Following Sun Microsystems’ announcement of a merged NeWS and X.11 window management system (CI No 681) members of the industry have expressed thoughts that this may become the standard. The president of Applix Corp said that his company whole-heartedly supported the merge towards a standard. Electronic Data Systems thinks that, following its success with the Network File System, Sun may have another winner with the merged system saying that Sun is following the tradition of NFS with this announcement of a strategy to unify window system platforms across workstation vendors. Intel Corp considers that NeWS alone is sufficient and considers that NeWS will be the chosen OEM windowing system for the Unix-based 80386 products. London-based Whitechapel Workstations sees Sun’s way forward as the best way to support the valuable but less sophisticated X standard. X/Open technical manager Hans Strack Zimmerman of Siemens at The Instruction Set’s Unix Directions conference the other day, descibed his views on the windowing problem with a discussion of window managers designed to stir up opinion, in comparing X Windows and NeWS. And probably summed up much of the industry’s thoughts and fears concerning the issue. He descbibed X as taking a very traditional approach that lacked the sophistication and facilities available with NeWS – a view made rather more controversial by the fact that the X/Open Group is looking to standardise on X.11. This version, he pointed out, is on alpha release and he claimed that originator the Massachusetts Institute of Technology is not allowed to release it because other standards bodies are also looking to adopt it. As an example of the X approach, he noted that X thinks in terms of pixels – different fonts are handled by being pre-computed into a pattern of pixels and loaded into the display station. This immediately restricts the ability of the workstation to manipulate the result. Saying that the Xlib library of window management routines were too low level and a higher level set of Widgets – presumably the tools being developed by the X consortium – which would appear maybe next year. But higher level toolsets would probably never be standardised – most companies would have their own proprietary toolsets that fitted their own product strategy. With X.11 comes an extensible set of facilities, with some opcodes unassigned to allow for definition of extra functions – but he was worried about the result potentially being a set of non-standard extensions. In contrast, the highlights of the PostScript-based NeWS, he said, included the facility to to express complicated thoughts – but the language is also not so easy to use. While extolling the power of a system that uses a programming language, interpreted in the display station, as a protocol, he pointed out that the technology is not all proven for workstations. I think it’s a little early in time, he added. And in contrast to the well- understood X, it is not easy to speed up Postscript functions by implementing them in silicon. He gave as an example of the flexibility of NeWS the way fonts are handled – stored not as a bit map but as an outline, which is passed to the workstation; the workstation is then able to enlarge or tilt the result as required. But hope of combining some of the benefits of NeWS with X is at hand – he said that some of the PostScript model will turn up in the display areas of the first X standard – and displaying PostScript information in X windows could come close to providing the required flexibility with a standard.
ASCII terminal’s days are numbered
When a member of the audience dared to ask about the possibility of windowing systems for non-graphics terminals, Zimmerman said that all attempts at one scheme that deals with both bit-map and alphanumeric terminals have failed. If X/Open had been formed five years previously, perhaps it would have looked at a standard for character-based terminals – but he felt that the days of the character-based terminal were anyway numbered. That’s a common view
which seems to be inevitable in the long run, and now that you can get a 16-bit micro for UKP500, the future of the dumb terminal looks distinctly perilous: perhaps it’s just myopia, but the resellers and manufacturers at the sharp end of support for the Unix market simply don’t see it like that – last year they were selling small boxes with dumb terminals, next year they’ll be selling bigger boxes with the same dumb terminals.