💾 Archived View for bbs.geminispace.org › u › gyaradong › 5913 captured on 2024-03-21 at 20:53:04. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2024-02-05)
-=-=-=-=-=-=-
My understanding is that a lot of the complexity in today's linux comes from cgroups. It's not bad or bloated, but the threshold for understanding is high.
2023-10-05 · 6 months ago
🦀 jeang3nie · Oct 05 at 15:06:
@dimkr I didn't realize you had an association with Puppy Linux. I learned how to use and abuse the shell on Puppy. Managed the first CE release way back in 2006 and was behind the Grafpup distro.
@jeang3nie Yes, I do, I'm a self-appointed "maintainer" of https://github.com/puppylinux-woof-CE/woof-CE, because nobody else seems to be interested in keeping Puppy alive by fixing years old bugs and migrating away from dead stuff like GTK+ 2, X.Org and aufs. I do some new development in https://github.com/vanilla-dpup/woof-CE/tree/vanilladpup-11.0.x and maintain dpup, which is the only viable path forward IMO.
You might like my story.
Ten years ago, frustrated by experiences like [1], and frustrated by my attempts to slice complexity off Ubuntu, I decided most software was parasitical [2].
After some time investigating Forth and OpenBSD and so on, I went off to build a new computing stack from scratch, intended above all to be easy to build, easy to run and easy to understand: Mu
I worked on this for 5+ years, paying particular attention to the number of languages I was using [reason]. For example, I think it's an abomination that Python requires gcc, and gcc requires Python.
I eventually gave up on this mostly because of the burden of building an OS and supporting hardware. I was kinda aware of the device driver problem going in, but I'd assumed I could build something lowest-common-denominator if I didn't care about performance at all. This assumption turned out to be optimistic. For example, I was reading and writing 1 byte at a time from disk and still only supporting 40% or so of hard disks. And there's also the problem of debugging on real hardware. I thought for a while that I was done after implementing something on Qemu, but this too was wildly optimistic.
These days I've become a little less dogmatic about a few things, and discovered a language I hadn't quite paid enough attention to back when I was researching Forth: Lua. My current setup feels like a good sweet spot between minimalism and capability.
— https://akkartik.name/freewheeling
🍄 Ruby_Witch · Feb 02 at 09:58:
@akkartik Just a random question that represents my own viewpoint: Why try to make your OS that you were writing have wide compatibility? I understand that it can be nice for other people to be able to enjoy and appreciate your work, but if it were me, I'd be focused purely on writing a system that was compatible with my own hardware.
Obviously, this kind of approach would limit adoption and outside contributions to a software ecosystem, but...so what? I guess what I'm saying is that everyone should have their own personal TempleOS if they're capable of doing such a thing. :)
Very interesting question. Probably just boils down to my own motivations. I never chased mass adoption but I also wasn't comfortable entirely foreclosing on the possibility. On the other hand, the hardware I built for was indeed precisely the hardware I had. So it didn't feel like the two were in tension.
Getting Sick of Linux — My pretty minimal installation of Xubuntu, running with dwm (no desktop or related crap) is beginning to feel like a lead weight. It is plenty fast on the old i5 ThinkPad (I've been scaling down - a couple of years I only used i7's), and I have few complaints, really. But I feel like I am drowning. It is running like 200 processes, without me doing anything taxing right now. I have no clue what 95% of these are. Some are downright scary sounding: idle_inject,...