sofiechan home

Hundred Rabbits' "UXN" ultra-portable personal computing stack

anon 0x76 said in #776 1y ago: 77

(https://100r.co/site/uxn.html)

They made a simple VM that can be implemented in a weekend by almost anyone that can run their whole suite of personal computing, artistic, gaming, music, creative, etc apps. This is one of the coolest projects out there and it proves that computing could be radically more resilient to economic decline etc.

Their stack is designed for their needs and interests (games, "roms", apps), and doesn't match mine at least. Devine (the designer of UXN) thinks everyone should have their own. I want us (vaguely) to build our own such computing stack at some point. I suspect we have unique needs and would rather escape the metaverse like 100r at some point.

So let's have a thread about radically independent computing stacks and how we might want to design ours if we did. To start us off, I'll suggest that we may be particularly interested in how it could facilitate high-agency computing by a community of people seeking relative sovereignty and material independence. What that means in practice IMO is some of the stuff I mentioned in the other thread on personal computing (reading, writing, discussing, drawing, calculating, automating) especially in its social aspect. I suspect that if I get around to doing this, it would be in the context of giving sofiechan and our community a more solid basis for our communications and social governance infrastructure than the web. What stack would it take to provide highly resilient sovereign communications on basically arbitrary hardware and over arbitrary high-latency low-connectivity post-Internet links?

The other personal computing thread:
>>772

They made a simple V 77

anon 0x77 said in #781 1y ago: 22

UXN, for example, does not seem to support reading PDFs [0], and so is basically useless for research-dependent writing.

[0] https://github.com/hundredrabbits/awesome-uxn#applications

referenced by: >>782

UXN, for example, do 22

anon 0x78 said in #782 1y ago: 22

>>780
>>781
I don't know. It doesn't seem to hold them back. Maybe they just need a quick PDF converter to get images and text out. But a VM designed to run simple applications doesn't need to support every convenience of the modern bloat borg to not be "useless". There are a lot of things it doesn't do. The point of it is the things is does do. An e-reader does PDFs. As long as you can move files around with unix or whatever, why do you need your app vm to deal with PDFs?

I don't know. It doe 22

anon 0x79 said in #783 1y ago: 22

Again, I do not mean to quash efforts at first principles experimentation, but every use case described thus far seems better addressed by a _practice_ rather than a tool.

Because, on the one hand, the technological affordances of networked information ingest strongly support the goal of deep engagement with a subject matter of interest.

https://www.robkhenderson.com/p/how-i-read

> “Almost every idea that you have is downstream from what you consume. When you choose who to follow on Twitter, what book to read, what podcast to listen to, you’re choosing your future thoughts.”

Thus, you want to be able to opt-in to the broader networked knowledge web, which requires a hardware and software stack that recapitulates no little of all existing personal computing technology.

On the other hand, for those times when you do not want to be networked, to immerse yourself in fully local thought, extant hardware and software is fully capable of receiving vast productivity bandwidth - emacs is a flutter of the fingers away, and is available on all major platforms.

I do not think that undertaking an OS and networking project has expected positive return _with respect to the goal of deep thought productivity_, moreso than learning an existing tool and wielding it with exacting discipline.

However, if the goal is to _erect a network barrier_, then yes, the massive interoperability friction and costly investment of a bespoke software (to say nothing of hardware) system is of tremendous utility - you just very likely may find yourself on the outside of that barrier as well.

> An e-reader does PDFs. As long as you can move files around with unix or whatever, why do you need your app vm to deal with PDFs?

Why do you need the application VM if it doesn't provide the applications you need?

That a twee they/them artists duo need not read PDFs (yet still presumably watches sailing videos - on their phones, I expect) does not seem an argument against the ball of mud, but evidence that most of their impressive endeavors (i.e. circumnavigating the Pacific as amateur sailors - their games, music, art, etc. are not impressive) could have been accomplished with the existing consumer hardware and software stack.

referenced by: >>794

Again, I do not mean 22

anon 0x7a said in #784 1y ago: 22

> To start us off, I'll suggest that we may be particularly interested in how it could facilitate high-agency computing by a community of people seeking relative sovereignty and material independence.

> What stack would it take to provide highly resilient sovereign communications on basically arbitrary hardware and over arbitrary high-latency low-connectivity post-Internet links?

Decentralized p2p sofiechan instances installed by default within the OS, with the post database integrated into the file system so that conversations could be grep/awk/sed.

referenced by: >>794

Decentralized p2p so 22

anon 0x7b said in #785 1y ago: 22

These are not proposed solutions, just data points for learning:

New OS built in Rust:
https://www.redox-os.org/

Minimal person server built on Linux:
https://sive.rs/ti

Minimalist Linux distribution:
https://www.alpinelinux.org/

referenced by: >>788

These are not propos 22

anon 0x7c said in #788 1y ago: 11

>>785
Redox is where I would start if I was committed to dropping Linux while still not separating myself from all existing software like Urbit. You want libc & partial POSIX integration.

referenced by: >>794

Redox is where I wou 11

anon 0x7d said in #794 1y ago: 22

>>783
> most of their impressive endeavors could have been accomplished with the existing consumer hardware and software stack.

Read their explanations and try to rise above basic HN pooh pooh. The existing stack failed for them and they tried a bunch of wacky stuff and settled on this interesting result. It's not supposed to be impressive, it's supposed to be sustainable for them in their circumstances. It's just interesting to see how radically they could compress their computing stack down to something that could be implemented in a week or even weekend on basically any hardware. That is legitimately cool.

>twee they/them
You're just mad.

>>784
Maybe. I think basically the way this would work is you occasionally sync your local copy of the db with the server. It's not fully p2p because there's actually an inherently centralized trust accounting service sofiechan is built around. What's p2p is the network routing over various kinds of async tunnels.

Also, re: sed/awk/grep I think we need a new platform that isn't the web and isn't unix, but somewhere in between and radically simpler than either.

>>788
I'm not that interested in dropping linux in general. Linux can get so small and is a pretty good way to make arbitrary hardware available. Rather I think we want a platform on top of the true metal-interface OS that focuses on providing the right portable programming and UX model. Like the web, but again, not the web.

referenced by: >>801

Read their explanati 22

anon 0x7e said in #801 1y ago: 22

>>794
> I think basically the way this would work is you occasionally sync your local copy of the db with the server. It's not fully p2p because there's actually an inherently centralized trust accounting service sofiechan is built around. What's p2p is the network routing over various kinds of async tunnels.

By "occasionally sync your local copy of the db with the server", do you mean that the central server is aggregating the local views of the network available to each client, or that the server is the primary holder of all traffic, which the clients then mirror?

> You're just mad.
I am certainly not maddened by the twee they/thems, but neither do I think philosophy can be separated from the philosopher.

> Read their explanations
I have, and find it interesting to compare their intellectual journey with that of Stanislav Datskovski of Loper OS. While a Borgesian picture book for children [0] is a charming project, the means of its production is not the razor sharp tool shaped for a maximally agentic mind.

[0] https://100r.co/site/thousand_rooms.html

referenced by: >>802 >>803

By "occasionally syn 22

anon 0x7f said in #802 1y ago: 11

>>801
In addition to Loper OS (http://www.loper-os.org/?p=284), two other projects of interest for portable, personal, resilient, empowering computing that is actually powerful, not the first three at the expense of the last [0]:

Cosmopolitan Libc: https://justine.lol/cosmopolitan/

> Cosmopolitan Libc makes C a build-once run-anywhere language, like Java, except it doesn't need an interpreter or virtual machine. Instead, it reconfigures stock GCC and Clang to output a POSIX-approved polyglot format that runs natively on Linux + Mac + Windows + FreeBSD + OpenBSD + NetBSD + BIOS with the best possible performance and the tiniest footprint imaginable.
> Your program will also boot on bare metal too. In other words, you've written a normal textbook C program, and thanks to Cosmopolitan's low-level linker magic, you've effectively created your own operating system which happens to run on all the existing ones as well.

Reproducible builds of the Guix System from a reduced binary bootstrap seed: https://guix.gnu.org/manual/en/html_node/Reduced-Binary-Seed-Bootstrap.html

From 245Mb down to 60Mb counting, until the only thing that one will need to trust is a Mes binary (https://www.gnu.org/software/mes/), itself formed from 5,000 lines of Scheme code. This is more than a weekend's implementation to reproduce, but it yields much more than a weekend project's worth of return on investment.

https://guix.gnu.org/en/blog/2020/guix-further-reduces-bootstrap-seed-to-25/
> The Stage0 project by Jeremiah Orians starts everything from ~512 bytes; virtually nothing.

[0] I categorically reject "we need to develop minimally featureful, efficient software to moderate our energy expenditure to be good environmentalists" - features should not creep, and tools, whether general or targeted, should be properly scoped to their purpose. Efficient software can be justified by SWAP [1] constraints for missiles, or because stealth is advised, or so as to take advantage of Jevons Paradox and enable the doing of more things for the same wattage. However, the overall thrust ought be towards ever greater energy consumption, until we force an advance to the next Kardashev level (which is to say, the first [2]), and the concomitant fueling of ever greater means, adroitly matched to more magnificent ends.

[1] https://www.infinitioptics.com/glossary/swap-size-weight-and-power

[2] https://en.wikipedia.org/wiki/Planetary_civilization

In addition to Loper 11

anon 0x80 said in #803 1y ago: 22

>>801
>the server is the primary holder of all traffic, which the clients then mirror?

This. The server has the authoritative state, everyone else syncs to that, but only the part they are permissioned to see.

There's room in the world for actually decentralized comms, but that's a different layer. A public square (sofiechan) inherently has central governance. What philosophical discourses you have in your own space (decentralized DM groups, self-hosted Akademia, etc) is up to you. I basically want a decentralized protocol for peer-to-peer data syncing, but the thing you actually do with that is have particular hosts host particular conversations. Sofiechan is unapologetically monarchist in being a particular hosted conversation.

>neither do I think philosophy can be separated from the philosopher.
Good point.

>Cosmopolitan Libc: https://justine.lol/cosmopolitan/
Tunney is truly a dark wizard.

>Guix etc
That's neat

>Efficient software can be justified by SWAP [1] constraints

I agree. If I'm going to take something into the woods for a week to do some writing and low-intensity computing, I want size, weight, and power to be very heavily considered as well as intelligibility. These all favor smol computing.

When I want to do big iron work, I would need big iron computing. Here's the thing though: I've almost never had the occasion to do any big iron computing because computer UX right now is not particularly well suited to easily specifying and feeding the computations I occasionally have in mind.

90% or more of the size weight and power of our modern computing stack is used for pure waste because of a growing tower of bullshit abstractions. Is the spec of the web longer than the bible yet? Who controls that? This is an unsustainable situation.

To get to better computing, I'm convinced we should go back to the drawing board with respect to the web. Again linux kernel and C and all that can stay for getting the most out of arbitrary hardware, but for most actual user-facing computing abstractions, something between the web and unix in nature but small like somewhere between uxn and Lua in complexity is called for.

referenced by: >>805

This. The server has 22

anon 0x81 said in #805 1y ago: 22

>>803

> The server has the authoritative state, everyone else syncs to that, but only the part they are permissioned to see.

Copy.

> To get to better computing, I'm convinced we should go back to the drawing board with respect to the web. Again linux kernel and C and all that can stay for getting the most out of arbitrary hardware, but for most actual user-facing computing abstractions, something between the web and unix in nature but small like somewhere between uxn and Lua in complexity is called for.

QSL, this charts a clearer problem statement. What kind of computations do you occasionally have in mind? Have you experimented with the "computational notebook" model of Mathematica, Jupyter, etc.?

referenced by: >>807

Copy. ... 22

anon 0x82 said in #806 1y ago: 11

Because the transition from "minimalist computing in the woods" to "big iron" is more spectral than binary, homogeneous scalability, without having to retrain on multiple radically distinct UI/UX, would seem relevant to any stack simplification project. At the hardware level this is computronium [0], but the affordance can be extended to software as well; Plan 9 exhibits, to a certain extent, this "transparent modularity".

Phenomenologically, a truly humane personal compute device should be a small slime mold membrane [1], that can be joined with other slime molds as needed, through no interface more complicated than a concatenative plug.

[0] https://en.wikipedia.org/wiki/Computronium

[1] https://www.sciencedirect.com/science/article/pii/S1567832610000287

referenced by: >>808

Because the transiti 11

anon 0x83 said in #807 1y ago: 22

>>805
I haven't played with mathematica etc enough. I really should, because that probably fulfils a lot of what I want, or at least tells me what ways other people have solved some of these problems.

I haven't played wit 22

anon 0x84 said in #808 1y ago: 22

>>806
Yes plan9 is a good model. Basically, you have a really simple user-facing abstraction of basic capabilities via "everything is a file" or whatever our equivalent is. That can be pretty simple, but can also interface to arbitrarily complex and heavy capabilities through the standard interfaces.

Yes plan9 is a good 22

anon 0x85 said in #809 1y ago: 22

see also on the general topic of minimalist computing: Suckless
https://suckless.org/

see also on the gene 22

You must login to post.