> Portable Genera, an official port of the VLM to Intel and ARM done under contract for existing customers. While this version isn't publicly available as of this writing, it's still actively developed.
The jury is still out of how much official this is, seeing that the legal situation for Genera is a quagmire.
Wilder is that the MIT Lisp Machine branch is also being actively developed, and it has a clear legal situation today. To the point that there are several (simulated) CADR's on the Chaosnet talking to each other over the Internet. :-)
Build quality has diminishing returns. Rolexes and Bugattis are impressive feats of engineering, but I was aiming for the most utility per dollar, which would probably peak at a PCB, mechanical switches, group-buy dye-sub keycaps, and 3D printed cases.
My evil plan would be to go on for other types of incredibly rare hardware that could run in emulation while the physical interface with the human can be replicated with reasonable accuracy with modern technology.
A PC keyboard is not ideal for a Xerox Star, or a Symbolics, but is quite fine for an Amiga, Atari ST or Archimedes
My intention was both snark and to point at real vs perceived value. Whoever downvoted me seemed to miss that. HN is not very perceptive; this is not a very nice place.
Anyway, yes, small-batch artisanal handmade keyboards cost a lot of money.
> Build quality has diminishing returns.
Of course it does. But the point where the gradient tilts might surprise you.
I have tried a Unicomp modern replica of an IBM Model M. It was not a very nice keyboard, IMHO. It felt and sounded cheap and plasticky, with poor quality mouldings, rough edges, and so on.
This was a $150 replica device. It did not feel like hundred-and-fifty buck hardware. It felt a bit cheap and nasty.
(The owner told me it died not that long afterwards.)
This is the problem: proper quality kit costs, even if it's a modern reproduction of a mass-produced item from 30-40 years ago.
I agree: it's a ridiculously expensive keyboard. But the cheaper repro I tried disappointed me badly.
I have a Unicomp PC122 and it feels solid, but it sounds a bit like late IBM, much more than 1980's IBM. I heard the newer units are heavier, but I don't know much about that - I suppose they might be using a heavier steel plate, because changing the molds would be both ridiculously expensive and would invalidate their claim to be building original IBM keyboards.
In any case, the layout is more important than key feel (unless it's absolutely terrible) - for many machines you NEED those keys and, if they are in the same positions than they used to be, the experience is a lot better.
Is there any good literature on building keyboards? I’d love to try my hand at building vintage-like keyboards for things like DEC terminals, workstations (Xerox Star and Apollo Domain rely so much on their keyboards it’s hard to experience them on regular PC keyboards).
I've just completed a keyboard inspired by the Lisp machine keyboards but aimed toward use on a modern Unix system. Making your own one-off hand-wired keyboard is time consuming but pretty easy if you can solder. I probably spent $400 all told, including a fully custom keycap set from https://fkcaps.com/custom/
I've been meaning to write a blog post about it but I'm still finishing shakedown (currently chasing a wiring problem that messes up multi-key chords on the homerow)
I joined a discord for keyboard makers. I'm trying to learn electronics and circuit design and thought it would be a nice way to get into it.
The conclusion I drew after researching the subject and talking to people was it's not viable unless you're planning to turn it into an actual product and sell large numbers later on. Looks like the manufacturers are assuming you'll follow up with large orders.
No idea, this is sorta the project that runs on fumes right now. Wanna do something together? The keycaps are always the issue, making, remaking the PCB for the keyboard is just a days work even making it work with something like QMK or some Arduino thing.
But then you want the keycaps, and each special keyboard has its own set .. double injection moulds are expensive unless you do a large batch.
I know very little about keyboards (not much beyond reading a matrix, which is something I did once, 30+ years ago), but I’m in.
I would imagine by now there would be a KiCAD wizard that would generate a standard keyboard with extra rows and columns if needed and all customisation would be dragging the switch footprints to their positions for the non 1:1 keys.
Keycaps can be blank, laser eroded and resin filled, dye sub, or any other technique.
injection molding is intractable for small runs, but there are services for one-off UV printing and dye sub; you'll have to compromise on accuracy to the original, but could still make something nice
sorry for the late response, I've used WASD, there's also Max Keyboard, and I've heard some good things about Goblin Techkeys. They'll all do one-offs.
By compromise on accuracy I just mean they won't be the chunky doubleshot ABS keycaps like the originals. You can definitely still make something that looks nice and has the symbols you'd want.
When I was working on trying to update the CADR software, I found it helped to check in to a VCS all the versions of a particular source file over the previous version, what was your reason for going back to checking them in separately?
So many mysteries: which customers, and which customers with the capacity to pay not only for an ongoing service contract but a port? Some unknown wealthy benefactor? Someone managing 30 year-old ICBMs?
There were some financial giants that used Symbolics for fraud detection; possibly it just makes sense to keep it running than to rewrite it. I remember American Express being a large customer, from what I read.
From what I recall of various discussions over time:
Around 2001, AMEX fraud detection system was still developed on Genera, then OpenGenera running on Alpha workstations, but for deployment it was compiled using Franz's Allegro Common Lisp to run on "normal" servers.
About 2014-2015? I remember seeing more than 100k USD in publicly visible support contracts for US federal agencies related to Symbolics systems, which might have been related to the part stock and repairs done by symbolics remnant mentioned in the article - I know for certain that repair work continued to happen until at least 2018 (I talked with one of the people doing the repairs on contract).
There does seem to be a small residual market, enough to support a little pro work. "Who" and "where" might be confidential, but "how big" and "why" or "what" could be fascinating.
> Portable Genera, an official port of the VLM to Intel and ARM done under contract for existing customers. While this version isn't publicly available as of this writing, it's still actively developed.