Applications Software Programme Review

CFTAG's Recommendations to CFC

1. INTRODUCTION

The review of the Computing Facilities Committee's Applications Software Programme was one of the first major tasks of the Computing Facilities Technical Advisory Group (CFTAG).

In July 1986, the Engineering Board instructed CFC to carry out this review and to report back in two years.

In the event the new EASE policy proposed by CFC is an integrated policy which includes all aspects of CFC's areas of responsibility. The review of the Applications Software Programme has been fed into the discussions which have resulted in the EASE proposal.

This paper nevertheless presents the specific aspects relating to the review of the Applications Software Programme.

2. BACKGROUND

The Applications Software Programme (ASP) was introduced in 1978, at the same time as the Interactive Computing Facility (ICF).

The ASP provided for a number of Special Interest Groups (SIGs), many of which used specific engineering expertise at the Rutherford Appleton Laboratory (RAL). The SIGs were charged with the task of drawing together a community in each specific area to identify, and where necessary produce, appropriate applications software for that community. This software was made available to the community and supported through the ASP.

Although the SIGs were generally successful, a major criticism levelled at the ASP in recent years has been that little or no attempt has been made to phase out those which had reached a 'steady state' and introduce SIGs in new areas.

The two SIGs still active are in the areas of Artificial Intelligence and Electromagnetics. The former is utilising expertise in Edinburgh and the latter at RAL.

Other current ASP activities are continuing specific applications activities initiated by SIGs which no longer exist (eg in Finite Elements).

CFC was charged by the Board to undertake a review of the ASP to ensure that it had an applications programme which related more closely to the current needs of engineers rather than one which simply continued a programme introduced more than 10 years ago.

3. REVIEW PROCESS

CFTAG decided to utilise workshops in specific area as its major review mechanism. A proposal to hold 4 Workshops during 1987/88 was approved by CFC, subject to an overall financial limit of £25K.

The following Workshops were held:

• USER INTERFACE MANAGEMENT SYSTEMS
• DATABASES
• NUMERICAL METHODS
• TOOLS FOR INTEGRATION

Each one was organised by an Advisory Committee supported by ECFE staff from RAL.

Attendance was by invitation and covered a wide range of disciplines and organisations, both academic and commercial, all with experience or requirements as users, designers or investigations of computing facilities.

The objectives of each Workshop were:

• To survey current activity in a particular area to avoid duplication and wrong directions of effort.
• To educate the community at large and the funding and technical committees about the topic.
• To produce a report to those committees which identifies the main points and draws together any consensus achieved in the direction the community believes it should take.

4. CFTAG CONCLUSIONS AND RECOMMENDATIONS

CFTAG considered the workshop reports at its meeting on 4 May 1988, and reached a number of conclusions and recommendations, policy aspects of which have been incorporated in the EASE proposal being submitted to the Engineering Board in July.

There was considerable agreement at all Workshops on the major thrusts for the future. These were:

1. Education
2. Introduction of Standards
3. Technology Introduction.

CFTAG identified a number of main activity areas for each of these major thrusts and produced a set of recommendations for immediate and longer term action.

Main Activity Areas

1. Education
   1. X-11
   2. Databases for Engineers
   3. Graphics Standards
   4. User Interfaces for Engineers
   5. Data Exchange Formats
   6. Languages for Engineers
   7. Numerical Software.
2. Introduction of Standards
   1. X-11 on WW
   2. NAG - via 'CHEST'
   3. STEP/EDIF/CAD*I/IGES
   4. Fortran 77 (8X?)
   5. CGI/PHIGS/GKS
   6. SQL
   7. ADA (Do the users want it?)
   8. 'C' - follow ISO Standard
3. Technology Introduction
   1. Support full range of NAG Numerical Library products
   2. Pre and Post Processing
   3. Full range of 'other' NAG products
   4. REDUCE/ELLPACK - via CHEST
   5. ADA - with support tools
   6. TIGEE/PA-SET/BLOX
   7. NEWS
   8. Relational Databases (RTI Products)
   9. Suggestions from EB committees on software requirements.

The recommended priorities for Immediate Action are:

1. Educate and enlighten UK research community about X-11 and its toolkits via workshops addressed by experts from US and Europe (Done).
2. Set up X-11 toolkits working party (under Prof P Brown University of Kent) to gather and disseminate information on X-11 Toolkits and make recommendations on required enhancements.
3. Convert WW Package to run on top of X.
4. Purchase and benchmark PASet, TIGER and BLOX.
5. Make NEWs available as part of EASE.
6. Purchase and make available software packages like ELLPACK, REDUCE, MAPLE.
7. Make all NAG's libraries available to SERC grant holders, including the adoption of the Harwell Sparse matrix routines ie GENSTAT and GLIM.
8. Keep an eye on developments in parallel architectures.
9. Evaluate progress which could lead to useful enabling software in the areas of mesh generation and intelligent front ends.
10. Improve liaison between engineering community and the database community by way of 'educational exercises' such as joint product reviews.
11. Undertake review of database products relevant to engineers.
12. Encourage use of project management tools and initiate pilot projects, involving tools such as PMW, PROMPT and ARTEMIS.
13. Organise a series of 1-2 day awareness courses on new computing techniques and developments in engineering.
14. Monitor PCTE development.
15. Monitor and participate in developments in data exchange eg STEP, EDIF, CAD*I, IGES.

The recommended priorities for Longer Term Action are:

1. Graphics standardisation: products implementing emerging standards such as GKS-3D and PHIGS should be monitored, with a view to including them in EASE when the corresponding standards are ratified. CGI capability should be integrated with window management support.
2. Object oriented programming systems: these appear to have particular relevance and potential for user interface design and prototype both revolutionary (eg Smalltalk-80) and evolutionary (eg objective C or C++) approaches should be considered.
3. Evaluate the community requirements for ADA and ADA Support Tools ie commission assessment study of YORK offering against other commercial products.
4. Evaluate one or more IPSEs for use in one or more engineering applications.
5. Set up training advisory group with full time staff to direct a training and awareness programme. This group also to be responsible for liaising with other bodies and institutions such as BCS, IEE, CB.
6. Standards: activity participate in defining and supporting standards from design to manufacture.
7. Study techniques for software re-use. Initiate the construction of a national software component library.
8. Evaluate systems analysis techniques for their utility in engineering computing.
9. Fund work on the provision of mixed language environments based on standard languages.
10. Evaluate advanced programming techniques for engineering and their integration. examples are:
    programming techniques Examples are:
    • Specification using VDM, Z, METOO
    • Object-oriented programming
    • 4 GLs.
11. Attempt to define suitable engineering database interface definitions and encourage their implementation by vendors.
12. Carefully monitor the progress of development projects such as POSTGRES and PRODABAS which offer a wide range of datatypes and greater functionality than existing database systems.
13. Assess the portability of PC interface design tools and the possibility of transferring PC products to EASE. This should make available a wide range of low-cost packages if a suitable support route could be established.

5. SPECIAL INTEREST GROUPS

CFTAG recognised the benefit of Special Interest Groups and the responsibility of CFC to initiate the establishment of new ones, in consultation with Subject Committees. In particular CFC should be seen to be drawing specific communities together.

However it was also recognised that CFC's responsibility should not be open-ended, leading to the sort of criticisms levelled at the current ASP.

CFTAG recommends that CFC should take responsibility for setting up and funding new Special Interest Groups. The funding should be for a maximum period of 3 years, with responsibility for further funding beyond this period being assumed by some other funding source (eg a Subject Committee). CFC should coordinate discussions on such future funding.

In the case of the two existing Special Interest Groups, SIGAI and SIGEM, CFTAG recommends that CFC funding should continue for a maximum period of 2 years from 1 April 1988, preferably less. Discussions to identify possible sources of future funding for these groups should begin as soon as possible.