Held at University of Leicester on 6th April 1978.
Present: D. L. Fisher (Chairman) Leicester University J. Adams Hydrographic Dept., Ministry of Defence. R. W. Breward Mapping and Charting Estab (MCE), Royal Engineers (RE) P. Dimmer Leicester University D. A. Duce Rutherford Laboratory K. Y. Kwok University of Manchester Regional Computer Centre R. C. F. Langridge CAD Centre, Cambridge D. H. McLain Sheffield University D. Rosenthal CAAD, University of Edinburgh T. Sancha Cambridge Interactive Systems M. D. Sayers Hatfield Polytechnic H. Tillotson City University M. Williamson CAD Centre, Cambridge K. W. Brodlie (Secretary} Leicester University
As convenor of the working group, David Fisher welcomed everyone to Leicester. As this was the inaugural meeting of the group, he proposed that he himself take the chair, and that Ken Brodlie act as Secretary. If necessary these posts could be reviewed at the next meeting, and certainly ought to be reviewed annually. This was agreed.
Apologies for absence were received from:-
Each person present was asked to introduce himself, and give a brief description of his interests in computer graphics:-
Fisher gave the historical background to the meeting. Recent activity in graphics standards began with the suggestion from the BCS Standards Committee that GINO-F be offered as a standard. This prompted the BSI to propose an ISO working group on computer graphics, with himself, as a member of the BSI committee on programming languages (DPS/13), nominated as chairman. Some sixteen people from seven nations were assembled for a meeting in London in February 1977.
Meanwhile a meeting of international experts at Seillac in May 1976 revitalized the standards movement, especially in the U.S. Following the Seillac meeting, the ACM SIGGRAPH group set up a committee, called the Graphics Standards Planning Committee (GSPC), to define the kernel of a graphics system. The first draft of their report, known as the GSPC Core Report, was ready for the ISO meeting in London in February 1977.
In fact, the ISO meeting was dominated by a page-by-page discussion of the draft. Unfortunately, the GSPC produced a revised draft shortly after the meeting, before the ISO group had a chance to feed back their comments on the first draft. The final draft of the GSPC Core Report was published in Computer Graphics, Vol. 11 No. 3, 1977.
A second meeting of the ISO working group was held in Toronto, in August 1977, at the time of the IFIP conference there. The minutes of that second meeting were circulated with the intimation of this BSI meeting. Fisher resigned as chairman at the end of the meeting, and the chairmanship and secretariat passed to the Netherlands. The new chairman is Paul ten Hagen.
However, it can now be said that the ISO group is on a firm footing. At a meeting of ISO in The Hague in November 1977, the working group was officially ratified as ISO Working Group TC97/SC5/WG2. All decisions on graphics in ISO must pass through WG2. It should also be remembered that ISO is above ANSI in the hierarchical tree of standards organisations.
It was noticeable that other countries nominated their representatives on the ISO working group by means of some well-defined mechanism. In fact, at the next meeting to be held in Bologna in September, all representatives must be ratified by their national standards organisation.
The U.K. ought to reflect the same structure as other countries, and Fisher saw the formation of this BSI group as a first step. The group will be responsible for nominating representatives to the ISO working group. For the next ISO meeting, which will be considered as the inaugural meeting of TC97/SC5/WG2, Fisher will attend in his capacity as past-chairman. Any others who can raise the necessary finance are eligible for selection.
Fisher suggested that the terms of reference of the group be left as general as possible, namely that WG5 should act as an advisory body for DPS/13 on all matters relating to graphics. (The structure is as follows: WG5 reports to DPS/13 which reports to the BSI.)
Fisher saw the membership as a small body of highly motivated people. Broadly, people who attended would be regarded as members; people who did not attend but corresponded would be regarded as correspondents; failure to attend three consecutive meetings (without apology) would relegate a member to a correspondent.
Members were asked to supply names of people who might wish to join the group, or act as correspondents. There was a general feeling that better contact with commercial and industrial users of graphics was required.
Fisher introduced a general discussion of the GSPC Core Report (mentioned earlier in Section 2). The impression gained at the ISO meeting in Toronto was that the ACM SIGGRAPH had virtually decided on the standard, and that only cosmetic changes to the Core Report would be considered. However, different implementations in the U.S. have highlighted a large number of areas of the report which are not well-defined. It is reported that a review body has been set up to reassess the Core Report, so perhaps major changes will now emerge. This increases. the significance of the ISO WG2 meeting in Bologna, as there seems a greater chance now that the Americans will listen to other ideas.
Definite information on the current status of the GSPC Core Report is urgently needed. Mike Sayers indicated that he was due to visit the U.S. twice in the course of the next two months, and volunteered to make contact with Bob Dunn and Jim Michener, two leading contributors to the Core Report. Tom Sancha predicted that the two men would have different views: Dunn is likely to be anxious to press ahead with the definition of a standard, Michener is more likely to wish some redefinition of the GSPC proposals.
It was agreed that, at this first meeting, there was nothing to be gained from working through the Core Report page-by-page. Instead it would be better if some specific issues were discussed.
David Rosenthal felt that the GSPC were addressing the wrong problem. Instead of defining a software package, they ought to be defining a standard pseudo-device. Sancha agreed.
It was also remarked that the GSPC Core Report assumes a high band width between the applications program and the display. David Rosenthal and Sancha strongly disagreed with this assumption.
The various possibilities for current position (CP) were discussed: no CP, one CP, or several CPs (e.g. one for drawing, one for text, one for input, etc.). Sancha felt there should be no CP, so that for example the routine to draw a line in 2D would be specified as
VEC (X1,Y1,X2,Y2).
Removing the concept of CP avoided problems of how to define CP after a transformation, and after text output.
It was suggested that, while a system without CP should act as the basis, a layer could be added on top in which the concept of CP was used.
Sancha remarked that the proceedings of the Seillac conference contained a good discussion of the issues surrounding CP.
Sancha summarized three current views on graphical input:
Sancha mentioned that there will be a further Seillac meeting, in June 1979, which will concentrate on graphical input.
Rosenthal agreed with Sancha's views on input, but wished to go further: graphical output should also be expressed in terms of character strings.
Fisher pointed out that Mike Larkin had put forward the quite reasonable view in Toronto that the manipulation of graphical files and data was merely a subset of data base technology.
Sancha said that he knew of three distict views on the nesting of segments:-
There was much support for (b), he said, but it was difficult to define a clear way of doing it. The problem occurred in invoking segments at an intelligent satellite. Thus (a) tended to be accepted.
Sancha felt that the mapping to attributes should be done at device driver level.
John Adams raised the question of the continuous variation of attributes, such as continuously varying line-width, and urged that this problem be given consideration.
Fisher suggested that this BSI group should restrict its view to passive 2-D graphics, and try to get that right first. Sancha agreed, but Rosenthal strongly opposed the suggestion, feeling that any standard which did not cater for graphical input would be useless.
Fisher gave the history behind this project. He had long felt there was a need for a library of high-level graphical routines, just as there is a need for a library of numerical analysis routines - a need which has been satisfied by the NAG library.
An initial effort was made at Leicester to build a transportable library of graphical routines based on the GHOST plotting system, but before the project got off the ground, the management of NAG expressed an interest in the idea. They saw the inclusion of a graphical output facility as an important addition to the library. Thus it was natural that Leicester should channel its effort into helping NAG create a graphics chapter.
Thus a working party was set up to put forward proposals to the NAG executive, and these will be discussed by the executive on 17th April. (Added later: Proposal to create a NAG graphics chapter agreed by NAG executive - Secretary).
Brodlie gave a brief summary of the proposals, which were described in the paper circulated with the intimation of the meeting. Of main interest to this BSI group is the design of the graphical interface, a small set of primitive graphical routines which will act as an interface between the NAG library routines and the underlying graphics package at any site.
It seemed generally agreed that the set of routines which had been chosen as the interface were well suited to their purpose. However the suggestion that the NAG interface routines might form the basis of a passive 2-D standard met with opposition: the fact that the concept of current position was used made it unacceptable (see earlier discussion in section 5).
The problem of highlighting was discussed, and a distinction was made between highlighting and specifying attributes. In the case of highlighting (as defined by the GSPC), it is of no importance which attribute is used to make information stand out - thus highlighting is device dependent.
Another area where it was felt that DPS/13/WG5 should become involved was in the transmission of graph plot files. Fisher mentioned that the Inter-University Software Committee was planning to create a working group to study the transmission of graphical information. He hoped to keep in touch with that group.
The group was urged to take note of the ARPA network graphics protocol (Reference: Sproull, R.F. and Thomas, E.L. A Network Graphics Protocol, Computer Graphics vol.8,1974, pp 27-52).
David Duce agreed to send a copy of the network independent file transfer protocol, prepared by the High Level Protocol Group. This protocol is used within the Atlas Computing Division of the Rutherford Laboratory.
Adams recommended that the group make contact with DPS/19, the BSI group responsible for standards in numerically controlled machine tools, who could well have interests in common with DPS/13/WG5.
It was agreed to hold the next meeting in London, on Wednesday 31st May, starting at 11 a.m.
Ken Brodlie, Secretary. 25th April 1978
May 31st 1978 : Next meeting of DPS/13/WG5 in London
Sept 28th 1978 : IMA Conference on Mathematical Methods in Computer
Graphics and Design: Leicester University.
Sept 27th-29th 1978: Next Meeting of ISO TC97/SC5/WG2 in Bologna, Italy.
June 1979 : Further meeting of experts at Seillac,
to be concerned with graphical input.
J Adams Hydrographic Department, MOD J F Bell ADCD, Ordnance Survey R W Breward MCERE, K W Brodlie, Leicester University P Dimmer, Leicester University D Bradley, NEL D A Duce, ACD, Rutherford Laboratory D Fildes University of Glasgow D Fisher Leicester University B Ford, N.A.G. C Hall NPL F R A Hopgood ACD, Rutherford Laboratory A Kilgour University of Glasgow K Y Kwok University of Manchester Regional Computer Centre, R C F Langridge CAD Centre, Cambridge. G D S MacLellan, Leicester University D MacPherson NEL T J Martin Culham Laboratory, UKAEA D H McLain University of Sheffield, D D M Ogilvie, Edinburgh Regional Computing Centre S Rae BCS Displays Group D Rosenthal Edinburgh CAAD, M Sabin Kongsberg Data Systems T Sancha Cambridge Interactive Systems Ltd M D Sayers Hatfield Polytechnic J M Sykes ICI H Tillotson City University E A Warman Perkins Engines Co Ltd P S Wells BSI D Whitfield, University of Aston, R M Williamson CAD Centre,
Present: D. L. Fisher (Chairman) Leicester University J. Adams Hydrographic Dept., Ministry of Defence. F. R. A. Hopgood Rutherford Laboratory K. Y. Kwok University of Manchester Regional Computer Centre R. C. F. Langridge CAD Centre, Cambridge T. Martin Culham Laboratory D. Rosenthal CAAD, University of Edinburgh M Sabin Kongsberg Data Systems M. D. Sayers Hatfield Polytechnic H. Tillotson City University E. Warman Perkins Engines Co M. Williamson CAD Centre, Cambridge K. W. Brodlie (Secretary} Leicester University
David Fisher, as Chairman, welcomed everyone to the meeting. Apologies for absence were received from:-
J.F. Bell C. Hall A.C. Kilgour D.H. McLain S. Rae
Fisher said that a division would have to be made between members and correspondents, in order to keep paperwork to a reasonable level and allow the speedy circulation of important documents to members. Appendix A contains the proposed list of members; Appendix B the list of correspondents.
A document register will be kept, and the initial register is listed in Appendix C. Documents will be referenced as DPS13/WG5/-.
The minutes of the previous meeting were approved subject to the following corrections:-
(i) Page 4, paragraph 1: The second sentence should be amended to read - David Rosenthal and Sancha strongly disagreed with the wisdom of this assumption.
(ii) Page 5, paragraph 4: The second sentence should be expanded to explain Rosenthal's point clearly:-
Sancha agreed, but Rosenthal strongly opposed the suggestion. He felt that to create a standard for passive 2-D graphics without consideration for graphical input, would almost certainly lead to difficulties when the standard was extended to cover graphical input at some later stage.
Ernie Warman commented that the title of the next Seillac meeting in June 1979 (mentioned on page 4; paragraph 8) would be Workshop on Methodology of Interaction.
Fisher reported on his telephone conversation with Bob Dunn, held during the previous week.
It seems that standards activity in the U.S. is now being organised by two separate bodies - the ACM SIGGRAPH GSPC, and ANSI. Dunn is more concerned now with the latter, and is chairman of ANSI X-3 SPARC Study Group on Computer Graphics Programming Languages.
This ANSI group will meet at SIGGRAPH 78 on 23rd August to preview formal recommendations for a graphics standard - the ACM Core Report is one proposal. Dunn expects the broad principle of a core to be accepted. In the following 30-60 days, comments on the recommendations will be collected. The consolidated comments will be put before ANSI.
Around 1st January 1979, Dunn expects ANSI to agree formally to set up an X-3 committee on graphics standards. He felt the ACM Core Report would be important input as a workable starting point in broad terms.
Dunn's group are currently concerned with matters of principle and broad technical matters - topics such as current position being regarded as fine technical detail. However, they would welcome comments - at worst these would be held pending the launch of the ANSI X-3 committee.
Dunn also mentioned that the ACM SIGGRAPH were developing their own line of progression.
Hopgood asked whether the Americans would be coming to the ISO meeting in Bologna in September with a specific proposal for a graphics standard. Fisher thought they would not be in a position to do this, but confirmed that Dunn would be attending the ISO meeting.
Mike Sayers reported on discussions he had had with Jim Michener in the U.S. on 25th May 1978. The discussions centred on the report of the GSPC Winter Meeting, held at Boulder, Colorado in March. A copy of this report is being circulated to members and correspondents of BSI/DPS 13/WG5 (Reference DPS 13/WG5/4).
Sayers confirmed that GSPC and ANSI seem to be moving in different directions. He ran through the main points of the GSPC Winter Meeting report. The GPC have set up six subgroups to pursue specific technical areas: core refinements; core extensions; core partitioning and protocols; graphics program structures and techniques; higherlevel requirements; and foundations of computer graphics standards. The core refinement subgroup have isolated a large number of issues in the GSPC Core Report to be reconsidered in the light of comments and criticism. Appendix C of the report gives details of these new subgroups. Appendix D gives information on the current ANSI activity.
Sayers felt that the GSPC would have relatively little to report at the ISO meeting in Bologna, and he expected that they would make slower progress now than before.
Fisher commented that the production of the GSPC Core Report had been a considerable achievement, and it was somewhat disappointing to see it being torn apart. Hopgood felt the Core Report was still worthwhile, even though it only provided a checklist for the next effort - many people would have been unhappy if the present Core Report had been agreed as the standard.
The meeting then considered what should be the U.K. contribution to the standards effort. Fisher said that the contribution should be divided into the two areas:-
- general principles - fine details.
Malcolm Sabin felt that the BSI group ought to concentrate on general philosophy, since past proposals had suffered from bottom-up design.
David Rosenthal argued in favour of defining, not a standard graphics package, but a standard pseudo-display. Furthermore, the standard ought to be defined formally, and not informally as in the case of the Core Report. He felt it would be possible to define the syntax, or grammatical structure, of the standard using BNF, and to use ISP (Instruction Set Processor: a hardware description language) to formally describe a standard pseudo-display. Rosenthal's proposal is detailed in his paper Techniques for Formalizing a Graphics Standard (Reference DPS 13/WG5/12), which is being circulated to members and correspondents.
There seemed general support, at least in principle, for Rosenthal's suggestion. Hopgood and Bob Langridge welcomed it as a new approach to graphics standards.
Rosenthal went on to pose four questions which ought to be considered when defining a graphics standard:
Rosenthal noted that the GSPC appeared to have ignored this question.
John Adams felt it would be difficult to settle on the level of pseudo-display interface. He saw the interface steadily shifting as the intelligence of displays grew. Rosenthal envisaged several levels of pseudo-display to cater for the different intelligences of devices.
Sayers agreed in principle with Rosenthal's proposals, but questioned whether manufacturers would be as ready to standardize at the pseudo-display level as at the application program level. Rosenthal felt they would be more likely to standardize at the pseudo-display level. Sabin, speaking from his experience at Kongsberg, felt that manufacturers would be prepared to adopt a standard of the type envisaged by Rosenthal.
Fisher felt it was important that the BSI group should put forward some definite views, both to the ISO meeting in Bologna, and to Bob Dunn's ANSI Study Group. Warman. agreed that the British contingent must have something concrete to put before the ISO meeting. Rosenthal and Hopgood were asked to prepare a paper, developing the ideas in Rosenthal's paper mentioned above (DPS 13/WG5/12). It was agreed that Rosenthal and Hopgood should concentrate on ways of formalizing a standard at the pseudo-display level, and not on particular facilities which should be included in the standard pseudo-display. They were asked to have the paper ready for circulation to members by mid-August. The paper would be discussed at the next meeting of the group, which would decide the British line for the ISO meeting.
It was agreed that there was a need to compare different protocols which are currently in use (for example; GHOST'S GRID file, GINO-F's PSEUDO file). Rosenthal drew attention to the Network Graphics Protocol defined for ARPANET by Sproull and Thomas (Reference: DPS 13/WG5/13). He was presently implementing this protocol at Edinburgh.
Sabin pointed out that his digestion of the literature on software for computer graphics (prepared for the 1976 Seillac meeting) contained a comparison of different protocols; though now out-of-date, it might provide a useful starting point for any new comparison. A copy of Sabin's paper is being circulated to members and correspondents (Reference: DPS13/WG5/9). Of particular relevance is the part on Graphic Stream Formats and Protocols.
Members will be asked to contribute details of the protocols used by the graphical systems at their local sites. This activity will be coordinated by Henry Tillotson, who will circulate members in due course.
Fisher reminded members that all delegates to the next meeting of ISO TC97/SC5/ WG2 must be officially nominated by their country's standards organization. He felt that a U.K. contingent of 3-5 people was appropriate. It was noted that Warman was already nominated as one of two official IFIP representatives at the ISO meeting.
Fisher planned to attend as chairman of this group and past-chairman of the ISO group. Of the others present, only Hopgood was certain that he was free and could raise funds to attend. It was hoped that the BSI could assist with finance to allow one or perhaps two others to attend: in that case, Fisher hoped Tom Sancha would be able to go, and Langridge indicated that he might also be able to attend if given financial assistance.
The BSI representation will therefore be Fisher and Hopgood (definitely), Sancha and Langridge (possibly).
Warman pointed out a number of other meetings in Bologna around the time of the ISO meeting - these are listed in section 10 of the minutes. BSI representatives would be welcome to attend the IFIP WG5.2 meeting as observers.
Ken Brodlie gave a progress report on the NAG graphics chapter. Plans for the chapter have now been approved by the NAG executive. The one outstanding difficulty in the design of the graphical interface (see Graphical Interface for the Proposed NAG Graphics Chapter - Reference DPS13/WG5/6), that of highlighting, has hopefully been resolved and final plans for the interface will be put to the next meeting of the NAG graphics working party at the end of June.
Langridge asked whether users would be encouraged to call NAG interface routines directly. Brodlie replied that the interface routines would probably be available to users, but he anticipated that users would call routines from the underlying graphics package if they wished to add annotation, say, to the plot. Users would have to call routines from the underlying package to set up an appropriate coordinate system. Langridge warned that users could upset NAG library routines by applying certain transformations; Brodlie accepted that users would have to be relied upon to set up an appropriate coordinate system.
Fisher reported on contacts with standards groups in areas related to graphics in general, and to the transmission of graph plot files in particular.
There will be an exchange of documents with the following groups:
In general papers from these three groups will not be circulated.
Fisher reported that he had also tried to make contact with Paul Ellison, University of Exeter, who is chairing an Inter-University Software Committee (IUSC) working group on the transfer of graph plot files. He awaited a reply.
At this point, Sabin gave a short exposition of his views on the transmission of graphical information. The details are not minuted here, as Sabin has since written a position paper on the topic and this is being circulated to members and correspondents (Reference: DPS 13/WG5/10). Sabin's talk could be regarded as a preliminary version of that paper.
There was some discussion on the degree to which DPS 13/WG5 ought to become involved in problems of transmission. It was agreed that questions of file structure were outside the scope of the group. The group should be concerned with the contents of the file to be transmitted, and some members made the point that such details were already being considered in efforts to define a standard pseudo-display. Other members, in particular Sayers, felt there were two separate problems: the compression of graphical information so that it could be sent along a communications line was a topic worthy of consideration in its own right.
Fisher felt that this group ought to concentrate on defining a standard pseudo-display, but it also ought to keep a watch on other standards activities such as the IUSC group. He brought to members' attention the report by the Edinburgh/ Glasgow RCO on a low-level graphics transfer protocol, this report (Reference: DPS 13/WG5/9) is being circulated to members and correspondents.
Rosenthal gave a quick assessment of the graphics system ILP, proposed by Paul ten Hagen and others at Amsterdam. (Reference: DPS 13/WG5/3). Posing the four questions listed in section 4, Rosenthal summed up the system as follows:
As a standard, Rosenthal felt it was too complicated, and being incompatible with FORTRAN, had no chance of being adopted. Hopgood pointed out that ILP was never intended as a proposal for a standard.
Hopgood commented on a point made by Sancha and Rosenthal at the last meeting, namely that graphical input (and output) ought to be mapped onto a character string. Hopgood felt that multiple streams of characters must be considered, and mentioned that the kind of syntax needed to define multiple streams was similar to the syntax needed for study of human behaviour. There have been a number of studies in the AI area where Production Systems have been used to model human behaviour (e.g. Young, Newell, etc. at Carnegie-Mellon) and they are much less dependent on a rigorous ordering of tokens than the standard syntax analyser.
Langridge mentioned that a 2-D graphics package was under development at the CAD Centre. The package will include device independent pseudo-code, and he agreed to bring details to a future meeting.
Sayers mentioned that the BSI group responsible for virtual terminals ought to be kept informed of DPS 13/WG5 activity: in particular they should be aware of the need for multiple input/output channels.
The next meeting will be held at the University of Leicester on Monday 11th September 1978. The contribution from Rosenthal and Hopgood will be considered, and the meeting will act as a briefing session for BSI delegates to the ISO meeting in Bologna.
21st - 25th August 1978 - SIGGRAPH 78, Atlanta, Georgia.
11th September 1978 - Meeting of BSI DPS 13/WG5, Leicester.
18th - 20th September 1978 - Three-day tutorial course on State-of-the-Art Computer Graphics,
Bologna, Italy.
2lst - 23rd September 1978 - Interactive Techniques in CAD, ACM Conference,
Bologna, Italy.
24th - 26th September 1978 - IFIP WG5.2 working meeting: "Successes and failures in CAD",
Bologna, Italy.
27th - 29th September 1978 - Meeting of ISO TC97/SC5/WG2, Bologna, Italy.
28th September 1978 - IMA conference on Mathematical methods in computer graphics
and design, Leicester.
June 1979 - Workshop on Methodology of Interaction, Seillac.
Secretary Ken Brodlie, 19th June 1978
J Adams Hydrographic Department, MOD K W Brodlie, Leicester University D L Fisher Leicester University F R A Hopgood ACD, Rutherford Laboratory K Y Kwok University of Manchester Regional Computer Centre, R C F Langridge CAD Centre, Cambridge. D MacPherson NEL T J Martin Culham Laboratory, UKAEA D Rosenthal Edinburgh CAAD, M Sabin Kongsberg Data Systems T Sancha Cambridge Interactive Systems Ltd M D Sayers Hatfield Polytechnic H Tillotson City University E A Warman Perkins Engines Co Ltd
J F Bell ADCD, Ordnance Survey R W Breward MCERE, P Dimmer, Leicester University D Fildes University of Glasgow B Ford, N.A.G. C Hall NPL A Kilgour University of Glasgow G D S MacLellan, Leicester University D MacPherson NEL (and D Bradley) D H McLain University of Sheffield, D D M Ogilvie, Edinburgh Regional Computing Centre S Rae BCS Displays Group J M Sykes ICI Paul ten Hagen CWI B Walker Ferranti Cetec Graphics Ltd H Tillotson City University P S Wells BSI D Whitfield, University of Aston, R M Williamson CAD Centre
1. GSPC CORE Report, ACM/SIGGRAPH "Computer Graphics", vol.ll, No.3, Fall 1977 - members asked to acquire their own copy by their own means.
2. Minutes of the meeting of "ISO TC97/SC5/WG2 held in Toronto in August 1977.
3. ILP, Intermediate Languages for Pictures, T. Hagen et al, Oct. 1977 - available from DLF on demand.
4. Report of the GSPC Winter Meeting, R. Heilman and B. Herzog, March 1978.
5. Minutes of 1st Meeting of DPS/13/WG5 held on 6th April 1978.
6. Graphical Interface for the NAG Graphics Chapter, K.W. Brodlie, April 1978.
7. RCO Plotter File Standard, D. Rischmiller, 1977.
8. Minutes of 2nd Meeting of DPS/13/WG5 held on 31st May 1978.
9. Software Interfaces for Graphics, M. Sabin, May 1976.
10. Issues Relevant to the Evaluation of Picture-Description Codes, M. Sabin, June 1978.
11. Screen Management, P.R. Brown, DPS/13/WG3, May 1978.
12. Techniques for Formalising a Graphics Standard D.S.H. Rosenthal, May 1978.
13. A Network Graphics Protocol, R.F. Sproull and E.L. Thomas, Computer Graphics, vol.8, No.3, Fall 1974.
14. Papers for ISO TC97/SC5/WG2 meeting in Bologna (Sept. 1978) from ten Hagen dated 2nd June 1978.
Present: D. L. Fisher (Chairman) Leicester University D. Bradley Leicester Polytechnic P. Dimmer Leicester University D. A. Duce Rutherford Laboratory K. Y. Kwok University of Manchester Regional Computer Centre D. Rosenthal CAAD, University of Edinburgh M Sabin Kongsberg Data Systems M. D. Sayers Hatfield Polytechnic H. Tillotson City University K. W. Brodlie (Secretary} Leicester University
Apologies for absence were received from:-
R. Brown, R. A. Earnshaw, R. Langridge, and E. Warman.
The minutes of the previous meeting were approved.
Robert Brown has replaced John Adams as representative of the Hydrographic Dept. on the working group. R. A. Earnshaw (Leeds University) and Robin Forrest (University of East Anglia) have joined the group as member and correspondent respectively. Roger Hubbold (University of Manchester Computer Graphics Unit) has been invited to join the group, but is currently in the U.S.
Tom Sancha's new address is given in Appendix A.
The Secretary was asked to collect the home telephone numbers of members of the group.
All documents circulated to the group from now on will be given a yellow front sheet to aid identification.
Malcolm Sabin indicated that a copy of any document in his bibliography (DPS13/WG5/9) can be obtained from him on request.
The current document list is, included as Appendix C.
David Fisher reported that Tom Sancha had been unable to reach agreement with the BSI regarding expenses for his visit to Bologna for the meeting of ISO TC97/SC5/WG2, and had decided therefore not to go. Fisher proposed that David Rosenthal should go in place of Sancha, and this suggestion was approved by the meeting. The BSI representation in Bologna will be Fisher, Rosenthal and Bob Hopgood.
Fisher distributed a copy of an ISO working document giving a glossary of computer graphics terms . He asked for comments on the document to be sent to him by 13th November 1978. A copy of the document (Reference DPS13/WG5/17) is enclosed for those not present at the meeting.
Fisher read out a letter from Bob Langridge in which he suggested that the Core Report had failed to consider the question of code economy. It is clearly desirable to use the intelligence of a device to produce graphics facilities, but the true value is only gained if one can avoid loading simulation software for these facilities. Langridge enclosed a working paper describing a possible solution to the question and a copy is enclosed with these minutes. (Reference DPS13/WG5/20).
Fisher said that the Core Report had made a deliberate attempt to exclude hardware frills. A list of essential hardware facilities had been drawn up, and it was intended that other facilities should be simulated in software. He agreed this could be expensive, and felt that Rosenthal's idea of a standard pseudo-device was one solution, with the simulation code being placed in the device driver.
Malcolm Sabin commented that not only hardware designers included frills which were easy to implement, but software designers were just as guilty. He felt it would be valuable if Langridge explained how code economy affected the design of a graphics package. (Some discussion is included in section 5 of Langridge's paper - Secretary).
Rosenthal presented his paper Techniques for Formalizing a Graphics Standard, which was circulated vith the agenda for the meeting. (Reference DPS13/WG5/12 - replacing an earlier version of the paper).
Having attempted to write formal descriptions of a Calcomp 563 plotter and a Tektronix 4012 storage tube (Appendices A & B], Rosenthal now felt that ISP was perhaps unsuitable as a formal description language, since it produced a rather illegible description of devices. Also there appeared to be no formal definition of ISP itself. APL had been suggested as an alternative, but it seemed likely to produce an equally illegible description. Concurrent PASCAL was a possibility, and would allow the description of parallel events (i.e. input/output). He still had to decide on the most suitable description language.
Rosenthal explained the formal descriptions of the plotter and storage tube. Writing a description of the plotter had been straightforward, but the storage tube had been much harder. In particular, he had found no way of describing the erase action in ISP.
Rosenthal felt that one important issue had emerged from his work: there was a need to establish criteria for comparing one picture against another, in order to validate different implementations of a standard. He suggested that comparison might be made at the level of an array of pixels, rather than at the level of line drawing.
Sabin felt that Rosenthal had made useful progress, and added some comments on the paper. He felt that the mapping defining each procedure call (page 3, section 4.2) ought to be formally defined. Rosenthal commented that he felt there was no need for a standard graphics package, just a standard pseudo-code.
Fisher thanked Rosenthal for his work and encouraged him to continue his efforts.
Sabin presented his paper Issues Relevant to the Evaluation of Picture-Description Codes, circulated with the minutes of the previous meeting (Reference DPS13/WG5/10).
There was a discussion on a number of issues. In particular there was some opposition to the suggestion that relative co-ordinates were unnecessary - it was felt that these could be useful for data compression. On the question of legible format, Rosenthal felt it was important to separate the notion of tokens in a protocol from their representation in any particular context, say as ASCII printing characters.
Mike Sayers said that the wide range of graphical applications gave another reason for standardizing at the pseudo-code level - it would allow different graphics packages for different applications.
Sabin summarized his paper by suggesting that it would provide a useful check-list for the evaluation of any proposed protocol.
Ken Brodlie gave a summary of a paper just received from the German standards group - Implementation of the Core Graphics System GKS in a Distributed Graphics Enviromnent (Reference DPS13/WG5/18). Copies are enclosed for those not present at the meeting.
Brodlie commented that GKS vas essentially a standard for 2-D passive graphics, although the implementation described in the paper also included input functions based on the GSPC Core Report. The Germans had selected a minimal set of output primitives (polyline, text and marker), with attributes of primitives being handled in an interesting way. At the start of a program, the user associates attributes of colour, linestyle, line-width and intensity with various pen numbers. A change of attribute in the course of drawing is achieved by selecting a different pen.
Fisher was asked to obtain a copy of Functional Description of the Graphical Core System GKS as a Step towards Standardization, by Wisskirchen et al, which gives a fuller description of the GKS system.
Henry Tillotson reported that he had received about 20 replies so far to his survey. Some replies were quite detailed, others brief. An assessment of the replies will be made in due course.
Sayers mentioned that the GINO-F Users Group had carried out a survey of devices, and agreed to pass information to Tillotson.
The next meeting will be held at City University, on Monday 30th October, when reports from Bologna and possibly SIGGRAPH'78 vill be heard.
21st September, 1978. Ken Brodlie Secretary.
D. Bradley Leicester Polytechnic K W Brodlie, Leicester University R Brown Hydrographics Dept, MOD R A E Earnshaw University of Leeds D L Fisher Leicester University F R A Hopgood ACD, Rutherford Laboratory K Y Kwok University of Manchester Regional Computer Centre, R C F Langridge CAD Centre, Cambridge. T J Martin Culham Laboratory, UKAEA D Rosenthal Edinburgh CAAD, M Sabin Kongsberg Data Systems T Sancha Cambridge Interactive Systems Ltd M D Sayers Hatfield Polytechnic H Tillotson City University E A Warman Perkins Engines Co Ltd
J F Bell ADCD, Ordnance Survey R W Breward MCERE, P Dimmer, Leicester University D Fildes University of Glasgow B Ford, N.A.G. A R Forest University of East Anglia C Hall NPL A Kilgour University of Glasgow G D S MacLellan, Leicester University D MacPherson NEL (and D Bradley) D H McLain University of Sheffield, D D M Ogilvie, Edinburgh Regional Computing Centre S Rae BCS Displays Group J M Sykes ICI Paul ten Hagen CWI B Walker Ferranti Cetec Graphics Ltd H Tillotson City University P S Wells BSI D Whitfield, University of Aston, R M Williamson CAD Centre,
15. Description of device independent code produced by GINO-F codegenerators SAVDRA and SAVPIC, CAD Centre.
16. GPGS - 16 bits device independent picture code, D. Groot, 0ct. 1975.
17. ISO Vocabulary for Information Processing, Section 13: "Computer Graphics and Computer Micrographics, (ISO/TC97 /SCl 655), June 1978.
18. Implementation of the Core Graphics System GKS in a Distributed Graphics Enviromnent, P. Wisskirchen et al., July 1978.
19. Minutes of 3rd meeting of DPS13/WG5 held on 11th September 1978.
20. Basic 2-D Graphics System Interface for Software Systems Development, R. Langridge, 1978 (CADC WP/95/78).
Present: D. L. Fisher (Chairman) Leicester University D A Duce Rutherford Laboratory R A Earnshaw Leeds University F R A Hopgood Rutherford Laboratory K. Y. Kwok University of Manchester Regional Computer Centre R C F Langridge CAD Centre Cambridge D. Rosenthal CAAD, University of Edinburgh M Sabin Kongsberg Data Systems M. D. Sayers Hatfield Polytechnic H. Tillotson City University K. W. Brodlie (Secretary} Leicester University
The Chairman welcomed Ray Earnshaw to his first WGS meeting.
Apologies for absence were received from A. C. Kilgour.
The minutes of the previous meeting were approved.
Bob Langridge asked for a reference to Concurrent PASCAL, mentioned in section 5 of thThe programming language Concurrent PASCAL, IEEE Trans. Software Engineering, Vol. SE1, pp.199-207. See also Software Practice and Experience, Vol. 6, 1976, pp.141-149 - Secretary).
David Fisher dealt first with the ISO vocabulary on Computer Graphics and Computer Micrographics (Reference DPS 13/WG5/17). The ISO graphics working group at its recent meeting in Bologna had voted against the document, as the terms were out of date and the vocabulary was badly structured. Fisher proposed that the BSI group also vote against the document, and this was agreed.
Fisher then reported in detail on the ISO Bologna meeting. There had been no French representation at the meeting, and only one American, Robin Williams - and he only for one day. The Germans, therefore, made a particularly strong impression, and they stated that they expected to have a draft standard ready for circulation by Christmas 1978. Their plan was to define a national graphics standard, and also to put it forward as an international standard.
The ISO group have decided to set up an Editorial Board to review standard proposals. Membership of this board will be limited to those with a standard either already in existence or in the pipeline. This would seem to limit membership to the Americans, Germans and Dutch (as chairman). However, the Norwegians plan to submit a modified version of GPGS which would ensure their representation on the committee. Bob Hopgood suggested at the meeting that the BSI group might put forward a modified version of GINO-F as its proposal for a standard, and this in turn would ensure a UK voice on the board.
David Rosenthal's talk was well received and it was clear that the Germans were taking far more account of communications within the graphics package than the US Core Report had done.
Minutes of the ISO meeting are enclosed.
Discussion centred on Hopgood's suggestion, made at the ISO meeting, that a modified version of GINO-F be submitted as a proposed standard.
Mike Sayers said that he detected a feeling at the CAD Centre that a successor to GINO-F ought to be considered, and that the GINO-F Technical Committee ought to be reconvened.
Fisher asked Langridge where his paper (DPS 13/WG5/20) fitted into this picture. Langridge said his new package was aimed at systems development, and was still at an experimental stage. On the question of a successor to GINO-F, he felt there was some resistance at CAD Centre to change as GINO-F itself cost more money than it brought in. However the Centre might respond to pressure from outside.
Fisher agreed to write to CAD Centre on behalf of WG5, asking for effort to be put into a GINO-F successor. He also agreed to urge the GINO-F Users Group to put pressure on CAD Centre for change.
Sayers felt a successor to GINO-F was definitely needed, but he suspected that the problem among GINO-F users was the fact that it was people without money, the academics, who were urging change, and that industrial users were keen for the status quo to be maintained.
Hopgood indicated that if CAD Centre put some effort into a GINO-F successor, the Rutherford Lab. might be able to add a corresponding effort. The Rutherford would not be prepared to tackle the project on its own. He felt a new version of GINO-F as most definitely needed, as the current version was unable to take advantage of new graphics devices which were coming on to the market.
Fisher mentioned that Culham had begun to write a new version of GHOST. He agreed to contact John Prior at Culham in an effort to bring GHOST changes and additions into line with current thinking in the standards field. The possibility of getting representatives from CAD Centre and Culham together was mentioned. Rosenthal felt that if Culham were to increase the interactive capability of GHOST, the German input proposals should be studied.
Langridge presented his paper Basic 2-D Graphics System Interface for Software Systems Development. He felt the main improvement suggested in the paper was a restructuring so as to take advantage of device facilities without also loading simulation software in the host computer. The package interface did not refer to specific device facilities - all implementation decisions were taken at device driver level. The package had been implemented at CAD Centre: its total size was about half the corresponding size in GINO-F.
Sabin asked if Langridge was pushing the interface between device independent and device dependent parts of the package upwards. Langridge agreed and noted that the structure of device driver routines had to be altered to allow sharing of conman pieces of code.
Sabin noted that when transmitting graphical data from one computer to another, Langridge's proposal pushed more work to the receiver of the data. Sayers felt there could be several levels of protocol within the structure. Hopgood agreed there ought to be at least one level lower down, in addition to the high level interface proposed by Langridge.
Discussion turned briefly to the internal structure proposed by the Germans. It was thought that their plan was to have a state table, rather like a telephone switchboard. Device drivers were formed by collecting together a suitable set of modules and entering the linkages in the state table. For clarification of these ideas, the German standards proposal was eagerly awaited, and Fisher agreed to circulate the document as soon as it became available.
Hopgood asked Langridge further questions on his paper. The value of logical device co-ordinates, Langridge explained, was to allow the user to work in physical units, e.g. millimetres. Image transformations referred to transformations generated at a terminal - the problem of returning cursor co-ordinates was noted.
Ken Brodlie reported that the NAG graphics working party had finally agreed on a set of interface routines. The.outstanding problem of highlighting had been resolved by allowing the user to set up a number of conceptual pens, with attributes of colour, linestyle and text font. Highlighting within a routine would be achieved by a change of pen.
The working party was now concerned with the problem of validating graphical routines, and Brodlie felt this was an area where WG5 might be able to give some advice. One idea currently being examined was to have a version of the interface routines which printed out the routine name and its arguments; this might allow some form of automatic checking. It seemed likely, however, that comparison of pictures by eye would still play a major part in validation.
Henry Tillotson reported on his survey. Sixteen replies had been received, but the variation in replies indicated that his questionnaire had been too vague. He agreed to write up the results he had obtained.
Members of the ISO working group have been added to the BSI mailing list. Comments on the minutes of the last BSI meeting, and on Rosenthal's paper (DPS 13/WG5/12) have been received from Jim Michener, a member of GSPC. A copy of the letter (Reference DPS 13/WG5/22) is included with these minutes.
The meeting discussed the points raised by Michener. The first comment on the minutes was challenged by Fisher - he stuck to his statement. The second comment was accepted, but the third was opposed. Rosenthal emphasized that he was not against user interface standardisation, but felt there must be further levels of standardisation between the user interface and the actual pictures. The fourth comment was accepted - indeed the European feeling has been along these lines for a long time.
In reply to the second comment on his paper, Rosenthal said he was fundamentally opposed to Michener and the GSPC view on validation. There was no way in which a package could be verified by regarding it as a black box. While it might be impossible at present to compare actual pictures, the aim should be to certify as much as possible of the process. There should be at least one standard interface between the user interface and a picture.
Hopgood agreed and felt that the GSPC had a naive view of graphics standards. The Germans and British were in agreement over the matter of validation.
Despite the group's disagreement with some of the comments, Fisher expressed the group's appreciation of Michener's letter.
The next meeting was arranged for Monday 22nd January at the City University, starting at 11.00 a.m. Two major topics for the agenda would be consideration of the German standards proposal, and future effort on GINO-F.
20th November, 1978. Ken Brodlie Secretary.
D. Bradley Leicester Polytechnic K W Brodlie, Leicester University R Brown Hydrographics Dept, MOD D A Duce ACD, Rutherford Laboratory R A E Earnshaw University of Leeds D L Fisher Leicester University F R A Hopgood ACD, Rutherford Laboratory K Y Kwok University of Manchester Regional Computer Centre, R C F Langridge CAD Centre, Cambridge. T J Martin Culham Laboratory, UKAEA D Rosenthal Edinburgh CAAD, M Sabin Kongsberg Data Systems T Sancha Cambridge Interactive Systems Ltd M D Sayers Hatfield Polytechnic H Tillotson City University E A Warman Perkins Engines Co Ltd
J F Bell ADCD, Ordnance Survey K Bo RUNIT, Trondheim R W Breward MCERE, J Butland University of Bradford R L Crowhurst Directory of Standardisation, MOD P Dimmer, Leicester University R M Dunn DRSEL-GG-C U.S. Army Electronic Command J Encarnacao technical University Darmstadt D Fildes University of Glasgow B Ford, N.A.G. A R Forest University of East Anglia R A Guedj Thomson-CSF C Hall NPL B Herzog University of Colorado R Hubbold Computer Graphics Unit, Manchester A Kilgour University of Glasgow G D S MacLellan, Leicester University D MacPherson NEL (and D Bradley) D H McLain University of Sheffield, J C Michener Intermetrics Inc, Cambridge, Mass D D M Ogilvie, Edinburgh Regional Computing Centre W A Prior Culham Lab, UKAEA S Rae BCS Displays Group J M Sykes ICI Paul ten Hagen CWI B Walker Ferranti Cetec Graphics Ltd H Tillotson City University P S Wells BSI D Whitfield, University of Aston R Williams IBM, San Jose R M Williamson CAD Centre P Wisskirchen GMD, St Augustin
21. Minutes of 4th Meeting of DPS13/WG5 held on 30th October 1978.
22. Comments on the Minutes of the 3rd Meeting of DPS13/WG5 and Document DPS13/WG5/12 by J.C. Michener.
23. Minutes of ISO/TC97/SC5/WG2 - GRAPHICS Meeting held in Bologna on 27-28th September 1978.
Present: D. L. Fisher (Chairman) Leicester University D A Duce Rutherford Laboratory R A Earnshaw Leeds University R Ford Compeda W A J Prior Culham Lab D. Rosenthal CAAD, University of Edinburgh M Sabin CAD Centre, Cambrisge M. D. Sayers Hatfield Polytechnic H. Tillotson City University K. W. Brodlie (Secretary} Leicester University
Apologies for absence were received from Bob Hopgood, Bob Langridge and Ernest Warman.
The minutes of the previous meeting were approved and there were no matters arising.
In an effort to keep the mailing list under control, correspondents had been asked to confirm that they wished to remain on the WG5 mailing list. As a result of this exercise, it was proposed to remove the following, who had al1 failed to reply, from the list of correspondents:
Mrs. J. Butland (Bradford University) P. Dimmer (Leicester University) Dr. B. Ford (NAG) C. Hall (NPL) G.D.S. MacLellan (Leicester University) D.H. McLain (Sheffield University) D.D.M. Ogilvie (Edinburgh RCC) S. Rae (BCS Displays Group) J.M. Sykes (ICI) R.M. Williamson (Cambridge Interactive Systems)
Each person would be informed of the decision.
The following are to be added to the correspondents' list.
Dr. A.M. Addyan .(Manchester University)
Dr. M. Bellardinelli (Italy)
Dr. G.M. Bull (Hatfield Polytechnic)
Prof. A. Luehrmann (USA)
David Fisher recapped on the situation at the time of the last meeting, on 30th October. ISO had set up an Editorial Board to review standards proposals, and its membership was to be limited to those countries with a graphics standard already formulated or in the pipeline. It seemed that the UK might not have a voice on such a board, and therefore Bob Hopgood had suggested that a UK proposal, based on GINO-F or a development of it, should be drawn up. The Rutherford Laboratory would be prepared to develop GINO-F in conjunction with CAD Centre.
Fisher had been asked to approach the CAD Centre, the NCC and the GINO-F Users Group in an effort to obtain support for such an idea. Almost unanimous support was forthcoming from the Users Group, and an encouraging letter bas just been received from the CAD Centre. This indicates that, while they are extremely busy for the next three months, they would subsequently be able to devote some effort. Correspondence with the NCC had been less useful - in particular their suggestion that the BSI should object to the formation of the Editorial Board was unhelpful.
In any case, the imediate problem has now been resolved, in that Paul ten Hagen has invited Bob Hopgood to co-chair the Editorial Board, and so the UK will be represented. However, Fisher still welcomed the promise of effort from the CAD Centre.
Fisher reported that he had been invited to attend the GSPC Winter Meeting at Boulder, Colorado. He took this as an invitation to the BSI group as a whole, and felt it was very important that the UK should be represented at this meeting. Mike Sayers indicated that he might be able to raise the finance and the spare time to allow him to attend. (It is now confirmed that Sayers will represent the BSI group at Boulder - Secretary).
Ernie Warman has suggested that the EEC may have some money to spend on the promotion of standards efforts. If so, this could raise the level of activity in Europe, and Fisher asked members to consider ways in which money could be put to good use. Two possibilities are the arrangement of a week-long meeting where some useful work might be done, and the invitation of German standards workers to spend time in the UK (and possibly instal their GKS at some site). David Rosenthal suggested that a suitable topic for a week-long workshop would be the certification of graphics standards. Malcolm Sabin urged that some of the money be set aside for travel expenses.
The meeting discussed the new German standard proposal GKS, and compared it with the GSPC Core. The following summary of the BSI group's joint views will be sent to Paul ten Hagen as co-chairman of the ISO Editorial Board, together with personal contributions from Bob Hopgood, David Rosenthal, Ken Brodlie, and perhaps others. Hopgood's paper was distributed at the meeting.
It was agreed that the major difference between the two GKS file. This almost certainly influenced many of the other decisions taken in GKS, and many of the differences between the two proposals can be traced back to this single major distinction. The idea of maintaining a display file was generally welcomed.
The following other points were agreed:
There were mixed views on the merits of a separate clipping and window rectangle as in GKS, with a small majority in favour.
(Note: Copies of the above minute were sent to Paul ten Hagen and Jose Encarnacao.)
Individual contributions towards the comparison of the Core and GKS have been received from Ken Brodlie, Bob Hopgood and David Rosenthal, and these are included as Appendices to these minutes.
Rosenthal presented his position paper for the Seillac meeting on Methodology of Interaction (DPS13/WG5/26). He explained that he was still formulating his ideas on the subject, and indeed had modified some of his views since writing the paper.
Most of the comments came from Malcolm Sabin. His main points were:
Rosenthal has summed up his current thinking as aiming to put forward a model of graphics I/O processing combining:
The following diagrams show the flow of information between modules in the Core system, and an alternative system in line with Rosenthal!s ideas.
(Note: I am grateful to David Rosenthal for sending me the above diagrams and summary of his current line of thought - Secretary).
Copies of the paper Graphics with SBASIC (DPS13/WG5/27) were distributed to those present. This paper was received from Dr. G. Bull (Hatfield Polytechnic), and gives an indication of the work being done by those developing a standard for the BASIC language. The current plan is to add a set of graphics statements to the language.
There was unanimous agreement that the BASIC language ought to use any ISO graphics standard that emerges. However there were two views on how graphics ought to be incorporated: some felt that there should be special plotting statements in the BASIC language (as currently proposed), others felt that a single set of graphics modules ought to be accessible from any programing language.
It was noted however that there was a need for simplicity in the graphics standard, to encourage its use from languages such as BASIC.
The next meeting was scheduled for 15th March 1979, at the CAD Centre, Cambridge.
ACM Computing Surveys, Vol. 10, No 4, 1978.
This journal contains a number of articles describing various aspects of the GSPC Core.
Computer Graphics, vol. 12, No 4, 1978.
This journal contains three articles of interest: one by Vinberg criticising the Core, and two others describing different implementations of the Core.
SIGGRAPH'79, Chicago, 6th-1Oth August, 1979.
EUROGRAPHICS 79, Bologna, Italy. 25th-27th October 1979.
Note: The Secretary has a machine-readable version of the GKS standard document (DPS13/WG5/25) and can supply a copy to anyone interested.
D. Bradley Leicester Polytechnic K W Brodlie, Leicester University R Brown Hydrographics Dept, MOD D A Duce ACD, Rutherford Laboratory R A E Earnshaw University of Leeds D L Fisher Leicester University F R A Hopgood ACD, Rutherford Laboratory K Y Kwok University of Manchester Regional Computer Centre, R C F Langridge CAD Centre, Cambridge. W A J Prior Culham Laboratory D Rosenthal Edinburgh CAAD, M Sabin Kongsberg Data Systems T Sancha Cambridge Interactive Systems Ltd M D Sayers Hatfield Polytechnic H Tillotson City University E A Warman Perkins Engines Co Ltd
A J Addyman University of Manchester J F Bell ADCD, Ordnance Survey M Bellardinelli Italy K Bo RUNIT, Trondheim R W Breward MCERE, G M Bull Hatfield Polytechnic R L Crowhurst Directory of Standardisation, MOD R M Dunn DRSEL-GG-C U.S. Army Electronic Command J Encarnacao technical University Darmstadt D Fildes University of Glasgow B Ford, N.A.G. A R Forrest University of East Anglia R A Guedj Thomson-CSF B Herzog University of Colorado R Hubbold Computer Graphics Unit, Manchester A Kilgour University of Glasgow A Luehrmann Univ California at Berkeley D MacPherson NEL (and D Bradley) T J Martin Culham Laboratory J C Michener Intermetrics Inc, Cambridge, Mass Paul ten Hagen CWI B Walker Ferranti Cetec Graphics Ltd P S Wells BSI H Tillotson City University P S Wells BSI D Whitfield, University of Aston R Williams IBM, San Jose P Wisskirchen GMD, St Augustin
24. The AGF Plotfile - Towards a Standardization for Storage and Transportation of Graphics Information, G. Enderle, I. Giese, M. Krause, H.P. Meinzer, November 1978.
25. Graphical Kernel System (GKS) Functional Description (DIN Graphics Standard Proposal), December 1978.
26. Methodology of Interaction - Position Paper, D.S.H. Rosenthal, December 1978.
27. Graphics with SBASIC, W.Y. Arms, January 1979.
28. Minutes of the 5th Meeting of DPS13/WG5 held on 22nd January 1979.
29. A Formal Framework applied to Graphics Standards (Proposals), R. Eckert, February 1979.
Both standard proposals specify a set of basic functions for a graphics system, independently of graphical output device, programming language or application.
Main points are:
Some particular areas are compared in more detail in the following sections.
Both the Core and GKS have similar routines to initialize and terminate the graphics system. The Core is structured into levels and so the user has to specify via the initialization routine the level of the system he is using. GKS allows a user-defined error procedure to be specified at initialization.
The Core supports just one work station, but that work station may have several view surfaces, any number of which may be selected at any one time. GKS supports a number of work stations each with one view surface, but only one work station can be active at any time.
In GKS, SET REQUESTED DISPLAY SPACE lets the user define a subrectangle, relative to the actual display space, for each work station. A paper quality can also be defined for plotters. The Core has NDC SPACE 2 which specifies a maximum width and height in normalised device co-ordinates and which applies to all view surfaces. The GKS scheme seems more flexible, but it is not clear if the normalised device co-ordinate space on a work station is in terms of the actual or requested display space - presumably it is the requested space. Also it is not clear if the normalisation is such that the maximum or minimum dimension is 1.0.
Both systems have a window-viewport mapping.
The Core allows the user-space to be rotated with respect to device-space (via the VIEW UP 2 vector) before the window is applied. There is no corresponding function in GKS.
Both allow clipping, but in the Core there is simply an on-off switch for window clipping. In GKS, the user can specify a clipping rectangle within the window. This is a big improvement because it separates the notions of mapping and clipping.
GKS allows the user to maintain a display file, the so-called GKS file. Segments written to this file can be recalled either for insertion in another segment, or for delivery to the application program, presumably for long-term storage. Likewise a segment can be passed from the application program to the GKS file.
As far as can be seen, inserted segments lose their own identity and assume the attributes of the segment into which they are inserted. Thus the GKS INSERT function is simply a shorthand for a sequence of output primitives. Ihis needs to be made clearer in the definition of GKS.
The Core does not maintain a display file, and thus the insertion of segments is not a possibility.
Both systems have a function to delete a segment, but only the Core has a function to rename a segment.
Both systems have a function to transform segments at a work station:
TRANSFORM SEGMENT in GKS, and IMAGE TRANSFORMATION in the Core.
The Core allows primitives within a segment to be detected using the PICK ID attribute. GKS has no such function.
Segment attributes of visibility, detectability and highlighting seem similar in the two systems.
In the Core, all attribute setting is done within segments. In GKS, colour, intensity, linewidth and linestyle cannot be specified directly within a segment. Instead the user must define a set of conceptual pens outside a segment, these pens having attributes of colour, intensity, etc., and any change within a segment is achieved by pen selection. This is a good scheme because it allows segments to be defined in a device-independent manner.
In both systems, character attributes are specified within segments. The attributes are similar - size, spacing, font and quality. One small difference is that in GKS high quality text is subject to the current line attributes (i.e. linestyle, linewidth) - not so in the Core. (Incidentally it could be argued that text font and quality should also be associated with a pen and specified outside a segment).
GKS has no CP, while the Core has one CP. As a result the GKS POLYLINE primitive replaces six Core functions MOVE, LINE, POLYLINE in absolute and relative coordinates.
The definition of a marker is similar in both systems - i.e. device dependent and without size or quality. In my view this is wrong - markers should have size and quality attributes.
GKS has a DRAW function to act as a generalised output primitive. There is no corresponding function in the Core. Probably a function such as DRAW is best specified in a higher level standard.
This paper does not attempt to give a full comparison of the two systems. One reason is that the degree of completeness in terms of the definition differs considerably. For this reason, precise comments can often be made about one system and not the other. Also, it is possible to interpret both systems in a variety of ways. A larger set of examples showing implementation strategies on a variety of devices (particularly storage tubes) would be welcome.
Both systems are defined on a level rather than module basis. Consequently, a storage tube or plotter user has to carry the full system. GSPC have shown signs of going away from the level approach. In the DIN proposal, there is no method of using it without implementing segments in some way. The DIN proposal is approximately levels 1 to 3 of the GSPC proposal but omitting 3D. The lack of any subdivision of the DIN proposal may be a criticism.
The DIN proposal has made the major change of defining no current position and insisting that all primitives are defined in absolute coordinates in the user coordinate system. The problems experienced with current position indicate that this is a sensible way to go and is preferable to the GSPC approach.
As a result of this decision, many of the GSPC primitives translate to a single DIN primitive. For this reason, there is little difference between the proposed primitives in the two systems. Both include polyline. The DIN proposal does include a generalised primitive identifier which allows a whole series of higher level primitives. It is interesting to note that these include area fill as well as more complex arc drawing primitives. It is important that these additional primitives are well defined. No information is given concerning these and it is suggested that they may be part of a separate standard. Would there be one or possibly more than one standard for these higher level primitives. Are they seen as derivatives from polyline or separate primitives with their own attributes. Is it possible to define new attributes for them.
The DIN proposal does not allow the concatenation of text and would expect this to be done by buffering prior to output if even spacing of low quality text is required.
Both systems define colour, intensity, line type, thickness, font, char size as primitive attributes of some kind. The DIN proposal appears to define all but character attributes as segment attributes. If this is true, it is likely to force the proliferation of segments with the corresponding overheads. The DIN proposal has a pen number attribute which implies an open-ended set of device dependent options. There is insufficient detail to make any comment.
Major difference between the two proposals is the lack of a PICK-ID in the DIN proposal. This means that each light button of a menu will reside in separate segments with the consequent overheads which, in the DIN proposal, appear large. This is a fundamental omission which needs to be assessed carefully. Most existing systems which lack this have found a need for it.
The major difference between the two systems is the introduction of a central file of segments in the DIN proposal. It allows segments to be manipulated and output to a number of devices using this facility and the INSERT function.
The DIN proposal specifically allows the user to control the flow of traffic from system to device. Renaming and appending to segments is presumably done by using the INSERT facility in the DIN proposal.
Both proposals provide image transformations of individual segments. This goes against the concept of divorcing picture drawing from modelling. For example, see the DIN test program in their document.
The DIN proposal is only 2D and consequently much simpler. There may be problems in extending to 3D if compatibility with 2D is to be retained. The DIN proposal has a separate clipping window from the viewing window.
Both systems have the same 5 input functions. The DIN proposal does not have the concept of an event and a sampled device. Instead a request is made for a specific number of inputs fro a particular device and only one device can be active at a time. This seems to be far too restrictive and will certainly cause problems. It severely restricts the operator's method of working.
Valuators in the DIN proposal are restricted to the range (0,1) for no apparent reason.
There is no minimal set of available input devices in the DIN system. Consequently, many device dependent features will have to be implemented in the application program. The GSPC approach of specifying a minimal set always supported seems preferable.
The GSPC proposal allows some connection between input and echoing. Most of the DIN proposals see this as device dependent. Consequently, the operator may get a significantly different view of the system on different devices.
The following is an attempt to identify the semantic differences between the graphics pseudo-packages specified in the ACM-SIGGRAPH Core Report and the DIN Proposal in the area of segmentation. It consists of a description of this area of each package in a compatible format, and a comparison of the two.
In GKS all segments have names, and the segment associated with a particular SEGMENT_NAME can be in one of five distinct states:
1. Non-existent. 2. Open. 3. "On" at least one work station. 4. "In" the GKS file. 5. "On" a work station and "In" the GKS file.
The GKS description is not completely clear about multiple work stations; the following assumes only a single work station.
Transitions between these states are caused by applications of the following GKS functions:
A. OPEN_SEGMENT B. CLOSE_SEGMENT C. DELETE_SEGMENT D. RESET_ GKS_FILE E. CLOSE_WORK_STATION F. GENERATE_GKS_FILE_SEGMENT_FROM_RECORD G. REQUEST_NEW_DISPLAY_SPACE
The transitions between states caused by each function are set out in the following table:
TO STATE
1 2 3 4 5
s 1 - A - F F*
F T 2 - - B B B
R A 3 CEG - - - -
0 T 4 CD - - - -
M E 5 C - D EG -
* I assume that function GENERATE_GKS_FILE_SEGMENT_FROM_RECORD in state "Work Station + GKS File Active" results in a work station segment as well as a GKS file segment, the document is not clear on this point.
In the Core, segments are of two main types. Non-retained segments have no names, but the retained segment associated with a particular SEGMENT NAME can be in one of three distinct states:-
1. Non-existent 2. Open 3. Closed
Transitions between these states are caused by application of the following core functions:
A. CREATE_SEGMENT B. CLOSE_SEGMENT C. DELETE_SEGMENT D. DELETE_ALL_SEGMENTS E. RENAME_SEGMENT
The transitions which they cause are set out in the following table:
TO STATE
1 2 3
S 1 - A -
F T 2 CDE+ - B
R A 3 CDE - -
0 T'
M E
+ See resolution of issue 3.1.6.
1. The Core permits access to the open segment by name (DELETE_SEGMENT, RENAME_SEGMENT and inquiry). GKS only accesses the open segment implicitly, and permits DELETE_SEGMENT only when no open segment exists.
2. GKS provides no RENAME_SEGMENT function, whereas the Core does.
3. The Core distinguishes two types of segment, "retained" and "nonretained". GKS distinguishes two types of segment, "work station" and "GKS file", though a single name may have both types of segment simultaneously associated with it. These concepts differ as follows:
3.1 A11 GKS segments have names. Only retained Core segments have names.
3.2 GKS work station segments survive until explicitly deleted or their work station is closed. Non-retained Core segments disappear at a new frame action.
3.3 GKS work station segments may undergo image and visibility transformations, if the work station supports them. Non-retained Core segments may not undergo such transformations.
Thus it appears that all GKS segments are "retained".
4. The support of image transformations is a property of the GKS work station, and all segments on that work station may be transformed. The support of image transformations is a property of the Core level, and only segments of appropriate type may undergo image transformations.
5. The Core provides, from level 2 up, only one place where segments may be stored, in the display f1le. GKS provides three places, namely "On" the work station, "In" the GKS file, and in a user record. GKS rejects the argument (a) of Core issue 3.2 and insists that all implementations include at least one display file (the GKS file). This decision means that there is no prohibition on functions parsing the display file (e.g. INSERT_SEGMENT).
It is not clear whether the definition of normalised co-ordinates, and their relation to the actual view surface matches that of the Core's normalised device co-ordinate space, or, if it does not, what the definition is.
The meaning of the words "relative co-ordinates" in the description of the SET_REQUESTED_DISPLAY_SPACE function is unclear; should they not be "normalised co-ordinates".
A distinction is drawn between "user co-ordinates" and "actual user co-ordinates" (eg pg.24). The difference is unclear.
The rationale between separating the window specification from the clipping rectangle specification is unclear.
The function SET_REQUESTED_DISPLAY_SPACE should presumably generate an error if the P.Q. rectangle has zero area.
Since clipping seems to be regarded as a work station function, the need for the clipping rectangle and indicator to appear in the "GKS File State List" is unclear.
The precise meaning of "deleted" in the context of functions CLOSE_WORK_STATION, REQUEST_NEW_DISPLAY_SPACE and RESET_GKS_FILE is unclear. Since there are two places from which the segment can be deleted, it should be explicitly stated each time to which of them the word refers.
The effect of DEACTIVATE_WORK_STATION on segments is unclear. For example, after:
ACTIVATE_ GKS_FILE
ACTIVATE_WORK_STATION(N)
OPEN SEGMENT (NAME)
.....primitives.....
CLOSE_SEGMENT
DEACTIVATE_WORK_STATION(N)
DELETE_SEGMENT (NAME)
ACTIVATE_WORK_STATION(N)
it seems to me that NAME is both an available segment name, and the same of a segment "On" work station N.
Pg.13 states that ACTIVATE_WORK_STATION is allowed in state "Work_Station_Active" whereas pg.l7 states that this causes error No.4.
The example program is misleading; it confuses normalised and user co-ordinates. REQUEST_SET_OF_LOCATORS returns user but TRANSFORM_SEGMENT accepts normalised.
If the initial sequence of states is:
GKS closed → GKS Open → GKS File Active
how can the initial value of "actual window" in the "GKS File State List" be "maximum display surface" since this refers to a particular work station which has not yet been selected?