wmb@MITCH.ENG.SUN.COM (12/11/90)
> > I sincerely believe that the language desperately needs improvment. > > I agree. I just don't think a "Standards" body is the place for it. If there were another effective forum for improving Forth, I would gladly use it, in preference to all the difficulty, stress, and expense of an ANS standard. However, I have come to believe that no other method exists. I have tried the "publish and promote" method. I have been publishing and promoting a "stdio-like" file system interface for 7 years, at various conferences and in personal contact with other implementors. The result: every serious vendor now has his own file system interface, and there is not a shred of compatibility between them. I have tried the "put out a system" approach. This ends up fragmenting the community even more. Some techniques become popular in subsets of the community, but there remain several "domains" which tend to move in different directions. For instance, F83 was widely used by some groups, and widely ignored by others. A standard is the only thing that has sufficient force that every new and existing implementor cannot ignore it. > Imagine what a world this would be if the screw base of the common light bulb > was redesigned every four years. The ANS situation is not nearly so profound. It's more like the addition of the third prong to the US wall sockets. 2-prong plugs fit into both old and new sockets, 3-prong plugs fit best into 3-prong sockets and can be relatively easily adapted to 2-prong sockets. Computer technology has changed profoundly since the last standard. Back then, memory cost $1600/megabyte, and you could fit a megabyte on a Multibus card. Now, it costs about $50/megabyte, and you can fit 100 Megabytes on an SBus card (a little larger than a cassette tape). We are talking about a factor of 300 improvement in just 8 years. (This blows my mind!) In an environment that is exploding at the rate of computer technology, it's not surprising that programming languages need some updating. The improvements in ANS Forth codify techniques that have been around for 10 or 15 years. We're not talking about pushing the envelope here; we're talking about catching up to late 1970's technology. It is silly that Forth programmers currently have to roll their own file interfaces and string packages and error handling mechanisms and memory allocators (or learn different ones for every different Forth system). These are old worn out wheels; the Forth community gets negative points for continually reinventing them. As an amusing side note, technological improvements are finally driving screw base light bulb "standards". Compact fluorescent bulb with screw bases now offer very attractive long-term energy and cost savings. The only catch is that the bulb part is bigger and heavier, so they physically won't fit in some fixtures (i.e. they will screw in the socket, but the shade won't fit over the bulb). The need for evolution in standards is directly related to technological evolution. Incandescent light bulb technology has been relatively static for 30 or more years. Mitch Bradley, wmb@Eng.Sun.COM