This is exceptionally cool, but even for someone like me who has disposable income, a deep love of retrocomputing and several years of Lisp experience... it's just very expensive. For the machine, GPIO headers and USB keyboard support (really the bare minimum for interesting things) its over $300.
You have to really, REALLY love offbeat baremetal lisp machines to throw down that kind of cache for a system where your USB controller is probably more robust and powerful than the entire rest of the system put together. Hell, I'm pretty sure my TV's remote control clocks faster when it's on...
The price will come down, soon, as soon as I can build a batch of about 50. I'm thinking of creating a mass buy sign-up, so I can put in a big order and offer them at a reduced price. Or some sort of credit on future purchases for those who buy now. I'm open to suggestions, and ideas on what is an attractive/competitive target price.
I appreciate that you're doing small batch SBCs and that adds up. As an artistic accomplishment, I love your work.
I think that for me personally, the outer envelope for a SBA (with the ability to use a USB keyboard and some sort of minimal display attachment) is $150.
The GPIO expansion and the cute case are fine as secondary things and I wouldn't mind giving over for that.
I don't know if there is some sort of uart option for serial comms on the board, I didn't see that called out. But if there is you perhaps want to mention it is your option for folks buying on a tighter budget.
Yes, there is a USB/UART on the CPU card, I think I do mention it, but not as directly as I should. Once you have the expansion board, then real terminals or computers with serial ports are possible too.
If you like the idea of a small lisp machine, you should look at uLisp. Last time I priced out the components they came to somewhere in the $20 to $40 range.
I feel the same way. I've been watching this project for months. I really want to support it, but I don't think I'd get enough use out of it to justify the cost.
No. That's ridiculous. To some extent I want the fun of the retro machine without the hassle or $30k price tags of the days of old that helped inculcate that reverence in me.
I can't help but compare this to the Nvidia Jetson I just bought that can run SBCL just fine, costs 1/3 as much for the entire feature kit sans laser cut Labo-style keyboard, and uh... I could actually do things with. I don't expect to get exactly the same value from a limited run retrocomputing platform, but I at least expect full transparency on the source before I buy, and an emulator so if I do make something I like and the board faults out I'm not SOL.
Having a computer with a much simpler (like several orders of magnitude simpler) software stack has its own appeal.
Though I wouldn't buy it either, since I have already more computers which can run Lisp than I have time to. Including a Nvidia Jetson Nano and actual Symbolics machines.
(There's one that I'm horribly, horribly overdue to get back to the authors about archiving on the Internet, actually. Anyone who's based in North America and happens to have the hardware and the expertise to recover data from both 5.25" and 8" floppies in some CP/M format?)
The first Apple ][ LISP I ever played with was "MICROLISP", written in APPLESOFT BASIC, included with the Apple Software Bank Volume 3-5, published in 1978 (despite the Internet Archive listing saying it was published in 1974, which would have been impossible since the Apple ][ was launched in 1977).
Text APPLESOFT BASIC source code -- It's only 201 lines! Beautiful!:
60000 HOME : PRINT "MICROLISP/16-JUN-78": PRINT "COPYRIGHT 1978 APPLE COMPUTER INC"
60010 PRINT "IT WILL TAKE APPROXIMATELY 1.5 MINUTES"
60027 PRINT "TO INITIALIZE THE ARRAYS."
I can't find an MP3 of the original cassette tape it was distributed on ;), so you'll have to suffer with the DSK image file:
The following discussion assumes the user is already familiar with the principles and technical vocabulary of the LISP language.. The user should consider this document more as reference material on how MICROLISP differs from other versions of LISP, and what features it implements. A good reference for learning LISP is Weissman's LISP 1.5 Primer (see Bibliography). We also can recommend Friedman's The Little LISPer as an enjoyable introduction.
MICROLISP is a cassette-based microprocessor implementation of an interpreter for LISP, a programming language useful for symbol processing. The MICROLISP interpreter is written in APPLESOFT II BASIC. More complete documentation will soon be available when Hayden Press publishes a book on Microlisp by Ole Anderson. This book will include an annotated source listing of the interpreter, an explanation of its theory of operation, a detailed reference manual, an application section describing how to do graphics from MICROLISP using BASIC, a short LISP primer for the inexperienced user, and a bibliography of LISP literature useful to the
MICROLISP user.
Many programming languages (BASIC for example) have been created to deal primarily with numerical data, and programs written in those languages can solve a large range of problems. But there is an equally interesting area of programming that deals with symbolic data.
If you want to be "in" with the artificial intelligence crowd, you might consider learning a symbolic manipulation language. Symbolic data might be chess positions, equations (treated as equations, not as expressions to be evaluated), or natural-language sentences. LISP is a language for list processing (hence the name: LISt Processor). Lists are the data items of LISP, just as numbers are the data items in BASIC. Because list elements are symbols not restricted to numbers, LISP makes it possible to solve many types of symbolic manipulation problems as easily as simple numeric problems are solved in BASIC. LISP is used primarily for artificial intelligence applications such as processing natural-language input, performing symbolic differentiation and integration of numerical formulas, and proving mathematical theorems. MICROLISP is a version of LISP that will allow you to learn the elements of LISP by writing and executing LISP programs.
BASIC goes a long way toward making the power of the personal computer available to the average person, but it lacks several critical features found in many other high-level languages: naming of subroutines and calling them with parameters, use of variable names with more than the first two characters significant, recursive subroutines and dynamic allocation of variables. These features are all found in MICROLISP.
Implementing MICROLISP in BASIC means that LISP programs do not execute very rapidly. However, it makes it possible to see how the language has been implemented. On the other hand, APPLESOFT II supports real (floating point) arithmetic, string functions and trigonometric functions, and these are all available to the MICROLISP user (often they are not available in larger versions). Another convenient feature in APPLESOFT II is the ability to
PEEK and POKE memory locations, and to CALL machine-language routines from BASIC. This is often useful for interfacing special hardware and for applications where speed is critical.
Another reason MICROLISP was written in BASIC is for tutorial purposes.
Many personal computer users desire to learn about the techniques involved in implementing high-level computer languages. This program is an example of a complete language processor which accepts characters typed from the keyboard, builds symbols from them, and processes these symbols according to the syntax of the LISP language. Thus, it uses many of the important algorithms employed by all other interpreters, compilers and language processors. MICROLISP can also be used as an educational tool for introducing elementary list-processing concepts.
The "skimp" page also includes a list of several 8-bit LISP implementations, including several for the 6502 [1]. This discussion lists more [2]. One useful implementation discussion (of many!) is R. Kent Dybvig's "Three Implementation Models for Scheme". University of North Carolina Computer Science Technical Report 87-011 [Ph.D. Dissertation], April 1987 [3].
neslisp [4] is a small subset of Lisp that generates 6502 code (NES); the as of version 0.1 it is released under the GPLv2 license (and thus is open source software). An older version is available via neslisp on Google code [5].
Picobit [6] (GPLv3) looks interesting. You program in Lisp (specifically a variant of Scheme) on a larger system; it then compiles to a VM that runs on the target. Currently it does not support the 6502, but the system is designed to be portable and it looks to me that it'd be relatively easy to port to a 6502.
Lysp [7] (MIT license) is another small implementation of Lisp. It doesn't directly support 6502 and is in C, but it's small... so it might not be hard to transliterate. A related program from the same developer is Maru [8], which can self-host. This could be a reasonable starting point for creating a 6502 Lisp.
Turning the Apple //e into a lisp machine, part 1 [9] refers to this Lisp interpreter for the Apple //e [10] (MIT license). They also show a clever bootstrap approach: they send code via the audio jack (!).
There's more that can be done; Lisp was originally developed in 1965, on even smaller systems. The book "Lisp in Small Pieces", Program Transformation in Lisp [11], this Scheme page [12] and these details on implementation [13] are all of interest. R. Kent Dybvig's "Three Implementation Models for Scheme" [14] shows how to implement Scheme while still keeping most stuff on the stack. Tagging could be difficult, but an alternative (that I believe was done for early Lisp implementations) is to partition the memory into regions; to determine the tag, just check which memory region it is in. It might be possible to start with PicoLisp [15] to create a 6502 version. There is a trivial ACL2 implementation of Lisp (GPL) [16] that I converted into Common Lisp ( compiler [17], run-time [18]; See Wolfgang Goerigk's "Compiler Verification Revisited" [19] for more).
I know that there was an "InterLISP 65" available for Atari's 6502-based line.
Various full LOGO implementations were available as well.
Another 'bare metal lisp' is http://www.ulisp.com/ , which has some fun demos and is supported by different hardware, 16 or 32 bit. It's also a single .ino file, e.g. [1]
I've never heard of this happening before, and I can't even think of a way in which uploading uLisp could break an Arduino, but contact me directly (my email address is on the uLisp site) and I'll try and help you sort it out.
Related impractical fun: you could fit a small Lisp (such as standard R7RS Scheme) in some PC motherboard flash chips used for the BIOS/bootstrap, using Coreboot. Maybe ThinkPad X200 or a Coreboot-friendly Chromebook.
In the Lisp community, there's an infamous story of a Lisp Machine made out of Nokia Communicator (the early, DOS-based model with 80186 cpu) by Timo Noko using Nokolisp.
What happened was that the DOS flash image got corrupted somehow, so Noko implemented missing bits in his lisp implementation and made the phone boot directly into it. Since all the software he used was written in his Lisp, there was not much of a difference in usability :-)
I played with this thing for like a half hour at Vintage Computer Festival a few months back, it's a very cool thing. The computer packed fully in to a case built around a VGA display and usb ortholinear keyboard, and in to a shoulder bag. The "Lisp Machine" itself speaks serial in/out. The keyboard plugs in to a USB-to-Serial interface, and the screen in to a serial terminal to VGA emulator.
Cool stuff, and the creator's goal of creating a machine for interactive learning with a hand-coded scheme-alike is pretty wonderful.
EDIT: It looks like it went on sale two days ago and the full kit is already listed as out of stock. Hopefully they can increase stock. I've wanted one of these since I first put eyes on it.
I'll adjust the stock count on Tindie, I have a few more than what is shown there, but I am getting low and I have to order more parts soon. Thanks for your interest and positive comments !
I didn't make that github repo, I don't know who did. But it's old. The most recent archive on makerlisp.com is from a few days ago. Go there if you want the most recent stuff.
This is a snapshot of the source preview I put up just before my trip to the Bay Area Lisp & Scheme Users Group in February, before I had shipped any products. Apparently someone was just trying to help me get more exposure. I have no problem with this, the attributions and links are all correct and it is a proper snapshot. As a reading / demonstration example, to get the feel of the system, this is fine. But I update the archive on my website every few days, and that's where you can find the latest version.
There's an archive posted on the Download page of makerlisp.com with all the source and lots of examples. And you can run the code on x86/Linux to get a feel for the system.
I don’t think it’s a Minivan, could be a Vortex. If you search for “split spacebar 40 keyboard” you can see similar commercial and custom boards.
It’s what’s called a 40% layout (as in 40% of the keys of a standard 104 key board). To save space it doesn’t have a number row, you have to use a fn key to type numbers.
It’s not ortholinear, as another poster claimed. See the Planck keyboard for an example of that.
(PS If you’re the type of person who can fall down a rabbit hole for a subject with lots of lore to learn and that allows infinite customization, you should probably avoid the world of custom mechanical keyboards ;)
I am that kind of person and have already spent way too much money on this. By this time next year I’ll have an atomic purple Iris with mito laser keycaps that light up on individual key presses. I’ll also have an ergo infinity with symbiosis space cadet 2 keycaps. All that I’ve built myself, hopefully without issues.
The real kicker is the waiting. If delayed gratification made a noise it would sound a lot like cherry mx switches.
I chose this processor because it was available for under $10.00 and had the capability of addressing at least 1 Mb of external RAM. Regardles of architecture, when I searched, there were only two processors that I could buy on Digi-Key or Mouser that had these parameters, and they were both Z80 variants ! And then, bonus, it can also run retro software. It has been an exceptionally easy and capable processor to use. And although I'd always like more performance, for the kind of of software experience I'm trying to recreate, it's plenty.
That sounds better, but still pretty underpowered. I suppose if you wanted something high performance, you wouldn't want to use baremetal lisp anyway and this is more for fun and experimentation so a cheap and easier to understand processor wins.
You have to really, REALLY love offbeat baremetal lisp machines to throw down that kind of cache for a system where your USB controller is probably more robust and powerful than the entire rest of the system put together. Hell, I'm pretty sure my TV's remote control clocks faster when it's on...