Jump Over Left Menu
Engineering Computing Newsletter: Issue 46
- CHEST Deals for Engineering
- Improving the Quality of Fortran Programs - Reviews
- New Community Club reaches 100 Milestone
- 100 Mflops/i860 chips added to Parallel Evaluation Centre
- OSF/MOTIF Course
- Announcing the Release of a Colour Matching Package
- ERCIM Workshop on Multimodal Human-Computer Interaction (HCI)
- Forthcoming Events
Unfortunately, the details of the future EASE Programme,following the cuts reported in the Editorial of the July issue, have not yet been decided. The meeting of the Panel fell victim of the summer recess and is now scheduled to be held on 15 September. Since the Panel has been asked to produce a programme which better meets the needs of the users, and I doubt the members will have time for an extensive consultation with you the users, you may like to provide your input to the Panel via me. You are invited to send your views to me by 14 September, preferably bye-mail: firstname.lastname@example.org and I will ensure they are made known to the Panel.
Mike Jane, Editor, Rutherford Appleton Laboratory
CHEST Deals for Engineering
CHEST administers a number of agreements for software and datasets. One which is about to begin is for the Compendex*Plus and Page One datasets. The datasets will be available on-line over JANET (in a similar way to earlier CHEST agreements for ISI and Embase), and should be in operation by Autumn 1993.
Details of the products (the on-line service will cover data back to and including 1972) are:
- Compendex*Plus (known as Engineering index (Ei) in print) is a comprehensive abstracts database which covers over 2,600 international engineering journals, conference proceedings and technical reports. 170,000 new abstracts are added annually from 175 disciplines and specialities within the broad engineering areas, i.e. mechanical, civil, environmental, electrical, structural, process, materials science, solid state and superconductivity, bioengineering, energy, chemical, optics, air and water pollution, solid waste management, hazardous waste, road transportation and transportation safety. Besides applied engineering, coverage also extends to manufacturing, quality control and engineering management issues. Over 5,000 new entries in applied physics are also planned for 1993.
- Page One holds references from the table of contents of over 5,400 journals, conference proceedings technical reports and trade publications world-wide. Coverage includes literature from all major engineering societies, commercial publishers and research and development organisations, as well as from university presses and government agencies. A year of Ei Page One has approximately twice as many entries as a year of Ei Compendex*Plus, but without abstracts.
This will be a 5 year agreement (from 1 September 1993), with site licence arrangements similar to those for ISI and Embase. The service may be used by staff or students at a participating site for any non-commercial purpose. Costs will be £2,800 pa for Page One alone, or £5,000 pa for Compendex*Plus and Page One.
Sites or individuals interested in Compendex and its use may wish to subscribe to an electronic discussion list called chest-compendex. Join the list by sending an e-mail message to mailbase@mailbase which has one line of text in the following form:
join chest-compendex <firstname> <lastname>
where <firstname> <lastname> are replaced by your personal names.
Annette Lafford, CHEST/NISS
Improving the Quality of Fortran Programs - Reviews
On 27(28 July 1993, the CFD Community Club (CFDCC) held a two day event aimed at improving the quality of Fortran programs being produced by the academic research community. The first day consisted of presentations on topics such as maintainability, portability, Fortran 90, static/dynamic analysis and metrics. These were followed by presentations and demonstrations from some of the commercial packages available. The second day was devoted to practical hands-on sessions where attendees tried out some of the available software packages. This was very successful, particularly since most of the students brought along some of their own Fortran code to test. Everyone got so involved in this part of the event that more time could easily have been spent on it. If the course is run again, then less time will be assigned to vendor presentations allowing more time for the hands-on sessions.
Following are reports from two attendees on the course. Copies of the presentations from the seminar are available from me.
Debbie Thomas, RAL
I am a PhD student and research assistant with a first degree in mathematics. I came to the seminar knowing that my programming was bad. First I wanted to be informed of what good coding practice was and during the practicals be offered helpful advice and constructive criticism.
I was taken by surprise when the first talk was slanted towards the commercial environment and the management of a large computer code construction. The next talk on maintainability was completely out of my depth and the points made were too specific. Thus it was a relief when the FTK vendor spoke and at last I saw some actual familiar Fortran. The Fortran 90 speaker was quite interesting, but the next three speakers were too much product orientated and not really targeting sales at people like myself.
However, let me not be too negative about the contents of the first part. Although I had not attended the course to learn about approximately 70% of the material covered, I think it will be very useful in the future to be aware of the topics and perhaps apply some of the theories in a very diluted manner to my own simple programming. Thus, it was interesting and entertaining.
The second part of the course was much more geared up to my needs. The evening talk on public domain tools was extremely relevant since these are the only ones that my department provides.
The practical part of the course was good fun. It was invaluable obtaining a flow diagram of a package I was using in my code and actually seeing how it was structured. Although I understood the theory behind the method I had not even attempted to pick through the code to see how computationally the solution was arrived at. I would be very interested in being able to access the tool that I used to produce this flow diagram through some service provided by RAL.
I found the course relaxing and enjoyable although perhaps not exactly what I had expected. It would be tempting to suggest splitting the course into two groups between people like myself and those actually considering purchasing such a tool. However, at least by merging the two it brings awareness of such issues to the first group and hopefully encourages good coding practice. For the buyers it may make them realise that buying the tool alone will not necessarily install quality into the programs of the people they are buying it for. Certain standards must be enforced and training given.
Caroline Lowe, Cranfield Institute of Technology
As a TQ (Total Quality) company with BS5750 registration, Perkins Technology has the objective of improving the quality of its products and this includes the software it produces for internal and external use. Towards this end, I attended this two day seminar.
Fortran programs in industry, I suspect, are mainly generated by engineers and scientists with little background in software generation and CASE tools. Although some of these programmers may be experienced, most of them will have learned their programming techniques through trial and error. However, well written and maintainable software requires more than correct programming techniques.
Good quality assurance techniques are required and to this, the first part of the seminar gave a reasonable introduction. For individual programmers and small groups generating software for internal consumption the seminar would have been very beneficial.
Some of the vendor presentations could be criticized for being aimed more at selling their software than providing useful information. The second day of the seminar was dedicated to hands on use of the tools, but was really too short a time to give them a fair assessment or test their full capabilities. The QA Fortran and LDRA Testbed tools may be useful for large software projects where automated verification, validation and testing are required. However, for small programming teams the cost of some of these tools may be prohibitive.
For companies with large amounts of old and unreadable Fortran 66 or academic code that they wish to convert to Fortran 77, the SP AG tool by Polyhedron Software might be particularly useful. I used the tool to convert some of our old unreadable Fortran 66 code which would have taken me several hours to decipher (literally) and it converted it into code that I could understand in less than a minute.
Overall, the seminar was well organised and very educational.
Colin White, Perkins Technology
As part of the concern of the CFDCC for the quality of software in academic research teams, licences have been purchased for three packages which were used in the above workshop. These are Fortran LDRA Testbed, QAFortran and plusFORT (SPAG). We would like to make these available to the academic community to try them out. This could be done by providing access over the network or by allowing researchers to come to RAL to run their code through the packages with help from Informatics Department staff. If you would like this facility to be made available then please E-mail Debbie Thomas and let me know which option you would prefer.
New Community Club reaches 100 Milestone
Membership of the newly created Parallel Processing in Engineering Community Club, announced in ECN 44; May 1993, has now passed 100.
A questionnaire has been sent to all those registering to identify their application areas and interests in parallel processing.
Initial analysis of returned questionnaires shows that the current use of parallel processing is spanning a wide number of applications domains (see Fig. 1 for a graphical distribution of the members current applications areas).
A full analysis of all members questionnaires is under way. Full results from the analysis will be passed to the Steering Group, currently being established to coordinate the activities of the Club. Membership is open to all engineering researchers with an interest in parallel processing. If you are interested, please contact us to register as a member. Send a message including your name, address, telephone and E-mail details to email@example.com.
C P Wadsworth, Head, Parallel Processing Group
100 Mflops/i860 chips added to Parallel Evaluation Centre
The new Intel 50MHZ i860-XP processing chips have been introduced to the Parallel Evaluation Centre (PEC), operating at Rutherford Appleton Laboratory (RAL).
Two i860-based parallel systems (Transtech Parastations) have now been upgraded to use the newest version of the i860 processor. This processor now operates at 50Mhz with the capacity to complete one integer instruction and up to two floating point instructions per cycle. Peak computation rates are thus 50 MIPS and 100 MFLOPS.
Each Parastation houses four i860 nodes. A Parastation can support 1-4 users each using one or all available i860 nodes. Each i860 node uses a T800 Transputer to provide communications support.
Work is currently under way to support the Intel Hypercube programming style on the Parastation systems. This will provide a local parallel resource to execute applications written for small to medium sized hypercube systems.
Engineering researchers wishing more details on the i860-XP processors and their Parastation enclosures can contact the Parallel Processing Group at RAL which provides technical advice and hands-on support to the PEC.
The upgrade of the i860 systems adds to the existing range of C40, 486 and T800 based parallel facilities available to users of the Centre.
For more details contact me.
Mr B W Henderson, RAL
This course was held recently at RAL and helped to introduce around a dozen people to some of the subtleties of this area of program development It was based around a Motif Gill builder called Xdesigner and the PHIGS graphics facilities made available through PEX under XIIR5. Most of the first day dealt with the GUI builder and this was a revelation to one who has always built such code by hand. I can see little point now in attempting to write widget based applications without such an aid. It was clear however, that such tools do not exempt the programmer from needing a good deal of X background knowledge. Without this a number of things would remain pretty obscure, in particular the use of callback routines and their data.
The end of the first day and the morning of the second day dealt with a brief introduction to PHIGS. Although this was fine for me I can imagine that anyone without experience of PHIGS would have found this rather daunting. Perhaps a health warning on the course might be appropriate!
The bulk of the second day then dealt with ways of using PHIGS to paint into a MOTIF canvas widget and handle events within that widget, i.e., how do you deal with picking PHIGS objects. Here this was done using X events and PHIGS escape calls to identify the objects.
Interspersed throughout the course were hands-on practicals on Sun workstations. These were also well supported with good handouts. Indeed, it is my experience that the handouts are often the most crucial aspect of such courses as one can rarely absorb all the information at the time of delivery. Only when one returns home and tries things out do some of the points really hit home.
So in all, a good, if specialised course for those involved in this area. It may also be the case that separate Gill builder and PHIGS courses would aid those without the background experience embarking on this area.
A J Keane, University of Oxford
Announcing the Release of a Colour Matching Package
As more and more people use colour monitors and introduce colour images into their work a problem getting hardcopies is created. When a hardcopy of an image on the screen is produced the colours have to be converted from the Red, Green and Blue (RGB) colour-space to the Cyan, Magenta, Yellow and Black (CMYK) colour-space. Due to a lack of understanding about the capabilities of the screen and printer this conversion is usually performed in a less than adequate manner. This causes the image on the colour printer to differ greatly from the original image on the screen.
The colour matching package gives a simple method of matching images on monitors and printers. The user does not need to know anything about the underlying algorithms, and does not need any specialist equipment. The only extra equipment required is a Kodak Reflection Density Guide (Kodak Publication Q-16) which can be found in most reprographics departments or ordered from photography shops for a small price. The colour matching package is based on Mike Lamming and Warren Rhodes' paper A Simple Method For Improved Color Printing Of Monitor Images, ACM Transactions on Graphics - Vol. 9, No.4.
The development of the package will be in two stages, this release is stage I. Stage I ensures colours that are grey on the monitor are also grey on the printer, and that the lightness of screen greys linearly maps to printer lightness values. Stage II establishes the amount of black that needs to be printed in addition to the Cyan, Magenta and Yellow, and also how much reduction in Cyan, Magenta and Yellow is needed to compensate. Stage II will also perform hue and saturation adjustment of the image. The colour matching package is written in C under Unix. It provides both command line and an XIIR5 / Motif user interfaces, although X-windows is required to perform the matching process.
The package performs colour matching on PostScript format images. If your images are not already in the PostScript format they can easily be converted using public domain packages like the San Diego Image Tools or the PBMPLUS package. Both are available via anonymous FTP.
RALcol_match is available from the University of Kent as part of the hensa archive. It can be obtained via anonymous FTP from unix.HENSA.ac.uk in the directory /ftp/pub/misc/unix/visual/match. Documentation explaining how to build and use the package is provided with the distribution.
Roger May, Informatics
ERCIM Workshop on Multimodal Human-Computer Interaction (HCI)
Nancy (INRIA-Lorraine), France, November 2-4, 1993
This workshop is focused on the design and implementation of: Natural Mu1timodal HCI for non-specialists and/or the general public, that is on multimodal HCI, involving:
- natural or pseudo-natural language (text and speech),
- unconstrained gestures (pointing, handwriting, drawing...),
- graphics and pictures, sounds.
TOPICS of interest include (but are not limited to):
- Incidence of multimodal communication on the overall architecture of user interfaces.
- Interpretation and generation of multimodal messages
- Practical applications of multimodality and new interaction techniques (cf. pen-based systems, data-gloves,...) in various interaction contexts
- Relevant formalisms for describing the form and content of multimodal communication.
Dr M D Wilson, Informatics
- World Transputer Congress 1993 (WTC'93), 20-22 September 1993, Eurogress, Aachen, Gennany
- BMVC93, 4th British Machine Vision Conference 20-23 September 1993, University of Surrey
- World Robot Championships, 23-25 September 1993, Royal Concert Hall, Glasgow
- Visualization in Electromechanical Engineering Research, 29 September 1993, Oxford University
- CFD Methods in Civil, Coastal & Marine Technology Applications, 17 November 1993,University College, London
- Parallel Processing for Visualization, 17 November 1993, University of Manchester
- COMPUGRAPHICS'93, 6-10 December 1993, Hotel Alvor Praia, Algarve, Portugal
- Data Modelling and Database Design Seminar, 9 December 1993, Rutherford Appleton Laboratory
- General Purpose Parallel Computing, 22 December 1993, University of Westminster
- Multimedia/Hypermedia in Open Distributed Environments, 6-9 June 1994, Graz, Austria