LISP [was RE: John McCarthy]

From: Liam Proven Sent: Sunday, October 30, 2011 9:00 AM On 29 October 2011 03:15, Rich Alderson <RichA at vulcan.com> wrote: [snip] Since my first exposure to Pascal was via the 3-part Tiny Pascal article in BYTE, where a subset compiler was written in NorthStar BASIC, that would be very interesting. Writing a Pascal compiler for the DEC-20 in BASIC+2 was how I stopped doing my Ph. D. work in linguistics. Back when I saw first a version of this, Emacs Lisp wasn't on the radar, and H'sMC was not written, so perhaps I should be quiet. Back in the early days, when dinoLisps roamed the earth, there was a very successful variant called MACLISP. This ran on PDP-10s, under ITS and TOPS-20 and WAITS (the Tops-10 variant on SAIL), and on a GE-645 under Multics.[1] Now, there was a desire to have a version of the EMACS editor for the PDP-10 on Multics, because it had built into it an awareness of LISP syntax, even though it was written in a far more arcane language called TECO.[2] A bright fellow, Bernie Greenberg, thought to himself that with just a few tweaks, MACLISP could be used to create a version of EMACS on Multics, and proceeded to do so. At the same time, over at the MIT AI Lab, people were experimenting with new progamming languages all the time, and the best language for implementing new programming languages was LISP, and the best LISP was MACLISP. These folks were looking into problem-solving systems, and the first successful language of this type was PLANNER. This was followed by CONNIVER. Then, like a little mammaLisp among the dinoLisps, there arose SCHEME. This was written, like the others in MACLISP, and showed its dinoLisp DNA in places. SCHEME fell into the hands of the smaller-is-better enthusiasts, and had more and more of its origins trimmed away with each new revision. It is nearly perfect. This makes it hard to use, hard even to grasp, except by those who spend all of their time concentrating only on Scheme.[3] MACLISP continued along in its imperfections, influencing the development of Lisp machines, such as Symbolics, LMI and the TI Explorer, and of new variants of Lisp on strange architectures like the VAX (for example, NIL under VMS and Franz Lisp under Unix). So many variants of MACLISP arose that some felt a need to reduce the clutter. But these were people of vision, so they invited not only those who worked in the MACLISP variants, but the Standard Lisp/Portable Standard Lisp folks what the common denominator of all these variants of Lisp might be. And so was born Common Lisp, which rather than being the refined intersection of all which was the ultimate fate of Scheme, became the enhanced union of all. In the mean time, those Lisp machines used editors which grew out of the Multics (!) version of EMACS written in MACLISP. The maintainer of the software for one manufacturer, LMI, who was the original author of EMACS in TECO, saw that this was good, and bethought himself that making this also available on Unix workstations (the wet dream of the day) would be a good idea, along the way to overthrowing the source-hoarding Unix hegemony. To do this, he created a version of Lisp explicitly tailored to writing text editors, and created a version of Emacs to run under his hypothetical OS, called GNU.[4] So while Scheme and Emacs Lisp and Common Lisp appear very different, they have at their hearts the same DNA. Hmm, I guess I did have a lot to say about it. Sorry 'bout that. [1] Or a Honeywell 6180 or follow-on. Anyway, a 36-bit Multics system. [2] Sure, TECO is an editor. But real TECO is Turing-complete. [3] OK, OK, IMAO. [4] GNU's Not Unix. Rich Alderson Vintage Computing Sr. Server Engineer Vulcan, Inc. 505 5th Avenue S, Suite 900 Seattle, WA 98104 mailto:RichA at vulcan.com mailto:RichA at LivingComputerMuseum.org http://www.LivingComputerMuseum.org/