Search
In this issue we have an article on X by Chris Cartledge. Further instalments will appear in the following issues. Comments and experiences with X are invited!
By the time this article is published the results of the PC graphics package evaluation, the recommendations to CHEST and the results of the subsequent negotiations with suppliers should be known. Also a report should be available on the results obtained. This report is being published as AGOCG Technical Report No. 3 under the title PC Graphics Software Evaluation. The report is available from Dr Anne Mumford.
The packages covered by the evaluation fell into three broad categories business/presentation-oriented, scientifically/technically-oriented, and specialist. Business/presentation-oriented packages tended to be good at presentations of bar charts, pie charts and word charts and were good at producing quality presentations quickly. Scientifically/technically-oriented packages tended to have better features for graphs, data analysis and curve fitting. Specialised packages tended to have graphics bolted on to another facility such as statistical calculations or spreadsheet. As with most categorisations of this sort it was sometimes difficult to place packages in one neat category.
The report on the evaluation contains information on how the evaluation was carried out, the criteria used for the evaluation, and detailed tables which can be used to compare the facilities offered by the different packages. General impressions gained by the evaluators on each package whilst evaluating them are also included.
It should be pointed out that the evaluation was carried out under a timetable which meant that all possibilities could not be covered. For instance the whole area of public domain software was not considered. This was not totally ignored but it was felt that issues of support, training and documentation would cause problems in choosing this sort of software for community-wide use. The evaluation did not consider libraries of graphics procedures since such software is already covered by CHEST deals (e.g. NAG Graphics and GKS).
The use of computer graphics in the community covered by CHEST now encompasses a wide spectrum of users. These are those with little experience of computers, who wish to produce high quality presentations, to researchers who wish to produce more technically-oriented graphs and charts for papers and reports. The most suitable packages for those at the first-mentioned end of the spectrum tend to be in the business/presentation-oriented category and for those at the other end tend to be in the specialist category. It was clear from the evaluation that no package would fully satisfy the whole spectrum of use.
It was therefore recommended that CHEST try to negotiate site licenses for two packages, one in the scientifically/technically-oriented category and one in the business-oriented category. It was also recommended that priority be given to the scientifically/technically-oriented category since a site licence deal would be more valuable in this area due to the need to provide support for teaching in networked classrooms.
At the time this article is being written (7 November 1990), negotiations are being arranged with suppliers of the scientifically/technically-oriented software. It is therefore not possible to give precise details of which packages were chosen in this article. This information can be found in the report mentioned above which will be available in January 1991.
The colour printer deal is now well under way. The agreement is with Sintrom for the QMS A4 and A3 thermal printers. These are both PostScript devices. The cost of the printers includes ink rolls, paper and transparencies and the A4 device comes with a memory upgrade. The deal also includes on site maintenance on the devices until 31 July 1994. The cost is fixed until April 1991 at £7,048.50 plus VAT for the A4 printer and £12,462.00 plus VAT for the A3. The deal has been negotiated through CHEST and like all CHEST deals is available to all members of the academic and research community through their site's CHEST contact. I would be interested in feedback on the style of the deal and the experience of using the printers.
We now have a number of CGM tools available in the community. All of these are available either as free source or as site licence deals in the case of the commercial offerings. Last month there was an announcement about the RAL-CGM software being available. This is for translating and viewing CGMs. The C source of this is now available. The LUT -CGM toolkit is a Fortran interface for writing and reading CGMs and it can be interfaced to graphics packages at the device driver level. It is available from Loughborough University. A list of known bugs has been drawn up and information on these can be obtained from me. Further manuals are also available.
On the commercial side we have software available for checking CGMs on Vax, Sun, Apollo, PC and for viewing on a Pc. This is the Metacheck and Metaview Software. We have also now obtained some software for the Macintosh which will interface to Mac products. Contact your CHEST representative or email chest@uk.ac.bath for more information.
Please let me know if you have any experience of setting up UNIRAS CGM profiles for various packages. I feel it would be useful to exchange such information.
AGOCG considers that the development of training materials is an essential part of its work. Training materials need to be available for all software if it is going to be viable. Ideally these will come from suppliers but this is not always the case. Training materials might include sets of OHPs with lecturer notes, student notes, self-teach materials, factscards etc. AGOCG currently has a set of UNIRAS Training Materials and an order form can be obtained from me. Sets of OHPs plus documentation are being developed for Graphics Standards, PRIGS, GKS, CGM, CGI and X. These are in various stages of development. Please contact me for more details.
I have had a question from someone who wants to know if anyone else is working on combining raster and vector images. The person concerned is working on the integration of a variety of bitmap representations and their combination with proprietary vector graphics standards. Areas of particular relevance include manipulation (scaling, zooming) of the combined picture and performance/portability issues. Please contact me if you are interested in chatting with someone else working in this area.
Students on the M.Sc conversion course in Computer Science at the University of Birmingham use Turbo Pascal for their initial programming exercises. We find that the integrated environment helps the less experienced students to get started on their programming. Since I make a point of discussing the GKS standard in my graphics lectures, I wanted to give them some practical experience of using this library in their programming exercises.
During the Summer of 1989, one of the M.Sc students wrote a level zero version of GKS using the Turbo Pascal Graph package and adding the extra level of code to look like GKS. This proved impossible to fit into the system along with the full Turbo environment and the network software. I then spent several weeks reducing the library in size and scope so that it would fit into our system. "Minigks" is the result. I made some use of it last year and this year it is available for the full class to use for their exercises.
Last Summer another M.Sc student produced a C.A.L package GKSTUTOR which worked very well when used by myself and the project assessment team. I shall use it with the full class of 50 students this term and by Christmas will be able to report on how it stands up to this demand.
To summarise, minigks is a library which uses the Turbo Graph library and looks like GKS to the users. It makes no claim to be a full GKS implementation, but I believe that those procedures which are implemented are as close to the correct GKS procedure as is possible within the graph package. It fits within the network software and the Turbo Pascal environment and has provided useful experience for student programs. The gkstutor is also available as an online introduction to the library. If anyone is interested in this software, please contact me.
New members to the list may like to know that they can get hold of all the previous messages. It is in the help information that is sent out to new subscribers.
To see what files are available for your list send a mail message to MAILBASE@NEWCASTLE containing the line
index chest-uniras
The files there are archives (taken regularly) separated into months. So the file for November is 11-1990. To get the file send a message to the mailbase@newcastle containing the line
send chest-uniras 11-1990
And repeat for other months. I think it goes back to 8-1990. Check with the index command.
Thanks to Cristyn Emmett at Newcastle for this information.
The major information flows in a simple computer (taking processor, memory as a unit) are as shown below.
There are many reasons for the display being remote from filestore on disc, data may need to be shared with remote users, for example. It is possible to pull the system apart at point A to give the traditional terminal to a remote system or at point B to produce a workstation system with remote filestore.
Neither approach is better overall, instead both are necessary. A terminal is not usually as desirable as a workstation, but if your application is not implemented on your hardware or it runs 100 times quicker on a remote super-computer then a terminal is necessary.
At Sheffield we currently use 6 different terminal emulators, for reasons too complex to explain here. The most sophisticated, EmuTek, emulates a graphics terminal, the Tektronix 4208. It has one graphics area with a local vector display list, one scrolling text overlay of one font. It has a sloppy serial protocol which means that users have to be careful not to press the right (or even worse the wrong) key at the wrong time. It has a friendly HELP key which does not give help on your application, but instead is a mine of information about the arcane device protocol. EmuTek has given us inexpensive access to this widely used graphics protocol, but it does not give support for the modem graphical user interfaces which are now ubiquitous on personal computers and workstations.
There is only one candidate to replace the babel of terminal protocols: the X protocol. As is not uncommon in computing, a new and obscure vocabulary is used by those in the know. An X terminal may simply be called just that, but it is also referred to as a server while an X application is referred to as a client. So it is usual to talk about X server software for personal computers rather than about X terminal emulator software. The IUSC Workstations Working Party who worked on the Emu-Tek deal is doing a survey of X server products for PC and Macintosh machines with a view to there being a bulk deal for UK higher education.
The X protocol has the basic facilities required for the remote support of graphical applications. It can output two-dimensional graphics, text in a number of fonts, handle bit maps and has specific support for multiple windows. It has mechanisms for handling mouse input, separate identification of button push and release, function key input as well as normal keyboard input.
The X windowing system was originally developed at MIT as part of Project Athena, which pioneered the large scale use of workstations on a campus. The workstations were of various types supporting different software. This gave strong motivation for the development of a terminal protocol. Development is now the responsibility of the X Consortium, with MIT and DEC leading the work and wide industrial participation.
The X Window System is a trademark of MIT, but the specifications and source for model implementations of X have been put into the public domain and can be obtained over the network or for the cost of media and handling. There is no restriction on the use of the software, however it would be improper to violate MIT's trademark, so some implementations of X are sold under another name.
A number of things are associated with a typical windowing system, like Microsoft Windows:
X has none of these - as standard. Instead the X protocol supports basic functionality, while other things are built on top. The X protocol needs a reliable carrier which may be interprocess communication within a single processor, between Unix systems is TCP/IP over Ethernet, between DEC systems is DECNET and in the future may be OS!. X is implemented on many systems including VMS and VM/CMS(!), not only on Unix and its derivatives. The X protocol is currently at version 11.4 but some commercial implementations are still based on earlier, less efficient 11.3 sources. Version 11.4 has some new features but there is a negotiation mechanism which should ensure that an application does not use features not supported by the server.
The X protocol is full duplex, so no echoing takes place locally. Whenever a character is typed, it must be sent down the line to the application for echoing there. Many users of Unix or DEC editors will be aware of the problems of delay on such an interface.
X considerable makes demands on an application in terms coping with events outside the application's control, like user action moving or resizing a window. X applications need careful design and structuring to limit the impact of these demands. Good versions of traditional graphics libraries like GKS and use of toolkits can isolate the application to some extent. A fundamental component of any X system is the Xterm application which allows text applications to use a window on an X server.
(To be continued in the next issue)
