Before any useful discussion can take place it is essential to have some agreed purpose in mind. A choice must be made between attempting to advance the state of the art on a theoretical plane, and attempting to consolidate our knowledge and help present users of graphics.
The field of graphics is changing so fast that advanced theoretical work is not likely to remain relevant for long, nor is this workshop a suitable environment for such work. The workshop should therefore concentrate on distilling and agreeing good practices from current methodologies. In doing this the participants should adopt the same attitude as other engineers have adopted over the last century in drawing up codes of practice. It is only in this way that programmers and designers can become software engineers.
Having agreed to a general purpose it is then necessary to identify a clear objective so that we can deduce some criteria for choices. A simple but all embracing objective is desirable:
To reduce the resources required to use computer graphics
These resources include hardware, manpower, software, media, etc. Some of the mechanisms by which this reduction will be affected are:
Any report produced must be scholarly and of a high academic standard - yet it must also be practical and down to earth. No principle or theory is sacred - all decisions must be made on the basis of sound arguments. Some useful criteria for these arguments are discussed below.
It is absolutely vital that agreement be reached on some topics however few. Disagreements are of no value to the community so time must not be wasted discussing topics on which agreement cannot be reached.
When agreement is reached it must be documented. Therefore all topics discussed must be rigorously and formally definable. The method of definition must be comprehensible, clear and consistent.
Proposals must be usable in a practical environment. To achieve this they must be simple, and ideally use a minimal set of concepts and principles. To achieve this each concept defined should be allowed in all its manifestations and no special cases should be required - that is all concepts should be as orthogonal aa possible. Despite this requirement for minimality the concepts must be natural, comprehensible and constructable (Cf. Algol 68).
The cost of adopting the proposals must be less than the benefits. This creates great difficulties when a variety of different application areas are considered. The provision of a facility for one use must not imply an unnecessary overhead for another use. This requires considerable flexibility and modularity.
The area of applicability must be well defined, and within this area proposals must meet the needs of the users without being unnecessarily restrictive. The degree of device, computer, and application independence must be agreed upon and the restrictions implied by this stated.
No proposals will be adopted unless they are expected to have a long life. To achieve this they must be stable against changes and have a high probability of continuing to represent the best techniques for a considerable time. To achieve this in the context of a changing technology proposals must be open ended and extendable. That is not to say that features which cannot be agreed upon must only be partially defined (Cf FORTRAN) but rather that they should be omitted and a mechanism for the later extension of the proposals into this area provided. Furthermore the proposals should not curtail future developments in related areas.
The proposals should provide a basis on which to construct robust and reliable software. Consideration must be given to the handling of errors, exceptions and failures, and proposals must be consistent with the latest advances in software engineering.
It is most undesirable for proposals to conflict with existing practices and standards. If conflict is unavoidable it must be clearly stated.
Graphics covers a very wide area and unless the workshop concentrates on a few specific topics it will not make any progress. The following restrictions to the area of discussion will increase the likelihood of useful proposals emerging.
The workshop should not discuss:
What is left?
This must be discussed in an open ended way such that the proposals can be extended to cover the functions omitted as and when agreement is reached on them.
The criteria discussed above provide a basis on which to make decisions. There must be very good reasons before any proposal not satisfying them can be accepted.