What is the closest one can get to using Genera (without actually using Genera) nowadays?
One could use mezzano, but I don't know if it's "production" ready yet.
I would even use something Unix based, I just want to use an useable, preferably FOSS system that is as close as possible to Genera.
Any advice?
Squeak Smalltalk and Pharo.
>>2 I like Smalltalk but was looking for something written in Lisp.
Mu current setup is Fedora Linux (soon to be Guix) + StumpWM and Emacs.
It's good, but I wish I could get closer to Genera somehow.
I'm not sure such thing would be very doable with what currently exists. McCLIM offers counterparts to miscellaneous Genera utilities (the command listener, debugger, inspector, etc), and CLIM itself was Symbolics' attempt at a portable, standardized Dynamic Windows; running them in a single image, and treating the command listener (with a few commands to replace UNIX shell miscellanea and whatnot) as your shell, and whatnot could bring you an inch closer, but there's barely anything written for CLIM, or cooperating with the premise, and even what currently exists doesn't fit together as cohesively as the Lisp systems did (though there's definitely work being done).
Another option I see could be Emacs with EXWM, and trying to do everything with Elisp utilities. Emacs mostly fits the bill for a single address space, image-based, largely introspectable environment, and there's a surprisingly large amount of Elisp utilities for all sorts of tasks; but compared to Genera it's still kind of pathetic.
Much of what makes Genera (and the Lisp Machine systems in general) magical isn't really skin-deep behavior, but rather a consequence of fundamental system design characteristics (single address space, image-based, introspectability and openness, the pervasive presentation-based toolkit, etc) that aren't really possible to replicate by haphazardly throwing together a bunch of loosely-coupled hacks on top of UNIX. You can try though, and I would be interested if anybody gets anywhere significant.
I just want to use an useable, preferably FOSS system that is as close as possible to Genera.
Why do some people have nostalgia for inferior operating systems? Are you a retrocomputing enthusiast? Present-day UNIX and even Microsoft Windows have so much more functionality than the pitiful Genera OS and those "Lisp Machines" that have been rightfully relegated to the dustbin of history.
>>5 What is it about using Lisp Machines that you disliked the most?
>>4
Perhaps it would be better if it was all virtualized. To have everything contained on a single Common Lisp image pretending to be an OS of it's own, and the underlying os would just be it's "hardware" of sorts. Would this be a possibility?
>>7 That's also something I've thought about, something like the Smalltalk environments or the Inferno and Oberon operating systems but in Lisp.
This way it could run on any kind of hardware in a virtualized form but could also run on certain bare-metal hardware (just like Oberon).
>>1 you could use the interlisp system with it's emulator.
Guix + exwm + Nyxt browser?
>>10
A lisp machine is not just a system that runs a lot of lisp.
Why do we even need these obsolete Lisp machines when we already have superior C machines?
>>12
Can you edit your kernel and apply the changes while your system is running?
Can you recover from system crashes using the debugger?
Can you save the whole system state (process tree in the Unix world) to a file and restore the sate or any previously saved state whenever you want?
These are just some of many things that you could do with Genera.
Anyone here got a MacIvory, or an Alpha station running the (more) reliable Genera emulator?
༼⍨༽
>>13
Does it have superior hardware too?
For example:
* Can you upgrade your CPU while your system is running?
* Can you replace your motherboard, RAM, hard disks, and power supply unit while your system is running?
No, today's hardware is obviously superior but I'm talking about the OS, the OS or at-least one that's like it running on modern hardware would give you the best of both worlds.
>>16
s/you/applications/g
>>18
What?
>>19
I'm guessing >>18 is referring to >>16 (who is in turn making fun of >>13, who is answering >>12). The "joke" is that LISP machines couldn't upgrade their CPU themselves. Which actually makes me wonder if anyone has ever tried to make a fully mechanized and automated distributed computer that would be able to tolerate the online upgrade of it's own hardware.
>>13
I've just imagined a browser which supports JavaSc^W GeneraScript and when you open something like https://twitter.com in it...
>>16
IBM mainframes allow CPU hot-swap. I've never seen anyone doing that, though.
>>16
Phantom OS can do that. Though you need to "pause" (shutdown) it before upgrading.
>>23
Maybe we could have "cluster" computers that run something like an Erlang runtime. Then all of >>13 and >>16 would be possible.
Are disposable machines really the ideal?
>>24
We need an intergalactic computer cluster to achieve full redundancy.
I tried this the other day running Nyxt and StumpWM from the same repl. Nyxt froze and I couldn't kill the thread from the repl. Then StumpWM froze and I had to switch to a different TTY and kill the whole xsession... That was fun, but there's definitely some work that needs to be done.
The properly lispy thing would have been to debug Nyxt from the working repl, but the thread in question was non-responsive, so I don't know how you'd debug in the thread. (I'm a noob though who mostly hacks in Clojure, so maybe there was a better way).
>>4
RISC-V and ArmV8.5 both have tagged memory extensions. I think that's really rather exciting from the perspective of future lisp-machines.
Some of the stuff you mentioned is outlined here: http://metamodular.com/lispos.pdf
I think it's pretty cool and I have some ideas on the topic, though I can't really start in earnest until current projects are complete. At least some of what made the LISP machines possible was simply the amount of money thrown at them, and the level of talent which ended up working on them. It was the early google of its day in a lot of ways, in that hackers would happily want to work on the development of LISP machines. Now, it's a bunch of oddballs on obscure web forums.
Oh well, if LISP is really as powerful as we think it is, and it really gives a hacker the leverage we think it does, then why shouldn't a bunch of oddballs on obscure web forums be able to build their own LISP machine? If it can't be done, then was it worth all the trouble to pine over it? And if it can, then we should do it.
https://github.com/gas2serra/mcclim-desktop
Still not a lisp machine, and I haven't even tried it, but I thought it was interesting nonetheless.
>>10,11
To add to that, EXWM/StumpWM and Nyxt are more Emacs-like than Genera-like, and Guix feels very UNIXy.
>>33
What would one need to add or change about StumpWM to make it more Genera like?
Remove unix.
>>35 How do I do that?
>>36 Good question?
>>34
Rewrite it from the ground up and use McCLIM.
i would like to be able to use genera fonts
there was a guy named "jthr" who had them
but the bastard's github is gone for eva .-.)
>>41
lol, are bfd ≠ bdf?
https://archives.loomcom.com/genera/symbolics/sys.sct/x11/fonts/bdf/genera/
https://github.com/Interlisp/medley Probably the closest thing to Genera which is FOSS and being actively maintained. Next closest thing would probably be Macintosh Common Lisp if you could run it under emulation. Besides that, idk, your best bet really would be just setting up Genera under VLM on Linux.
(And before you ask about the MIT CADR code dump, pretty much useless except from a historical POV. Same thing with LMI and TI Explorer system software)
Running Open Genera 2.0 on Linux
https://archives.loomcom.com/genera/genera-install.html
>>45
Thanks, I was actually looking for this.
What are your opinions on StumpWM, anons?
>>47
Fairly buggy, it feels like some features have been tacked on without any real testing, and then nobody ever used them, so they stay faulty. Frames are interesting as a concept, but ultimately a pain to use in practice. Iffy support for truetype fonts, and there's no granular control over font and its properties in different components (like message windows, toolbar, and so on). Overall, it seems to me like the design is not really good (though it might be just my personal bias), and there are not enough contributors to fix all the issues. Furthermore, the focus of the most active contributors is to add more stuff, that is also potentially broken, just adding to the pile.
Note that the last time I've interacted with it was probably months ago, so you'd need to evaluate whether any of what I've said is still true yourself.
>>48
Oh, but to not be entirely negative, its keybind system is really good. Much more flexible and powerful than anything I've seen in a similar program to date.