Computer history: Graphics facilities (1986 – 1989)
Graphics facilities: GX4000 and I2S
In 1988, TNO’s graphics facilities working group (ITGC), being a part of the IT-policy advisory group of the TNO-FEL Laboratory (ITBC), prepared the replacement of the Evans & Sutherland system. TNO decided on a SUN 3/260 – Raster Technologies GX4000-configuration and a small SUN 3/60.
The SUN 3’s were installed at the end of 1988. The GX4000-boards, monitors and other equipment were installed at the beginning of 1989 by Ponder Associates which had the responsibility for the delivery of the integrated system. The GX4000, a ‘state-of-the-art’ product, comprised PHIGS+ software and special graphical boards that could generate and display graphical vectors very fast. For 3-D effects, a hardware Z-buffer was available. On March 28, 1989, the system was introduced during an internal colloquium by Division 2.
Acceptance of the system could not take place because certain offered and documented functionality did not work. In 1990, Raster Technologies was taken over by the Alliant Company, a manufacturer of multi-processor mini-supercomputers. The PHIGS+ software and special hardware boards contained many errors. It was decided to postpone the acceptance procedure until a new version of the SunOS and Raster software was received. Many hours by System Management & Support (SMS) and others were spent trying to get the system operating as it should. After 2.5 years, it was decided to discontinue the system as a general-purpose system. The system could be used by some specific projects ‘as-is’. Users had to program around all errors, which required a lot of extra resources for the projects. The system was from now used as a UNIX workstation and delivered display facility by division 2. These experiences lead to the decision to stay away from “state-of-the-art” products intended as general user services.
VAXstation 3500/I2S
The grid-graphic COMTAL system was also due for replacement at the end of 1988. In early 1989, a VAX station 3500 was ordered from Eurosense Inc. The system controlled two IVAS-display stations (I2S-system). The VAX station 3500 was connected to the VAX 11/750 as part of a Local Area Vaxcluster (LAVC). The VAXstation 3500 had 16 MB of memory and a 300 MB disk. The acceptance process of this system was also far from flawless. On April 6, 1989, 43 problems were logged, part of which could be resolved quickly. The expectations about the graphical software, however, were higher than could be realised by the manufacturer. In contrast to the problems with the GX 4000, the system could be used for its purposes without the need to find complicated detours. Because the ideas regarding (raster-graphical) image editing within the Observation Systems division changed, the system was hardly used anymore in early 1993. Reason to put the system out of service on November 1, 1993.
High wire time: plotting and previewing
In addition to the plot facilities under NOS/BE and NOS/VE, users of the Central VAX (VC) expressed their need for plot facilities. As the CYBER system programmers had little experience with the VAX/VMS operating system commands, its compilers and editors, it seemed a nice challenge to get to know the world of DEC systems better. The challenge was to adapt the plot library developed under NOS/VE (create an ASCII file with the low-level plot subroutine numbers and the parameters) for the VAX environment. The interpretable file with the coded routine calls could then be transferred with FTP to the CYBER NOS/VE part using the ftp concepts as previously discussed to get the print files via the double gateway to the CYBER. Much to the surprise of many, including the VAX/VMS systems programmers, the complete package was compiled and ran within one day!
In early 1988, there was a large demand for plot-preview facilities under NOS/VE for several projects. A first estimate was that creating such a package would require a man-month of work if use could be made of the new Graphical Kernel System (GKS) software. At that time, nobody with SMS had any experience with GKS, so it was difficult to make a correct estimate of the efforts and time. On a Friday afternoon, a new book about GKS was discovered on the library shelf. The temptation was great: the book was taken borrowed for the weekend. A listing of the previously described basic plot routines under NOS/VE was taken home as well.
After reading the first chapters with the basic principles of GKS, every basic Calcomp routine (above that was a whole library of complex drawing routines) was rewritten as a subroutine with a set of GKS calls. The most tempting problem was writing the correct scaling routines. Already after the first error-free compilation, the first drawing could be previewed. The preview functionality was tested and included within a week as a production library!
Enthusiastic now about all these stunts, the VAX/VMS system programmers teased their NOS/VE colleagues. It should now be very easy to provide Preview to the user community on the VAX/VMS systems. A new and exciting challenge! It lasted not long before we got it working after we had transferred the source code of the GKS package using ftp to the VAX. And not only that. Some code fiddling with GKS allowed us to create plots with the Calcomp software library as an interface to other output devices, for instance, a colour printer and a Postscript printer. Reasons enough to write a Calcomp device driver for the VAX/VMS GKS system. In this way, previewed GKS drawings under VMS were transported via VAX/VMS and a DEC Ultrix gateway to a Network device interface (NDI), NOS/VE, NOS/BE and the Calcomp 906-controller to the Calcomp 1051 plotter where they finally were materialised on paper or vellum. The use of the TNO-FEL Uniform ISO/OSI Basic Reference Model at its best!