taylor (05/09/83)
Relay-Version:version B 3/9/83; site harpo.UUCP Message-ID:<178@sdcsvax.UUCP> Date:Mon, 9-May-83 15:30:44 EDT As an ex-employee of the UCSD Pascal project (via proctoring a freshman programming class taught (marginally) by Ken Bowles himself), I have a few thoughts on the P-System: First and foremost: It's a great idea whos implementation sucks! (Pardon the 'french'). I get very hostile when I have to wait 2-3 minutes to run a previously compiled program that DOES NOTHING!!!! (and on it's own machine, yet!!!) The single most needed piece of support software for the P-System is a p-code to native code translator: Think of the consequent advantages; you could write your program in the *cough* language of choice, Pascal or Fortran or Basic, then distribute the P-Code. The user could buy the P-Code then run it through a translator and get !!!voila!!! machine-code version of the program!!!! In the meantime... We have some neat machines here on campus (UCSD) not the least of which are the bit-mapped Terak 8510's. Nice lumpy computers. Here (of course) they run the P-System...funny things happen if you try to use the Basic compiler that doesn't really exist (according to the P-people) (hah!); the entire disk that you are running on gets trashed! YEAH!!! Neat trick, my friends and I think - how can we simulate that?? Ooops: can't get access to ANYTHING to do with the system from within a p-code based program, since the pcode would have to be specific to the processor/OS...which conveniently brings up the second major problem with the P-System: Portable Object code, or the Mythical ANSI standard for CPU chips... Yes, you guessed it! All the nifty chips that are currently floating around have UNIQUE instructions, and certain tasks are better implemented in different ways on the different chips! Meanwhile...the P-System says, 'naaahhh....no worrys with this; we'll let the INTERPRETER worry about this, since it has to be written specifically for each chip anyway!!!' But, as they say, there's the rub - the interpreter suddenly becomes a system unto itself, and the P-code might as well be BASIC source code fed to a FAST basic interpreter. At least the user could modify the code to his (her) own needs if desired!!! The crux of the matter is that I don't think that portable object code is the way to go: I LIKE SOURCE CODE!!!! If I find a bug in a program (which they assure us never happens....) I WANT TO FIX IT!!! ditto if I decide that certain functions should be added... Besides, isn't the proper direction on programming language standards, not generated CODE standards???? with a deep sigh of disgust, -- Dave Taylor trapped deep in the gloomy halls of UCSD, La Jolla, California!! ps: the tone in this is a bit wierd - post-midterm shock, I guess!