eric@aplvax.UUCP (08/09/83)
I have just finished reading Jerry Pournelle's article in Byte on the future of microcomputer languages. <FLAME ON> AAAAARRRRRRRRRGGGGGGGGGGGGHHHHHHHHH!!!! Why is reading this man like listening to nails scraping across a blackboard? Is he trying to become the man that Byte readers love to hate? How can he make such sweeping statements when he admits to not having studied the products offered. <FLAME OFF> With that out of my system, I would like to make a few, well ... several, comments about the article. I will confine myself to the critique of 'C', although there is a wealth of other possible criticisms here. (FORTH being swallowed by LISP, indeed. I can see a 16K PROM burned with a LISP system). Without anymore snide remarks, onto the article. Jerry starts off by complaining that C programs are not self-documenting. I am curious what he considers a self-documenting language (i.e. if one exists). C certainly allows variable names to be descriptive, and is a structured language on a par with Pascal, or his beloved CB-80. He goes on to state: Alas, the programs are almost readable, and while one is writing C programs the purpose of each line is quite clear, so that further comments seem silly and are often omitted. Six weeks later the program is nearly incomprehensible. This is hardly a failing of a programming language. How many programmers write a line of code whose purpose is not clear at the time they write it? This applies across any language barrier, not just limited to C. His next complaint is that a simple minded program (a "hello world" variation) requires 11K of run-time library linked in. I would submit that printf is not one of the simpler output routines. Also, does this implementation link all routines in a library together, even if only one is required? Several Pascal implementations (such as UCSD) do this, but the complaint is not raised against them. His final conclusion on C is It will certainly survive. However, I don't expect it ever to become a highly popular language, and if I had to bet on its future, I'd say that it will take a respectable niche, after which its growth will be quite slow in comparison to the microcomputer world in general. I would like to point out that the Byte issue that this article is in contains reviews on no less than 20 different C compilers for various equipment. Not to mention the fact that Digital Research has decided to do all future work in C. Perhaps the most gauling point is the last sentence: I have a large number of C compilers and hope to do an extensive comparison of them for a future issue. This leads me to think that he has little familiarity with C and yet writes it off as unimportant. <FLAME ON> It is just this self-righteous, pompous attitude that has put off so many readers of Jerry's articles. If you look back on his early articles in Byte, they serve a very useful purpose. They are in depth, intelligent reviews of hardware and software systems. Granted, they were often esoteric and beyond the means of those on limited budget, but they were interesting. Now, he seems to be trying to review every item that hits the market, and act as expert on each. The result, not surprisingly, is he often fails to give reasonable explanation for his views, and is often just plain wrong. If you are out there, Jerry (and I seem to remember an article posted to the net by you), please take more time to study the issues, before passing sweeping judgements. And if you feel the need to lambast a product, program, or language, please present your arguments with some depth. <FLAME OFF> For those interested, Jerry picks Modula-2 as THE future language. His current choice is either CB-80 (a BASIC compiler) or Pascal MT+. eric ...!seismo!umcp-cs!aplvax!eric
MEAD@usc-eclb@sri-unix.UUCP (08/10/83)
From: Dick <MEAD@usc-eclb> Perhaps all articles (and especially spec sheets/ads) should have the disclaimer at the begining: The following is only the opinion of the writer, and may have no basis in fact! Actually, maybe Jerry does it on purpose, to rattle cages, heh, heh. To folks like myself who don't take any of it too seriously, it seems humorous (sorry). Nobody believed Columbus, either. -------
dan@idis.UUCP (08/15/83)
I suspect that Pournelle still lives mainly in the world of 8 bit machines with 48k bytes of ram and small floppy drives. The people that share this world with him (his intended audience) are consumer types interested in appliance class machines. From this perspective, Pournelle's views are easier to understand. He has buried his head in sand. When he gets over his 16 bit fright he will notice that cheap hard disks are common, average people are more sophisticated about talking to computers, and 8 bit machines are only used as outboard processors to run ancient and unportable cpm programs that were common in primitive times (1983).
RG.JMTURN%MIT-OZ@mit-mc@sri-unix.UUCP (08/20/83)
system costing under $2000. Sure, you'll never get a fullblown LISP up on a Z80, but the Z80 is 5(?) year old technology. Given time, I could almost certainly get it up on a 68010 (actually saw one last week!) with a Winchester and a virtual address space. That technology will only run you $7000 or so this second! Personally, I don't think LISP would make a good micro development language for other reasons, but I certainly relish the idea of a low-cost AI and expert system development package. In fact, the interface to DJ Info Service using natural language which was recently announced in InfoWorld running on the TI-PC was developed in LISP at TI-CSL on one of their LISP machines, then ported. Would it also surprise you to know that Logo Computer Systems Incorporated (LCSI) uses three LISP machines to develop the different version of LOGO they market for home computers. I think you may see real LISP systems available sooner than you think. James Turner Hardware/Software Engineer Lisp Machine, Inc.