex|
|Summer 1986 USENIX Conference||June||IJG||Atlanta|
|DECUS ULTRIX Meeting||June||NBMC|
|EMAP Seminar on IPSE, AI/SE, Fault Tolerant Systems and Risc Architectures||June||HKFY|
|EUU Show at Olympia||June||IH||London|
|UKUUG Technical Meeting||July||NBMC||Manchester|
|ULTRIX SIG Meeting||July||NBMC||Sutton Coldfield|
|Software testing and Verification||July||CDR||Microcomputer Unit|
|Rev 19.4 Course||Sept||MEC||Hounslow|
|Primos rev 19.4 Upgrade||Sept||MHR||Hounslow|
|SERC Interviewing Course||Oct||MEC||Bath|
|System Generation Course||Oct||CDR||Borehamwood|
|C Programming Course||Nov||IJG||London|
|Advanced System generation Course||Nov||CDR, ND||Borehamwood|
|Unix, A Comprehensive Introduction||Nov||KFH||London|
|X25 Networking Course||Dec||CDR||Exeter|
|Primix Course||Jan||MEC, MHR||Hounslow|
|SERC Induction Course||Feb||IJG||Daresbury|
|Developing Skills for women in Middle Management||April||SAW||Sunningdale|
|Prime Systems Programming||April||MHR||Southampton|
|Unix Directions (Instruction Set)||April||MEC, RET||London|
|Seminar on IBM and Unix||May||RET|
|Prime Systems Principles||June||MHR||Southampton|
|Rev20 Upgrade seminar||June||MEC, MHR, JRA||London|
|NFS Implementation Course||July||MIW||RAL|
|Unix and C Course||August||SAW, ND|
|SERC Induction Course||August||KP||Daresbury|
6. DISTRIBUTED INTERACTIVE COMPUTING - ENGINEERING COMPUTING FACILITIES
6.1 Project Objectives
The objectives of the project are to provide a comprehensive interactive computing service to university based research workers. Equipment used includes Prime, GEC, VAX, SYSTIME and PYRAMID Multiuser systems, and PERQ, SUN, and Whitechapel single user systems.
Sun, Whitechapel and PERQ: Martin Prime
6.2 ECF Management
Mike Jane has been responsible for running the newly constituted Engineering Computing Facilities Project (ECF). This results from the combination of the previous Interactive Computing Facility (ICF) and the Single User System (SUS) Project. Management responsibility is under the Engineering Computing Facilities Executive (ECFE).
An early activity centred on the need to rescue something from the Needs of Engineering Computing Working Party Report which was not accepted by the Engineering Board (EB). A new, short paper highlighting the important policy issues for the Computing Facilities Committee (CFC) was taken to all the EB Subject Committees and some Sub-Committees in the April round of meetings to ensure it would have an easy passage through the July Board meeting. This exercise was successful and did provide a rare opportunity to have a discussion with these Committees.
More recently MRJ has been coordinating a CFC constituted Working Party to produce full and costed proposals for a Transputer Research and Application Infrastructure. The final report of this Working Party is now ready to be considered by CFC on 10 October and will go to the Board on 27 November.
- C Balderson (CB)
- A B Lowe (ABL) arrived 1.1.86
- A M Jackson (AMJ)
- R Parkes (RP)
- Miss I Vollmer (IV)
- Mrs P D Athawes (PDA)
- Mrs J R Smith (JRS) arrived 30.9.86
- R Tillotson (RT) arrived 1.9.86
The section provides services to both the major Group Projects - Alvey and ECF and also to the Division as a whole. Emphasis on the activities of the section has been on establishing facilities and services within the R1 environment, and maintaining continuity in the operational aspects of the other projects.
RP is ECF Site Manager for the RAL Site, and PDA has continued with Resource Management for the ECF.
RP is looking after Resource Management for the ECF Service machines located in Atlas, and provides a point of focus for administrative matters for all Prime computers on the RAL site. He was heavily involved with the RL.PA/RL.PB merge operation in setting up users resources and disk allocations on the new machine and assisting with the planning. He is also involved with the similar exercise on RL.PI/TL.PG.
RP has planned and overseen the movement and installation of a number of Prime Computers during the year including the move of RL.PD from Atlas to R1; RL.PH from R25 to Atlas and the upgrades of RL.PA/RL.PB from Prime 750s to a single Prime 9955.
Informatics Machine Room in R1 (Shirley Wood, Andy Jackson, and Ines Day in the background) Systime 8750, GECs, Primes
RP acts as secretary to the Prime Progress Meeting and attends both Prime and GEC Managers and General meetings as the RAL Site Manager. He has also attended RAL/UMIST Liaison meetings and given a presentation at one of these on BRMS (Backup/Recovery Management Service).
PDA has been the Resource Management administrator for the ECF Mini-Computers (Prime and GEC). As part of this task, she has helped with the specification of the new Resource Management Database and manipulation programs which will enable the function to be carried out by Site Managers. She has also helped RP take over the RAL site resource management role and started training for Technology Division to do the same for their machines. PDA has continued to provide a full Resource Management service to the remainder of the ECF community.
PDA was the principal planner in detail and implementer of the RL.PA/RL.PB merger already referred to. This planning led to a huge saving in disk space in the combined machine together with an efficient distribution of files tore . She is producing a report on this exercise, and is now preparing a similar plan for merging the two Prime 750s RL.PG/RL.PI onto a single machine. She also planned and did much of the implementation of the move of RL.PH from a Prime 2250 in R25 to a Prime 2655 in Atlas.
A major task undertaken by PDA was a study of usage of ICF facilities over a three year period, following publication of the Bush Report, to give analysis by SERC Subject Committees.
PDA has recently been undertaking some Capacity Planning work in preparation for the complete transfer of resource management to Site Managers.
CB has been involved in the project with the following activities:
- Prime 9955 installation project. Leading the project to install the machine and transfer user service from the previous RLPA and RLPB machines.
- Medusa evaluation using Prime 9955. Arranging movements of computer and monitoring performance trials. Now planning the move of the user services on RLPG and RLPI to a new Prime 9955.
- Coordination of equipment moves and machine upgrades for various systems.
- Attendance at Prime Progress Meetings.
- Planning change of emphasis from Resource Management to Capacity Planning and Performance Analysis as new Resource Management system become available.
CB was Chairman of the UNIX Liaison Meeting during the year although this meeting has been modified and is now to be Chaired by the Group Leader. He acts as Divisional representative for some aspects of the Central Computing Division facilities: eg attend Site Users meetings, Category Representative, Liaison with Telecoms, Operations etc.
6.4 Multi-User Mini Systems
Staff in post at the end of September 1986.
- R E Thomas (RET)
- P J Isserlis (PJI)
- M H Roberts (MHR)
- Miss S A Wood (SAW)
- N Davidson (ND)
- K Poole (KP)
- C D Rust (CDR)
- R Harris (RH) departed 11/85
- Miss L Sheather (LS) departed 3/86
- L C Peckover (LCP) departed 6/86
- M A Stainsby (MAS) departed 8/86
- I Harding (IH) Unix Systems
- S T Frost (STF) Unix Systems
- H K F Yeung (HKFY) Unix Systems
The year has proved very exacting, with the appearance of a new Operating System, new hardware and with many staff changes. In addition resignations in the Support area have increased the load on System staff. MEC has had his hands full in coping with these changes.
There have been a number of significant changes to equipment at RAL, which have affected the workload. A 9955 was purchased to replace both RLPA and RLPB. This machine needed the new version of the Operating system (19.4.5). In addition, much work was expended in assisting the merger of the two systems. RLPD was moved from R26 to R1 towards the end of November, and the subsequent problems of recommissioning affected the development of the new system. Staff have also been involved in the various upgrades to the five Technology Division machines, the library database machine and the schools machine.
More recently, RLPD has been upgraded to a 9750 (without too much difficulty) and, following a successful test in February, a 9955 model II has been delivered to run the Medusa software. Naturally, this latest Prime range requires the next version of the Operating machine in the system.
Other hardware problems concerned the 2250s at Hatfield and Middlesex Polytechnics. There was a period when agreement had not been reached over maintenance. Consequently, support was suspended, and the machines were not updated. Now this has been resolved, steps are required to reinstate them. At Imperial, further budget problems have caused the machine to become disconnected from the network (after a whole host of difficulties stemming from the form this connection takes). It will be necessary to reconnect this at some time as well. Finally, a new machine at Bath was officially accepted onto the SERC support list, which involved the building of yet another system.
The major work with Primos has involved the mounting of version 19.4.5 and the inclusion of the Name Registration Scheme (NRS). Because of staffing difficulties, the main work has been done at UMIST. However, much effort has been expended here by the Systems team in carrying out the modifications on the RAL machines; in some cases, making up for the lack of skill of site managers. Part of the work involved the removal of the Network changes from the kernel of Primos. This should make it much quicker to introduce new versions. PJI has just received a copy of 19.4.10, and is mounting this, so we will be able to test the theory! This new version is required for the latest Prime machine. However, Rev 20 is due at the end of the year, so more changes will be required before too long.
One of the major items of software requested by the users is Red Book Job Transfer. MHR investigated the offering from Prime (ISOCEPT) and found it was not suitable as it depends on a FTP not used at SERC. However, Salford (who wrote ISOCEPT) have agreed a contract to provide JTMP in a form which will run alongside our existing system. Work is progressing, and is expected to complete by October. UMIST are providing the necessary interface routines. RET is looking after the contract.
Last September, RH was still working on the beta-test version of Primix in use at Surrey. He produced a report which was much appreciated by Prime, particularly in the USA. When he left, LS took over and ran tests at RAL on one more pre-release version, and then on the first production system (1.0). When LS left, MHR continued the work. Primix was mounted for a few users on RLPA (following its upgrade), as well as on the development machine, and RET (among others) has been trying out various facilities. Tests to date confirm the view that Primix is not yet ready to offer a service. Prime have asked that we beta-test the new version of Primix (2.0). This requires Rev 20.2 of Primos! The new Primix, together with a pre-release version of 20.2, is expected any day. Given the currently stated timescales of the official release of Rev 20.2, it is unlikely that a full Primix service can be offered until late this year.
MAS joined the team in April: first as a student in the work placement part of his course and subsequently as a vacation student whilst awaiting his examination results. Apart from performing some of the routine tasks, he has written a report comparing the various file archiving systems available for our possible future use.
ND officially became part of the GEC group in October, bringing the group up to full complement. Unfortunately KP left the Laboratory at the end of August 1986, once again leaving the group short of people.
RLGK and the GEC Systems group moved to R1 during December 1985 which caused disruption to work for a few weeks.
Systems support for all GEC 4000 machines supported has continued with several new systems containing bug fixes and enhancements have been released to all the sites. Support for the PSS gateways continues, as does support for the Pad Print Server. The OTL Diamond service which used RLGK as a gateway to X25 was changed so that RLGB became the gateway.
Informatics Machine Room, R1 (GEC 4090 and Shirley Wood)
The NRS has been installed on the GEC 4000 machines with KP doing the majority of the work. CDR has been responsible for the creation of the tables required by the operating system, and has rewritten the OS4000 mail manual. He also rewrote the OS4000 primer.
As part of the move towards installing 4.15 of OS4000 from GEC, a major change within the RAL system has been the alteration of the packing/unpacking of ids from the RAL algorithm to the GEC algorithm. This required modifications to the ins tore system and the main operating system. ND was responsible for implementing the changes and coordinating the installation of the new system at all the remote sites. However, the deadlines for mounting 4.15 are very tight, still. In order to help the team meet the target, Jonothan Mills (an ex member of the team) has been persuaded to help implement the GEC X25 (he will be putting in the mods that he did before!). RET has managed to arrange for this after a certain amount of negotiation with Contracts (who apparently cannot take out a contract with an individual).
KP has produced the system routines required by GKS for OS4000. This will be incorporated in a new release later in the year.
At the start of the period, Ian Gunn (IJG) was Manager of the Pyramid. This included installing the DELIGHT package, which had been requested by some of the users. Eventually this was achieved although more effort would be required to make it useable. Subsequently, work on DELIGHT was passed to those most interested in it, and no further effort expended on the Pyramid. Peter Randall (PJWR) successfully mounted GKS.
Following the December reorganisation, IH has been involved with the Pyramid, implementing a revised version of the Autodump suite of programs.
This means there is now a common interface on all ID-UNIX System service machines for CCD operators to make tape backups during the night.
A new version of Emacs (Release V2. 02) from Unipress was tested by HKFY. There were a number of problems which were all related to the termcap definition of the Cifer terminal. As there is already a version of Emacs on the machine, there is no plan to release this version in the near future.
STF worked on a number of problems with the Pyramid C compiler found during the porting of various packages. He ported the Berkeley Ingres relational database system to the Pyramid for use on the Alvey Infrastructure Who's-Who project undertaken by the SUS section. Communications work on the Pyramid has been carried out as part of the overall Divisional Strategy, and hence is described in the section dealing with all UNIX Communications issues.
Pyramid intend replacing the York box X25 interface with a board from Morning Star which will handle X25 on-board. Coloured book software is also being mounted. The work is being undertaken by RAD, JRA and IJJ. It has proved a frustrating experience at times, due mainly to out-of -date software being supplied by Pyramid in the USA. It was necessary also to convince the people in the States that X25 was important (and not merely a European special). However, these problems have now been ironed out and the project is nearing completion. A second Pyramid has been installed to help with development work (previously, development was limited, since the machine offered a service). It is hoped to use the new board in service during the autumn. The board should provide a much more reliable service than the York Box, as well as much greater throughput.
When the Pyramid arrived there were initially no communication facilities. A version of the sendmail mail delivery system was mounted by Ian Gunn (IJG) so that users could communicate with other machines over uucp.
NFS is becoming available on machines other than SUNs. A version to run on the Pyramid has been purchased, and will soon go into production use.
Pyramid are very interested in our attempts to mount this software on the 98X. JRA visited the USA to learn more about the OSx kernel, as part of this effort. Once again, work on the Coloured book software has shown up more bugs, which have been reported to York.
6.5 Multi-User Mini Support
6.5.1 Staff in Post
The staff in post at the end of September 1986 included:
- G A Lambert (GAL)
- B A Alston (BAA)
- M J Loach (MJL)
- P C Phillips. (PCP)
During the year Steve Nightingale departed in December 1985 and Paula E Baxter arrived in May and departed in early September 1986, just when she was becoming useful.
Throughout the year the section has suffered from being grossly under complement which has resulted in being only able to satisfy the routine requirements of the user community rather than devote effort in providing some of the facilities that have been promised for many years.
Due to the shortage of staff the decision was taken early in the year that support staff would cover all systems rather than being allocated to a particular machine range. Currently support is being offered to the users on the GEC 4000 series, PRIME 9000 and 50 series, Pyramid, and all UNIX Service systems.
6.5.3 GEC Systems
The level of user activity is showing definite signs of decreasing and it is known that the HEP supported sites at Manchester and UCL will close in August 1987.
User problems are handled by all members of the section but a survey of queries has shown that 35% result from problems with Networking, FTP and mail. Other topics contribute less than 10% of the total.
Version 3.2 of KERMIT has been released and production of documentation etc and bug fixes continues to be handled by MJL. He also implemented the latest version of the NAG Graphical Supplement. The changes needed for NRS RUNOFF Version 5.03 have been made available by BAA and he has also produced a draft manual. A number of bugs with this software have been identified but have s till to be fixed due to pressure of other higher priority tasks.
The WS Editor Version 3.3. produced under a software contract let to Cardiff, has been made available and has attracted generally favourable comment. The final six months of the contract will be devoted to bug fixing and production of final documentation.
Two User/Managers meetings organised by GAL and MJL have been held during the year but it is obvious from the decreasing level of attendance that some other form of user forum is required if user participation is not to totally evaporate.
6.5.4 Prime Systems
The overall level of activity remains at a high level peaking with the introduction of Rev 19.4 of PRIMOS. This produced a significant increase of user enquiries largely centred around performance issues.
This has highlighted the minimal level of accurate benchmarking facilities available to us and BAA is actively involved in developing more extensive facilities in this area based around the PERUG package. Hopefully this activity will be sufficiently advanced to ensure that when Rev 20. n is introduced any pronouncements on performance issues are based upon observed statistics and not indifferent personal plumbing.
BAA has also started work on implementing NAG TOOLPACK on the Prime systems but again due to pressure of other activities this will take many months to complete.
PCP has now completed the final developments on the INFO based Resource Management Database. This allows automatic allocation of USER-ID's and the creation and distribution of User authorisation documents. He has also completed extensive modifications to the usage statistics package, but future work in this area will depend upon the completion of a rewrite of part of the accounting package currently being undertaken at UMIST.
Extensive rewrites of user documentation have been undertaken during the year and MJL has become expert in the conversi9n of the basic input data from UMIST and in fighting the inertia of the Laboratory Reproduction Services so that distribution of documentation can be completed without inordinate delays.
6.5.5 UNIX Systems
The section now has responsibility for providing user support on all UNIX Service machines - ie Pyramid, VAX's etc and the Alvey Infrastructure Series 63 machines. There has been a limited level of UNIX knowledge in the section and rigorous efforts have been made to eliminate this deficiency, with everyone attending UNIX based courses. It was intended that PEB was to concentrate her efforts in this area - see 6.5.1 for the result!
The section now runs the UNIX Service meeting - GAL in the chair, with BAA acting as secretary. It is intended that this meeting should form the main user interface, with the Liaison Meeting concerning matters of policy.
The level of user enquiries on the UNIX front remain fairly low, typically 10 to 20 per week but it is anticipated that this number will increase significantly if and when PRIMIX is mounted on the Prime systems and a wider range of application packages are made available on the Pyramid.
6.5.6 Future Activity
This will be dependent upon the availability of additional staff. Many of the present activities which are only able to stagger along could make rapid progress given that the effort was available. In this category the addition of application packages must have priority. CLADP on the Prime has been ordered and will need support. Prime Information is now available, but we have nobody with any knowledge. The Computer Aided Control System Design initiative will hopefully create interest and probably result in extra demands upon our time and effort.
6.6 Common Base Programme
6.6.1 Staff as at end September 1986
- Julian Gallop (JRG) transferred from CCD end 1985
- F M Childs (FMC) transferred from S&A division November 1985
- William J Hewitt (WJH) arrived September 1986
- L J Jones-Ng (LJJ-N)
- P Kent (PK)
- K M Lewis (KML)
- J C Malone (JCM)
- M J Prime (MJP)
- P J W Randall (PJWR)
- T A Watson (TAW)
- D C Frith left in April 1986
Fran Childs (standing) and the Pyramid still in the Atlas Centre
The SUS/Applications Section supports the Common Base Programme. There is also a small team (most of the year consisting of one person!) responsible for graphics on UNIX systems; increasingly this team will take on text processing on UNIX systems.
6.6.3 Summary of Progress
The turn of the year 1985/86 saw most staff in the section adjusting to new roles: some moved into support from other areas, some moved into software development from support, team leaders were new. Despite these changes, staff adapted to their new roles very ably, not to mention adapting to a new Common Base Manager. Although generally staff members were low, the support team was particularly short of effort between April and September.
The section also became responsible at the beginning of the year for both machine ranges in the Common Base Programme. The CBP was run on behalf of both the Alvey Programme and the SERC' s Engineering Computing Facility (ECF).
At the end of 1985 the SUN2 was replaced in Sun Microsystems range by the SUN3, a machine that is faster in most respects than its predecessor. This led to the situation that in 1986, a very high proportion of the purchased systems were SUN3. Throughout most of the year, there were expectations of a successor to the PERQ2. Two pilot PERQ3 systems were purchased by RAL, but the end of the manufacturing of PERQs in the UK was made public, when ICL announced on 1 August 1986 its new collaborative deal with SUN. At the time of writing, the implications for SERC of this new arrangement are being explored with both companies.
6.6.4 Common Base Support
The Common Base Support team is involved throughout the whole life-cycle of each user machine:
- Selection: liaising with Peter Hemmings in Operations/Support section, initial advice on machine selection and configuration is provided (PK).
- Purchase: machines are ordered using the Central purchase arrangements. There were long negotiations between Harwell Contracts, Swindon Office and RAL over the method by which the central purchase of SUNs was to take place. The chief effect of this was to collapse together the purchase of 60 systems at the end of the financial year. This meant considerable tracking by telephone, ensuring that systems reached their destinations from Frankfurt correctly (PK). Throughout this time, KML looked after the purchase of PERQs.
- Software PERQ: PNX SR, the special release of PNX for PERQ1 completed field trials by the end 1985; this field trial was coordinated by TAW. The new release of this (and PNX 5) made necessary new versions of associated software from ICL such as Pascal, which were unfortunately delayed. However a more serious delay was introduced by the production by external contract of 200 copies of 20 floppy disc, and their return at a time when the support team was short-staffed. KML and FMC have carried out the distribution.
Software - SUN: although the operating system in future will be distributed by SUN, the distribution will only take place after RAL has had the opportunity to beta-test it and check that other essential software works. RAL has had to delay the release of V3.0 to some users because of Pop log not being available with that release.
Because the number of supported computers is increasing, a SUN file server is being installed to be the centre for software distribution over the network (and also for Common Base information and news). In preparation for this, several methods have been investigated (PJWR).
On both machine ranges, Common Base software is distributed on a regular basis (KML on PERQ, LJJ on SUN).
Support: users now contact the Common Base Support service via the CCD Service Line, who report hardware faults to the manufacturer themselves. Some user queries still need to be dealt with by Common Base support and this service is now available 3 hours a day. At various times through the year, it has been staffed by KML, David Frith and FMC.
Some faults are recognised in-house. For example, when PERQs are running Newcastle Connection and other equipment is running the Berkeley protocols, the connected PERQs stopped communicating after a short period. Chris Crampton of MMI section and the ICL engineer ran many tests trying to narrow down the fault. Eventually it was discovered that bringing all PERQ2s up to the current hardware modification level would solve the problem!
One user meeting for each machine range has been held and the first combined Common Base User Forum will be held in October (LJJ, KML).
Current status: the number of supported machines is (at 30.9.86):
PERQ 1 51 (+89 machines no longer maintained) PERQ 2 70 SUN 2 57 SUN 3 100+
6.6.5 Common Base Software Development and Assessment
Projects undertaken this year have included:
- Transcript (troff to Postscript) on the PE~Q (JCM).
- X-windows from MIT: this is a portable window system that has been used by many sites in the USA. It has been mounted on the SUN and is available for use in-house (MJP).
- Assessment of University of Kent software tools. The Common Base section distributes to all users those software tools which have reached a certain level of stability and usefulness. The Kent tools have been used frequently for demonstrators both in-house and on-the-road (MJP).
- Language interfaces to allow Fortran and Pascal users to use WW (Janet Haswell).
- X25 on the PERQ has been packaged for Common Base users (TAW).
- Speeding up and reducing the space occupied by GKS on the PERQ: a paper has been produced, which gives advice on how users may do this (TAW).
- GKS on the SUN: at the start of the year, SUN's own system seemed to be real soon now and had advantages including being level 2C. Two beta-tests and eight months later, there are still problems, and the section is rapidly considering alternatives. (MJP worked on the beta-tests and found the problems. TAW is investigating the ease or difficulty of mounting RAL GKS).
- GKS-3D: an implementation of GKS-3D from GKS-GRAL is being tested. Initial tests seem promising, but currently the GKS workstation occupies the full screen, instead of lying within a window (MJP) , and there are other problems which occur when a session is finished.
- Coloured Books on ISO TP4/LLC1 on Perq: this will allow file transfer etc across Ethernet between Perq and other equipment (JCM).
- Miscellaneous: presentations and demonstrations of commercial software have been arranged - Cambridge Graphics dialogue software and Q'NIAL prototyping language (MJP).
6.6.6 New Common Base Services
The main activities have been:
- ECF Servers: it has always been envisaged that the best way to provide certain services is not to place them on each workstation, but instead to provide access via a fast local network. In August 1986, an operational requirement for a central server facility was sent to 20 manufacturers. It requires access to services for filestore, virtual memory, wide-area communications and ideally for printing. One system will be purchased this financial year but the aim is to purchase a larger number of systems next year for selected sites. Access to the server facility is to be via NFS and Newcastle Connection (JRG, PK, MRJ, Bob Day with valuable comments from other individuals in the division).
- Low-cost workstations: the section received enquiries as to whether a workstation at lower cost than anything currently in the Common Base was possible. Because of this, future products from ICL and Whitechapel had been awaited with interest. The ICL/SUN collaboration changes the situation and at the time of writing, the low-cost evaluation requires a policy decision to reaffirm it or not to proceed.
- High performance workstations: a Silicon Graphics IRIS 2400 workstation was placed at ABACUS, so that it could be tested with some 3D application programs. It will be compared with other high performance workstations, such as SUN3/200 series and Apollo DN580. Liaison with ABACUS on this is by JH.
6.6.7 Graphics and Text Processing on UNIX MUM's
RAL GKS Version 1 has been ported to various UNIX systems - Pyramid and VAX 4.2. The port to Ultrix was started, but suspended due to the imminent arrival of version 2 from CCD Graphics Section. Version 2 is now being put onto the Pyramid and good progress is being made: most of the problems were solved with version 1 (PJWR).
Improvements in the speed of fill area on the Sigma 5000 series were investigated. This is possible because a Sigma has a local area fill. However because it is seed fill, it cannot handle all cases. The problem is recognising the cases that are possible. Some progress has been made, but Version 2 of RAL GKS has taken priority (PJWR).
Some UNIX filters have been written to perform various tasks like converting a SUN raster image to a GKS metafile. It has not been possible to test this fully, as some fixes need to be made to the Versatec virtual machine on the IBM complex (PJWR).
Some testing of Visual Engineering's GKS has taken place. However, the current version does not contain some of the more interesting aspects of that implementation, such as Graph Cap (a method for the rapid addition of drivers) and the application programs (Shirley Wood and PJWR).
This team will take on a coordination role for UNIX text processing in the division: with the arrival of new staff, it should become possible to carry out this function.
As an interim successor to the old SUSSG, a management panel for the Common Base Programme has met 3 times: Single User Systems Interim Management Panel (SUSIMP). JCM is secretary of this.
- PERQ User Notes 1 to 8.
- SUN User Notes 1 to 4.
- Common Base Technical Notes 1 to 35.
M Sparks, J R Gallop, Computer Graphics Language Bindings in IEEE Computer Graphics and Applications (August 1986).
ACUCHE at Hull, July 1986. (JRG)
Control Engineering Workshop at UMIST, July 1986. (JRG)
6.6.9 Conferences and Courses
|Eurographics UK '86, Glasgow||March||PJWR|
|ISO Rapporteur Group for Computer Graphics Language Bindings||April||JRG|
|CHI Conference, Boston, USA||April||MJP, JH|
|Visits to other HCI Labs in Canada and USA||April||MJP, JH|
|State of the Art in Computer Graphics Techniques and Applications, Stirling||June/July||PK|
|EPUG Annual Conference||July||JRG, KML, MJP, TAW|
|SERC/CACSD Conference and Workshop||July||GAL|
|Eurographics '86, Lisbon, Portugal||August||JRG, PK, JCM|
|ISO Working Group in Computer Graphics TC97/SC21/WG2||September||JRG|
|BCS HCI Conference at University of York||September||MJP|
|IPIF 10th World Computer Congress||September||GAL, BAA|
|Communications Course at Exeter||KML|
|Unix Courses||FMC, PEB|
|Advanced Unix Course||LJJ-N|
|PRIMIX Course||BAA, PEB, PCP, GAL|
|Rev 20. Update Course||BAA, PEB, PCP, GAL|
|Networking Ethernet Course||MJL|
6.7.1 Staff in Post at end of September
P J Hemmings (PJH)
The liaison role has three components. These are dealing with SERC research grants in conjunction with Central Office at Swindon, liaising with Central Computing Division on matters of common interest, and a more general one of publicising the work of Informatics Division.
Bernard Loach retired in December 1985 and since that time PJH has been responsible for this activity.
6.7.2 Grant Applications
PJH has been engaged in Liaison with Swindon Office. This has entailed examining grant applications from Engineering Board users (including Alvey) which have a computing component, and from other boards which wish to use ECF facilities. The first group principally involves Common Base Single User Systems and ECF MUMs but also includes requests to use the IBM mainframes and the Computer Board supercomputers. A rough breakdown of numbers is as follows:
|December Round||April Round|
|MUMs, IBM or Cray||39||63|
|Common Base SUS||31||75|
About half of the SUS applications seem to require discussion and/or negotiation with the grant applicant.
All Common Base SUS applications are costed for purchase and maintenance. Comments and costings are produced interactively on a PERQ.
PJH retains the grant applications until they are approved, when all the paperwork is handed over to the respective support teams.
6.7.3 Informatics Division Publicity
PJH is responsible for a variety of activities to publicise the work of the Division. Work under this heading includes:
- Prototype Handout (initially produced for potential recruits)
- Editorial supervision of the newly introduced Divisional Newsheet
- Represent ID on the editorial board of FORUM
- Acting as an RAL guide to visitors.
6.7.4 Conferences, Reports, etc
- PJH attended the CRAY Open Meeting at Imperial College on 8 July and produced an IGLM paper describing it.
- PJH attended the SERC Interviewing Course in February.
- PJH routinely attends meetings of User Groups concerning facilities that have to be commented on ie SUN, PRIME, GEC, IBM and CRAY.
7. DIVISIONAL INFRASTRUCTURE
7.1 Divisional Services
J R Smith (JRS) is looking after all sorts of records of Divisional equipment including that installed in universities, and has become proficient in the use of the INFO database. Specifically JRS has set up and maintained various databases in INFO; viz: Blue-Book, Infrastructure Contacts, SUS distribution, Survey analysis of Infrastructure facilities, Alvey Infrastructure register of users etc (Blue-Book is the Inventory of Divisional Equipment giving location, ownership, contract details etc).
A B Lowe (ABL) joined us from S & A Division, as the AMPTE control programme was completed, and has been responsible for the installation of the Ethernet LAN within the Division. Of necessity, much of the early time was spent in training which has enabled him to become expert in the technology. ABL has taken over the running of the Local Area Network Operations Progress Meeting (LANOPS) which agrees the programme of activities for installations, and identifies requirements for facilities.
Informatics Workstation Room, R1 with Martin Prime and Tony Lowe
ABL is now taking up responsibility for the day-to-day operations management of the UNIX Service within the Division with the various UNIX Systems Managers reporting to him.
R Tillotson (RT) has only just joined us, with the purpose of becoming the System Manager for the Single User Systems which provide a public service. These are the file-servers and the various SUSD in Lab 11. RT will also look after the Network directory (or Yellow Pages).
A M Jackson (AMT) is responsible for VAX RL.VC and the Pyramid RL.PYR-A. These responsibilities are modelled on the duties of the UNIX System Administrator, with the addition of some operational activities and also the management of the Informatics end of the main Computer Room. Much effort has been expended in establishing, testing and overseeing production usage of the automatic security backup procedures on the Pyramid (autodump), and in establishing a tape-library. AMT has also provided assistance to ABL installing and testing new ethernet connections.
ABL has responsibility for the Informatics area of the Computer Room. This means ensuring that the room is kept in a safe and clean manner, as well as arranging for the installation of new machines when required, and the management of the overall space available.
AMJ is the secretary of the UNIX Liaison Meeting and also attends the UNIX Service Meeting, producing statistics on usage and performance for his machines to that meeting.
IV is responsible for the Lab 10 area and has been arranging for the provision of furniture and installation of facilities in the Manuals Reference Room and the Users Printing Area.
AMJ is also the Divisional Safety Officer and a member of the Safety Committee.
7.2 Divisional Administration
G P Dancey (GPD) continued as Divisional Admin Officer (DAO) throughout the year. However, her assistant Janet Fraser left near the beginning of the year. Rather than recruit a replacement, the Division Head's secretary, Carol Barnes, has taken on a number of the Admin duties.
A major activity this year has been to get the finances for the Division automated. GPD is now responsible for keeping an up-to-date record of all the Division's commitment and spend during the financial year, thus removing a level of administration and financial work from the individual Groups.
A continuing activity has been the move of the staff to R1 and the related activity in ensuring the correct furniture and fitments is available.
A Newsheet has been started to provide a vehicle for information flow in the Division.
7.3 Laboratory Committee
As well as the Division Heads Committee (DHC) attended by F R A Hopgood, there are a number of sub-committees of the Laboratory looking after particular functions. The representatives during the year have been:
- Finance Sub Committee - R W Witty supported by E B Fossey from CCD. This vets Divisional expenditure during the year.
- Accommodation and Building Sub Committee - M R Jane attended for the Division until it was abolished for a smaller body on which the Division does not have direct representation. M R Jane has continued throughout the year cajoling and negotiating with all relevant parties in an attempt to get adequate space for the Division in R1. Major achievements have been the addition of the TAMS Workshop and some space on the middle floor of the R1 Link which should allow the Division to vacate R32 next year.
- Manpower Deployment Sub Committee - C J Pavelin. The Division does not have direct representation on this Committee which approves recruitment action. C J Pavelin liaises with the Sub Committee on behalf of the Division and generally looks after our recruitment situation.
7.4 Computer Science Sub Committee
The Computer Science Sub Committee (CSSC) is the main Engineering Board Committee for funding research in the areas of interest of the Division. FRAH attends the meeting on behalf of the Laboratory in order to answer queries concerning facilities available and, also, to provide some support for the Committee where appropriate.
7.5 Divisional Talks
K F Hartley (KFH) has been responsible for the programme of Divisional Talks held during the year. These were:
- Image Processing in an Observatory
- Speech Recognition
- Early History of Atlas Computer Lab
- Stone Chips
- Harwell - Past and Present
- The Cosmic Onion Quarks etc
- The Transputer