The aim of the session was to describe what is currently available to universities in the way of graphics packages. Hence most of the talks were either about the packages themselves or users' experiences with them. However, most of the discussions centred (or raged) around the question of standardisation. This came close to being a pro-GINO versus anti-GINO battle but eventually got on to the more general aspects of standards. The basic question seemed to be at what level, or levels, should there be a standard. There was a general consensus that there are 3 possible levels at which a standard could be defined. They are:
(1) Pseudo code: ie. some well-defined code which could be generated by all packages which can then be post-processed by various 'handlers' for the graphical output devices.
(2) Subroutine call: ie. all packages should have standard subroutine calls for line-drawing, text plotting, etc.
(3) Application: ie. there should be standard subroutines for graph drawing, contouring, etc. Of course, these 3 levels are not mutually exclusive and packages could be standardised at all of these levels.
As is generally known, CADC have submitted GINO-F to BCS, ISO, AFIPS, etc as a standard for graphics packages. Many users and non-users of GINO-F were somewhat critical of basing a standard on GINO. However, CADC did say that they had proposed it as a standard rather than the standard. No conclusions were reached about a standard or even a level for a standard but everyone did agree that a standard would be a "good thing". It seems that virtually all of the packages described, except those from ACL, use some kind of intermediate output format which is then post-processed for output to a particular device. It was also commented that graphical input should be considered when defining a standard for graphical output.
Tony Davies, University College of Swansea.
The first speaker defined what he considered to be the basic requirements of a graphics package and its implementation.
The world not yet being Utopian, the above described package does not exist. For the Swansea site 4 packages were considered, all of which had disadvantages:
(1) GD3 from CERN
Not well accepted in this country.
(2) GHOST
Poor support and user documentation, not many high level routines, no 3-D.
(3) GINO-F
Inefficient in core usage, marketed via agents, PSEUDO format not available to users
so they are unable to write their own back-ends.
(4) SPROGS
3D not general enough, arrived rather late on in the consideration.
Eventually they chose GINO-F. Because they were a test site they had direct support from CADC, and so the implementation was handled well. Other users, especially those on ICL machines had considerable problems with the implementation by the agents. The agents for ICL machines are Dataskil and there was general condemnation of them.
Dorothy Sims, CADC
The usual GINO-F presentation. Various applications packages based on GINO-F are to be released to users some time in the future.
These will be charged for in a similar manner to the basic package. A new pseudo code is being written and when this is finished it may be released to selected users eg Tony Davies.
T Long, University of Liverpool
This talk was about the implementation of GHOST at Liverpool. GHOST has a few useful features such as contouring on irregular grids and graphs on log scales, which are not readily available at ACL. However, it does not have very many high level routines. The Culham manual was thought to be good as a reference manual but was rewritten at Liverpool for teaching purposes. GHOST seems to lack a central support facility and so there are now some variations between sites.
J Gilbert, ULCC
The package DIMFILM has been written recently for ULCC's Calcomp plotter. As well as the normal graphical routines there are routines for centring text and a selection of line types including, dotted, dashed and dot-dashed. The dotting and dashing can be continuous over several line segments. DIMFIIM seems somewhat reminiscent of SPROGS, both packages being aimed at similar types of users.
Martin Dowd, University of Bradford
At Bradford they have a very simple graphics package which provides little more than 2-D graphs and histograms. This is claimed to satisfy some 90 percent of their users. The graph-drawing routine has 12 arguments which apparently does not upset the users. They probably are not aware of any other packages and so accept what they are given.
Several people expressed interest in obtaining copies of SPROGS and SMOG to generate their own tapes for the FR80. Cliff Stone from UMRCC even brought along his own magnetic tape.
