Talk:Lisp machine

Latest comment: 3 years ago by Blah2 in topic Xerox machines

Nubus and TI

edit

I'm not a good enough editor to fix this with the right referencing. But the article currently states that the Nubus was developed at MIT for use with lisp machines. But that is not correct. The NuBus was developed as part of a machine that ran Unix on a 68000 processor. See http://en.wikipedia.org/wiki/NuMachine

TI later manufactured these in a big cabinet, and the LMI Lambda was a processor board set that fit into a NuMachine. — Preceding unsigned comment added by 69.46.227.226 (talk) 23:35, 12 February 2015 (UTC)Reply

Sony and Lisp

edit

One of the deleted edits mentioned rumors of a prototype Sony handheld lisp machine (presumably from around 1978). I couldn't find any information supporting this claim on the net. And I believe Japanese companies at the time were much more interested in Prolog than Lisp. --Zippy 02:00, 18 October 2005 (UTC)Reply

Yeah a Google test really only popped up this page as the only explicit mention of anything relating Sony to LISP, so I chalked it up to patent sillyness / vandalism / whatever you'd like to call it. It was in the article for quite a long time too, haha. --I am not good at running 22:47, 23 October 2005 (UTC)Reply
That's sort of my fault; during the AI boom, the Japanese were intensely interested (esp. the big companies- like Sony), and so I considered it plausible enough to leave. (BTW: it was the Europeans who were in love with Prolog.) --Maru (talk) Contribs 04:25, 7 November 2005 (UTC)Reply
I know Prolog was popular in Europe (The Sicstus implementation of Prolog was from Sweden, if I remember correctly) but I also remember that in the late 70s/early 80s when Japan's government and industry really got into AI, Prolog was their big language. I'll see if I can find a reference.
Ah yes, here's one. A history of Prolog: "To this day Prolog has grown in use throughout North America and Europe. Prolog was used heavily in the European Esprit programme and in Japan where it was used in building the ICOT Fifth Generation Computer Systems Initiative. The Japanese Government developed this project in an attempt to create intelligent computers. Prolog was a main player in these historical computing endeavours." --Zippy 08:56, 7 November 2005 (UTC)Reply
Actually, now I recant; I've read further in The Brain Makers referenced, and I found this:
"Complicating and confusing matter even further, Japanese companies participating in The Fifth Generation project were buying LISP machines from Symbolics. Even though PROLOG was the designated language of The Fifth Generation, ICOT researchers found that it wasn't as flexible for creating some of the more ambiguous human attributes as LISP was. Thus, some of them were sneaking around the party platform. One Japanses Fifth Generation member, Fujitsu, had even developed a LISP Machine it wanted to sell in the United States."- pg. 331
--Maru (talk) Contribs 01:45, 13 November 2005 (UTC)Reply
Actually reading through the old SLUG mails, there was some kind of collaboration with Sony and Symbolics planned (mid/end 1980s). It was the time when Sony had their NeWS workstations. But it was not realized. I also read that some company (not Sony) in the end 1980s wanted to develop a Handheld computer based on Symbolics' Ivory microprocessor. -RJ —Preceding unsigned comment added by Joswig (talkcontribs) 15:27, 31 January 2008 (UTC)Reply

Lisp

edit

I added a link to ICAD (a Knowledge Based Engineering system) as an example of a system that started in Lisp (on the Lisp Machine) and then ported to Unix as Common Lisp evolved.

Personally, I have worked with 9 variety of Lisp Machines in a commercial context (Symbolics (3 variety), LMI (2 variety), TI (3 variety), Xerox). It would be good to see other examples collected.

Related topics to be added: -- Moby (spelling?) memory - LMI's (?) work on distributed Lisp, -- tools such as KEE, Knowledgecraft, ..., -- applications - KBE on Unix accounted for buckets of dollars (cost savings) in very large system design and manufacture (aircraft), -- other approaches, such as IBM's implementation of KEE on the mainframe, ...

I've taken the liberty of doing a little formatting on the ICAD article. --Maru (talk) Contribs 23:45, 27 December 2005 (UTC)Reply
Thanks. I may have clobbered those changes. Sorry, I need to be more careful. (Turns out that I wasn't logged in.)

Added GBB as another example. Let's get these application engines recorded and recognized!!! jmswtlk 13:22, 28 December 2005 (UTC)Reply

What Lisp Machines brought (or wrought)

edit

Over the years I have heard many people talk about the power of the Lisp machine while rueing its departure from the scene. The Lisp that ensued on Unix took many years to get back a lot of the functionality.

In a sense, development was crippled.

This is an analog, IMHO, of a larger problem of 'green field' versus 'legacy' that, perhaps, ought to be discussed elsewhere; however, the Lisp machine experience offers a wonderful and concrete example. So, let's hear stories about pre and post. Those who liked the Lisp Machines liked Lisp. What became of those who had the 'bounty' of this experience (or better, could understand the significance)?

A recent example is ICAD which is in hiatus. Given comments posted on this page, there is something missing in the proposed replacement systems. Has anyone characterized what might be the missing pieces and why Lisp might still be useful? It's interesting that GBBOpen is Lisp-based. jmswtlk 13:22, 28 December 2005 (UTC)Reply

Much of the history is yet to be written. Remember that Intel was just starting to build a 32-bit chip at the time when lisp was very big with some rather wealthy US Government customers. In fact Intel cpu architects met with many big LISP customers before the i386 was produced, and met with people at LMI too, with an eye toward possible cooperation. What exactly was important about the CADR? Why would TI want to produce a LISP chip? How did this relate to the (Intel in particular) efforts at an Ada chip and other high-level language chip sets?

For the general market cheapness and price/performance were the only drivers. But other markets might like to have a bit more security at the hardware level.

How much human effort has been wasted in the last 35 years by people debugging memory buffer overlap bugs? Array out of bounds bugs? Enough to support 1000 companies the size of Symbolics and LMI combined. — Preceding unsigned comment added by 50.245.46.27 (talk) 11:38, 2 June 2016 (UTC)Reply

Ambiguity

edit

This is a great article! WP:FA, anyone? :) But while copyediting, I ran across the sentence

Often it was affectionately referred to as the "Knight machine", perhaps since Knight wrote his master's thesis on it); it was extremely well-received.

Two problems: First, WP:RS; if we're going to claim that the machine was affectionately referred to as such-and-such, we'd better have at least a primary source for that (e.g., a Wikipedia editor who claims to be an old-school Lisp hacker), or better a secondary source (e.g., a dead-tree history of Lisp). Second, an ambiguity: Did Knight write his thesis on the computer, perhaps using Emacs? Or did he write it on the computer, perhaps focusing on its history and development? Assuming the latter, I changed the underlined bit to "on the subject"; but if that's not right, please make the appropriate substitution. --Quuxplusone 05:39, 30 March 2007 (UTC)Reply

I worked with the CADR at MIT from the time of its birth, (including assembly, using the infamous debug cable, whatever was needed to get the job done, which was getting the application work done, whereas system and hardware work was just part of the pain we all had to pay to gain access to such an amazing tool). Well, I never once heard anyone refer to the thing as a "Knight machine" and I couldn't imagine Tom Knight going along with that. What was the CADR emulator (written in PDP-10 assembly language) called? The Greenblatt Machine? Ha. Such designations were just not part of the culture. Was the Connection Machine the Hillis machine? Now I am not saying that personal names were not added to some things, especially when there were software packages that represented different authors points of view or alternatives. There were different window systems, different object systems, so it was useful parlance to refer to things by different authors names, not necessarily with affection! — Preceding unsigned comment added by 50.245.46.27 (talk) 11:27, 2 June 2016 (UTC)Reply

Firsts?

edit

Laser Printing was developed first on Lisp Machines at Xerox? Windowing Systems? Computer Mice? Bitmapped Graphics?

I think they were available early, but they were not developed first on Lisp Machines. The things that Xerox developed (also not the first) were often done for their office system or for the Smalltalk systems. Those were often using similar hardware, but were not developed in Lisp.

I see two alternatives:

a) replace 'developed first' with something like 'developed early'. b) coming up with references that backup those claims. —The preceding unsigned comment was added by 85.176.6.193 (talk) 22:41, 7 April 2007 (UTC).Reply

The Alto was the first known system to use a bitmapped GUI, the mouse was introduced for the NLS... therefore, I cannot help that unless specific examples are provides, the 'first' statement is simply in error, so I have removed it. JamesFox 02:13, 29 June 2007 (UTC)Reply

Missing LISP machine ?

edit

During a talk at Cambridge University, I heard Arthur Norman refer to a Lisp Machine known as SLIP - though I have no references for it. Does anyone have any further information - or I might write to ACN himself to "remind" me ! AbruptlyObscure (talk) 12:55, 14 March 2017 (UTC)Reply

AbruptlyObscure, maybe this? https://en.wikipedia.org/wiki/SLIP_(programming_language) Although it's not a machine.

Were Lisp machines special?

edit

Lisp machines were general-purpose computers designed (usually through hardware support) to efficiently run Lisp as their main software language.

What type of "hardware support" did Lisp machines provide that made them better platforms for running Lisp code? Did they have more memory and faster processors than other machines of the day?

The article answers this. If you don't want to read the linked technical reports, the gist is that function calls were cheap, the dynamic typing overhead of checking the type of arguments to a function were built in along with a number of other helpful things in the microcode, the address space was large enough to support the kind of virtual memory Lisp (and garbage collected languages in general?) needs, etc. The Lisp machines were also really good compared to contemporary workstations like Sun's. So they were quite different, architecturally speaking, but they were not the most radical such architectures - at least compared with some research project optimized for functional programming have been, such as the ones which had graph reduction built into the hardware. --Gwern (contribs) 04:04 20 May 2007 (GMT)

Inconsistency with Emacs history

edit

This page says that Emacs was ported to the Lisp machines in 1975, but the Emacs article says the first complete implementation was in 1976.

Arghman (talk) 21:42, 30 November 2014 (UTC)Reply

I agree with Arghman's observation. The file EAK; EMACS LORE says In August 1976, a bunch of hackers decided it was time to write a new editor. A runnable binary file for EMACS version 24 is dated November 1976.
In contrast, the earliest known email to the BUG-EINE mailing list is from January 1978: https://ml.cddddr.org/bug-eine/msg00029.html This suggests EINE was ported in 1977. Lars Brinkhoff (talk) 13:31, 28 February 2019 (UTC)Reply
Though it's not visible here, the LMDOC; EINED 14 file is timestamped 1977-11-09. This puts EINE squarely in 1977, or possibly but unlikely 1976. Lars Brinkhoff (talk) 13:28, 10 March 2019 (UTC)Reply

Response to von Neumann architecture?

edit

Could it be that the Lisp machines from the 1980s is because of some critiques on von Neumann architectures (and "von Neumann programming languages")? See, for instance, [1].

Is there someone, more knowledgable than me, that could make the link in this article with Backus and others, and the von Neumann architecture or languages, or refute this link I'm seeing altogether? Hulten (talk) 11:57, 25 July 2018 (UTC)Reply

References

  1. ^ [theory.stanford.edu/~aiken/other/backus.pdf Can Programming Be Liberated from the von Neumann Style?], John Backus, Communications of the ACM, Volume 21, Number 8, August 1978

Xerox machines

edit

I have been doing some reading recently regarding the Xerox Dorado, Dolphin, and Dandelion and it appears to me that they are sort of a unique case. They are not all the same architecture. It is not clear to me whether the Dolphin has the same processor as the Dorado but with different software, but the Dandelion clearly does not have the same processor as the Dorado. All of them are clearly descended from the Xerox Alto, having user-programmable microcode; the Dorado and Dolphin at the least appear to be partially backwards compatible with it. To me Lisp machine implies that these machines were principally designed to run Lisp, which does not appear to be the case. Lisp was but one set of microcode that could be run on the machine, and appears to not be the primary concern of the manuals, which appear to have been written with the assumption that Mesa would be the primary operating environment and spend about as much time talking about Smalltalk as Lisp. --Blah2 (talk) 16:35, 25 April 2021 (UTC)Reply