Post Azs42tlQbv86obcFMG by dmbaturin@functional.cafe
 (DIR) More posts by dmbaturin@functional.cafe
 (DIR) Post #AzrqdehECEO1RKHgBM by hsivonen@mastodon.social
       2025-11-03T06:25:40Z
       
       0 likes, 1 repeats
       
       I dislike framing current Linux system on hppa, alpha, or m68k as reducing e-waste. If that was true, current Linux system on i386 (non-SSE) would be about reducing e-waste, but https://lists.debian.org/debian-release/2024/11/msg00459.html is the realistic take about that: There’s so much x86_64 hardware that’s about to go to e-waste unless someone accepts it for zero or near-zero money, that if e-waste is your concern, you should take an about-to-be-e-wasted x86_64 computer and retire the i386, hppa, alpha, or m68k hardware.
       
 (DIR) Post #Azrqdg2tBBJ7cnQUaG by hsivonen@mastodon.social
       2025-11-03T06:30:14Z
       
       0 likes, 0 repeats
       
       Running a current Linux system on hppa, alpha, or m68k isn’t about avoiding e-waste but about the hack value. If enjoying the hack value is your hobby, that’s ok per se. However, in practice, this often involves demanding that other busy people participate in your hobby when they don’t enjoy it as part of their hobby or job. That’s less ok.i386 was easier to retire, because it had been a collective thing instead of someone having been very invested in it as a hobby.
       
 (DIR) Post #AzrqdhBmva1JAgb75s by shironeko@fedi.tesaguri.club
       2025-11-03T11:50:25.491829Z
       
       0 likes, 0 repeats
       
       @hsivonen I know the ship as long sailed, but it's one of the few things that still makes Linux fun. Let's just make it official, Linux is about commercial interests and making money, hobbyist is only allowed as unpaid labor, your needs will not be heard.
       
 (DIR) Post #AzrtZvPwNwPm0I6wi0 by dmbaturin@functional.cafe
       2025-11-03T12:17:13Z
       
       0 likes, 0 repeats
       
       @shironeko @hsivonen I have a great admiration for those machines, but I can't get myself to advocate for mainstream support for them in modern OS versions.Running Linux on obsolete enterprise workstations and servers is a very niche hobby that is also mostly limited to a small number of countries where those machines used to be common and thus available locally without exorbitant shipping costs. For most Linux users in the world, hobbyists and professionals alike, they are practically unobtainable.If people want to do it, they can maintain their own patch sets or forks — it's FOSS.For my own hobbies, I'd rather see more improvements in the sound subsystem.
       
 (DIR) Post #AzrtZway0QpRemHGXA by shironeko@fedi.tesaguri.club
       2025-11-03T12:23:22.453378Z
       
       0 likes, 0 repeats
       
       @dmbaturin @hsivonen Firstly, those architectures are not that niche by hobbyist standards. Secondly hobby has a long tail, most hobbies are incredibly niche, so if this is the cutoff, it might as well be no fun allowed.
       
 (DIR) Post #Azs2XRkibfrv0YirqK by wolf480pl@mstdn.io
       2025-11-03T14:03:46Z
       
       0 likes, 0 repeats
       
       @hsivonenIMO keeping those platforms supported can be also seen as a protest against the "move fast and break things" attitude.Also, arguably, supporting more architectures prevents the software from growing too reliant on one arch's features, and creates a supply of develpers skilled in porting to different architectures, which makes it easier for new architectures to enter the market.
       
 (DIR) Post #Azs42tlQbv86obcFMG by dmbaturin@functional.cafe
       2025-11-03T13:55:31Z
       
       0 likes, 0 repeats
       
       @shironeko @hsivonen Are there statistics on user counts?
       
 (DIR) Post #Azs42usYSuQOGzxS6a by shironeko@fedi.tesaguri.club
       2025-11-03T14:20:40.324559Z
       
       0 likes, 0 repeats
       
       @dmbaturin @hsivonen random search for m68k http://www.linux-m68k.org/
       
 (DIR) Post #Azs4n6HiOIh8y6kTJ2 by ignaloidas@not.acu.lt
       2025-11-03T14:29:00.317Z
       
       0 likes, 0 repeats
       
       @wolf480pl@mstdn.io @hsivonen@mastodon.social You can certainly "move fast and break things" even while still supporting ancient architectures, so I don't really buy that argument?As for growing too reliant on one arch's features - I'd rather do that than be held back by another arch's lack of a feature.
       
 (DIR) Post #Azs88XqpOXYyPL5c5g by wolf480pl@mstdn.io
       2025-11-03T15:06:28Z
       
       0 likes, 0 repeats
       
       @ignaloidas@hsivonen You can certainly support the French government's policies even while setting cars on fire in the streets of Paris.But why would you do that?
       
 (DIR) Post #Azs8Yn25ZYJ11Ps51k by ignaloidas@not.acu.lt
       2025-11-03T15:11:14.966Z
       
       0 likes, 0 repeats
       
       @wolf480pl@mstdn.io @hsivonen@mastodon.social I don't see it like that?If I'm writing a kernel for antiquated systems, I see zero reason not to break stuff - nobody really uses them for anything serious anyways, this is all hobby stuff, and the people for who this is a hobby won't get too bothered by needing to change stuff if it's for the better.It's widespread, largely commercial use, that really cares that you don't "move fast and break things" - because that's work on their part.
       
 (DIR) Post #Azs97JC0Vbn2bsyDaq by wolf480pl@mstdn.io
       2025-11-03T15:17:29Z
       
       0 likes, 0 repeats
       
       @ignaloidas@hsivonen Yeah but if you run an existing project that has widespread use, and you adopt an aggressive "no regressions" stance, then continuing to keep support for a niche architecture shows that you really mean it, and you're willing to stick to it even if it's costly and doesn't bring any benefit to 99% of your users.If you run a project with widespread adoption, and intend to move fast and break things, there's no reason to support anything beyond x86-64 and arm64
       
 (DIR) Post #Azs9rZkDInGdGV225Y by ignaloidas@not.acu.lt
       2025-11-03T15:25:50.896Z
       
       0 likes, 0 repeats
       
       @wolf480pl@mstdn.io @hsivonen@mastodon.social I don't think dropping support for stuff that's barely used by anyone anymore is a sign of not having a "no regressions" stance? Like, FFS, Windows is quite heavily there because they do have a fairly strong stance on supporting old software - but they don't really care about the hardware in the same way, because hardware is relatively easier to replace and it happens anyways over time, because new hardware is faster, cheaper to run, more efficient, etc. Dropping old hardware support is just following what naturally happens. Itanium got dropped from Linux a couple of years ago, and nobody really cared - even though it was sold up to 2021 - because everyone moved away from it anyways. There are plenty of architectures in Linux that haven't been sold for longer than that, that absolutely nobody runs for anything besides curiosity - are they really worth keeping?
       
 (DIR) Post #AzsAC8jav5M3ZnY2dc by wolf480pl@mstdn.io
       2025-11-03T15:29:33Z
       
       0 likes, 0 repeats
       
       @ignaloidas@hsivonen Being costly is exactly what makes it a good signal.It demonstrates to your enemies and allies that you are have high resolve.
       
 (DIR) Post #AzsAJOKfKEnPynLUG0 by ignaloidas@not.acu.lt
       2025-11-03T15:30:52.856Z
       
       0 likes, 0 repeats
       
       @wolf480pl@mstdn.io @hsivonen@mastodon.social I have never seen spending time on useless shit (for business) as a good signal (for business), but idk, maybe for you it is
       
 (DIR) Post #AzsAb3a7DOmgpnzmrY by wolf480pl@mstdn.io
       2025-11-03T15:34:05Z
       
       0 likes, 0 repeats
       
       @ignaloidaswhy do you care about business?@hsivonen
       
 (DIR) Post #AzsAmw54vas12y8uJ6 by ignaloidas@not.acu.lt
       2025-11-03T15:36:12.808Z
       
       0 likes, 0 repeats
       
       @wolf480pl@mstdn.io @hsivonen@mastodon.social Because I don't care about moving fast and breaking things for non-business purposesIf I'm doing something at my own leisure, I'm down to change things up because of other people changing things up if I see that it's an improvement.
       
 (DIR) Post #AzsBi34pabflFcGD2G by wolf480pl@mstdn.io
       2025-11-03T15:46:31Z
       
       0 likes, 0 repeats
       
       @ignaloidasif I'm doing things at my own leisure, I often don't have time to update my stuff to the latest breaking changes in someone else's project. I have to pick my dependencies such that they won't make backwards incompatible changes too often. Or I have to vendor them, and make sure I can build them offline, so that I'm never forced to update.Otherwise, my personal infra would always be broken, and working on it would be very unfun.@hsivonen
       
 (DIR) Post #AzsEcn75oVtI9hVFLM by hsivonen@mastodon.social
       2025-11-03T16:19:11Z
       
       0 likes, 0 repeats
       
       @wolf480pl It should be possible to test that theory by assessing if keeping support around architectures like hppa, alpha, or m68k had helped the introduction of RISC-V or aarch64. It doesn’t look like it to me, but I haven’t looked carefully. *Maybe* the very weak ordering of Itanium helped aarch64.OTOH, the consolidation of machine features has enabled Wasm as an efficient compilation target.
       
 (DIR) Post #AzsFcaJzTRWMR608p6 by hsivonen@mastodon.social
       2025-11-03T16:14:45Z
       
       0 likes, 0 repeats
       
       @shironeko @dmbaturin Banner for Debian 2.2 and updated in 2000.
       
 (DIR) Post #AzsFcau9J0RcFEv1BQ by shironeko@fedi.tesaguri.club
       2025-11-03T16:30:21.241890Z
       
       0 likes, 0 repeats
       
       @hsivonen @dmbaturin lol, I do wonder how many of these machines get regular internet connection (if any)
       
 (DIR) Post #AzsHHoPG6wVFwlgiZs by parzivalwolfram@infosec.exchange
       2025-11-03T16:49:00Z
       
       0 likes, 0 repeats
       
       @wolf480pl Honestly, that ship has long sailed. A lot of hardware support for old i386 machines has already been stripped out of the kernel, so many can't boot it with the newest kernel anyway, and a lot of software will break on assembly on that old of a target as well, especially if they're already partially/wholly written in assembly (nettle, gnutls/openssl, libm, libffi, more) because they can't target literally every variant of i386 out there. Hell, the kernel lost mpcore support in 5.11 or so, which killed a lot of 32-bit ARM support (including for the 3DS port I was involved with) in one move. It's a little late to say "well it's a sign of good faith that they're not trimming the edges off the kernel" when it's had a large chunk taken off already.
       
 (DIR) Post #AzsHt7TUB9EDngSFn6 by wolf480pl@mstdn.io
       2025-11-03T16:55:46Z
       
       0 likes, 0 repeats
       
       @parzivalwolfram I wasn't specifically talking about the kernel, and neither was OP AFAIK.But yeah, this has been going on for a while now and it sucks.
       
 (DIR) Post #AztOJ2IIr6ge96vQH2 by ignaloidas@not.acu.lt
       2025-11-04T05:42:23.397Z
       
       0 likes, 0 repeats
       
       @wolf480pl@mstdn.io @hsivonen@mastodon.social There's a big difference between personal infra and hobby projects. I don't think a hobby project is worth doing if I would be bothered by a prospect of needing to completely re-do it, while for personal infra I have somewhat similar requirements as for business purposes, maybe even more extreme in some regards.And in my mind, hosting your personal infra on m86k is absolutely, certainly, a hobby project - no sane person would/should give up their time on setting up their personal infra on one of those, expecting it all to go smoothly, and then expecting it to keep running with updates and without problems.Like, I don't want my personal infra to ever run on architectures where you could go and write in your personal website "hey, this was served from an <arch> machine!" and it be impressive.Like sure, you can run your fedi instance on whatever weird machine - but nobody ever does for more than a joke.