Abstract
This paper examines two important objectives which a standard for computer graphics should achieve. An interpretation of these objectives is given, which attempts to cast them in a practical light. The resulting framework of a standard for computer graphics is presented.
Computer graphics has been recognized as a separate area of study for well over a decade. Many devices, techniques and methodologies have been developed to meet the needs of a wide range of applications. It is felt by some that the time is now ripe for an attempt to be made at specifying a standard methodology for a significant subset of the developed technology.
However, many of the proposals for a standard lack a specification of the objectives such a standard should address. This can lead to proposals that achieve goals that bear little or no relationship to the true objectives. Therefore, in order that a proposed standard may be evaluated, it is essential that the objectives involved are clearly defined. Indeed, any discussion of a standard must start with a discussion and agreement on the objectives.
For the purpose of this paper, two objectives are defined and discussed.
Neither of these objectives is particularly helpful in directing the formulation of a standard. It is necessary to specify the level of details at which the objectives are to apply. In order to do this the well worn analogy with the FORTRAN standard is used.
Relative to the first objective it is illuminating to identify which aspects of FORTRAN must be considered to achieve portability, and which aspects are really not important. It is conjectured that the aspects of prime importance are global in nature, such as data types, parameter passing conventions, and the avoidance of recursion. Aspects which are believed to be of secondary importance are those which have a local impact and can be converted to other conventions by automated methods; for example, the logical IF, the ordering of subroutine parameters, the names of subroutines and the number of characters in a name.
A very important point concerning these two aspects is the observation that the global aspects of the FORTRAN standard are more widely known than the local aspects. Very few people are familiar with details of the ANSI Standard.
In light of the above it is proposed that a standard for graphics should be concerned primarily with the global aspects of the organization of programs which use graphics. Indeed, it is believed that any attempt to specify details, such as subroutine names, order of parameters, of representations of transformations, will be ignored by the majority of people for whom the standard is intended. This may be because the details are too numerous to remember, or because they force too many compromises in any particular installation.
Such a proposal is consistent with the second objective in that the important aspects of a recommended methodology are global in nature. A recommended methodology should address such considerations as program structure, philosophy of interaction techniques, handling of slow speed devices, etc., all of which are global considerations.
The previous section, if accepted, would indicate that it is neither desirable nor useful to adopt a standard which rigidly specifies every last detail about the graphics content of a program. This implies that the creation of a standard by adopting one of the several graphics subroutine packages presently available (GINO-F, GPGS, DISPLA, etc.) is inappropriate. However, the adoption of a standard which specifies some combination of the global philosophies underlying several of these packages would be acceptable. Moreover, the appropriate choice of global aspects could allow several of the existing packages to conform to the standard. This is an important point in that these packages are already implemented at many installations throughout the world. Abandonment of these packages in favor of some new standard would represent an intolerable expenditure and simply would not happen.
There is a lowest common denominator effect in combining the global aspects of several packages. However, the high level of the standard requirement would avoid the subsequent generation of a standard which consists of a trivial subset of all the existing packages.
It is proposed that a standard for graphics should be concerned with the following global considerations:
The standard should not be concerned with such aspects as:
For example, consider the class of facilities having a single-user processor and a refresh display system which permits modification of parts of the picture. The standard for this class should specify detail to the following level.
While this list is by no means complete for this class of facilities, it is hoped that the examples given will illustrate the depth of detail which, it is proposed, is reasonable and useful to specify.
It is recommended that the discussion of any proposed standard be preceded by a clear statement of the objectives addressed. It is suggested that standardization at the level indicated is both practical and useful. An attempt to define a standard which excludes most of the existing widely available graphics packages should be avoided.
