Turns out we don't have any postmarketOS device port for x86 except in the archived category, and therefore we don't have any users on this architecture. So we are now dropping x86 (not x86_64!) from postmarketOS edge to save some build times. It never was in a stable version.

https://postmarketos.org/edge/2026/09/10/Dropping-x86-from-postmarketOS/

@postmarketOS
I still think pmOS should offer a generic x86-32 image for the greater good of the project.

Working early/mid-2000s PCs are still pretty easy to find, and they fit well as "lowest common denominator" or "worst case scenario"-platforms (especially the netbooks).

cc @fun @justsoup

@moses_izumi @fun @postmarketOS I mean, Alpine still supports x86 (32-bit) just fine so

I booted it on my first PC last year (Athlon 64 in Socket 754, 2GB of RAM, Radeon X800) and it mostly just worked (though motherboard is in dire need of capacitor replacement because it's getting more and more unstable)

@moses_izumi @postmarketOS the thing is x86 support has been basically broken for a while and nobody has complained (at least if i understand this correctly). If someone wishes to get it running and maintain it (and there is actual interest/active users) then making prebuilt images again would make sense

ofc it's a catch-22 a bit but imho our focus and compute is better spent pushing the envelope forward than languishing in obsolete tech (especially when it's clear that the tech doesn't tie into our mission).

@elly @fun @postmarketOS
i just think pmOS would be in a better position if it started as an initiative to build up Alpine into a comfortable desktop OS, waiting a couple years before getting into phone ports.

if it did to Alpine what Ubuntu, Mint or Antix have done to Debian, if you will

@moses_izumi @elly @fun @postmarketOS in what sense? i'd agree that we want and would benefit from more desktop users, but we have been supporting (and many of us on the team) daily driving pmOS on our laptops and desktops for years now - with systemd even - the thing i think we are lacking is that last bit of polish and broader attention/interest (the installation process for desktops sucked and while it's much better now it still has rough edges)

other than systemd + musl being a fairly unique combo (for what value that adds) we just dont have anything to attract the wider desktop community, definitely something id like to see improved

@cas @elly @fun @postmarketOS
> we just dont have anything to attract the wider desktop community
If you can preinstall and integrate Waydroid, you can probably do the same with chroots and distrobox (to run glibc apps without falling back on Flatpak) at some point.

I'd also suggest becoming another "Lightweight Desktop Distro", but I'm fine with that being relegated to some branch/fork dedicated to supporting deprecated devices.

having Alpine ports of MX Linux's "MX Tools" suite of GUI-based configuration tools would be neat (frankly every Desktop Unix should have them)

Hell, you could probably adress most of this by pinning an FAQ document to the desktop :D

@moses_izumi @elly @fun @postmarketOS the question i gotta posit here is why? becoming "another lightweight desktop distro" doesn't speak to me as value we can bring to the community--to the world

embedded devices (including phones!) are where the unsolved problems are, there is a reason mainstream distros don't support android phones , the spins that are starting to show up like the fedora mobile spin and such are all built on top of work done by the postmarketOS community

@cas @elly @fun @postmarketOS
true.
I just think Desktop Linux has problems that could be solved in more flexible ways.

for example: pmOS could've bypassed the issue of GNOME and KDE needing silly workarounds to run on non-systemd setups, by building their own desktop environment around one of the standalone window managers (using MX Tools or YAST2 as a DE-agnostic equivalent of their respective Control Panel clones).

@moses_izumi @elly @fun @postmarketOS we much prefer letting our friends at kde, gnome, etc do the hard work of building GUIs, integrating systemd is easily the best solution here for us, we have a great relationship with all of our upstreams and in the case of systemd many of our long term plans to deliver on the security and reliability dreams we have depend on features that only exist in systemd

@cas @moses_izumi @fun @postmarketOS I disagree on systemd (hence why I have a sticker "systemdeez nuts" on my work laptop) but yes, no point in trying to starting yet another DE just to differentiate ourselves. Building a good UI is genuinely hard

Besides, desktops/laptops aren't even our main focus

@elly @fun @cas @postmarketOS
sure, but I still think porting Yast and/or MX Tools would be a good move.
and having a DE-agnostic frontend for APK.

i think the Flatpak thing might be a bigger mark on the project honestly.
it makes sense as a stopgap for stuff that isn't available on Alpine's repositories yet, but it's only suitable for on devices that already have large or expandable storage (kinda hurts Linux's reputation as a resource-light OS)

[having a dozen different desktops/windowmanagers is cool, but every distro rolling their own package format causes more problems than it solves]

good night for now

@elly @cas @postmarketOS @fun
will also mention @alexia's blogpost about package management

https://cyrneko.eu/posts/linux-app-distribution.html

https://shrimp.starlightnet.work/notes/aocva9gla9r1vxnx

@moses_izumi @fun @elly @cas @postmarketOS hah, this is nice, ive been on and off mentioning how much i like app bundles on here tbh,

@Li @moses_izumi @fun @elly @postmarketOS imho it basically boils down to scope, we are doing revolutionary work on embedded kernel and bootloader development and trying to vertically integrate a whole OS stack on top. For us to decide we wanna "fix the world" and diverge from the mainstream on the OS side would introduce a lot of additional complexity, we would need to solve a lot of difficult problems that we just don't have the skills in

part of the model we follow is to develop relationships with and collaborate with upstreams, and in general this is not at all exclusive to the mainstream (although as we focus in on milestones towards production quality this is a growing priority).

in other words, from the distro maintenance standpoint we have kept ourselves open to all kinds of implementations, from desktop environments to init systems, postmarketOS is almost entirely made of constituent parts that our community have contributed -- if someone wants to use the project as an experimental testing ground for some new app packaging and distribution tech then they totally can. Getting software packaged in Alpine and pmOS is impressively easy and the structure of testing/community repos already lays the groundwork for going from a poc to something we would ship by default.

that's not to say we are willing to throw out good solutions, and in the case of app distribution and sandboxing i think you'd have your work cut out to match or exceed Flatpak enough to justify changing the default (i have to admit, mainstream buy-in is a challenge to overcome and the increased friction of going against the current would have to be justified)

@cas @moses_izumi @fun @elly @postmarketOS mm but you haven't considered: what if i consider my desktop environment and my package manager being a sandboxing system and requiring 2GB to install a single program to be an anti-features actually.

app bundles are like the only MacOS thing i like tbh

@Li @moses_izumi @fun @elly @postmarketOS dont get me wrong, flatpak is far from perfect, and this in particular is a nasty issue

the solution of interest for the flatpaks devs seems to be to just have your OS be built out of the same runtimes, which js neat but, doesn't really work for us obviously

but why is this a problem that postmarketOS has to solve? like i just wanna make da phone run Linux, i don't want to take a moral stand against big bad app distribution or whatever, that only serves to aliendate potential users who dont care about these issues which hurts the distro, and attract an increasingly niche and opinionated community with ever stronger in/out group dynamics (see also, Alpine systemd discussions, devuan, etc)

@cas @Li @moses_izumi @fun @postmarketOS exactly this. Flatpak isn’t perfect, but at least it’s a solution.

I remember when I started using Linux circa 2008 and big companies had absolutely no interest in supporting it because of packaging format mess. AppImages were the first “solution” to that problem (and in my opinion they absolutely sucked).

These days if you want to make sure your application will run everywhere (even on musl-based system like Alpine/pmOS) you just ship a flatpak.

Hell, I’m a Gentoo user (w/OpenRC) and I use flatpaks for some stuff that I can’t be bothered with compiling (like LibreWolf, Chromium, slopware like OrcaSlicer etc.) and currently typing this from a Chromebook running Chimera. I could just click and launch Steam if I wanted to (which you couldn’t do without Chroot or Flatpak).

(I won’t because I need to grab my bike and change trains (RE51 -> RE60) in ~15 minutes but that’s besides the point )

@elly @fun @cas @postmarketOS @Li
the one thing I unambiguously like about Flatpak is that every application keeps its config files in /home/username/.var (makes it easier to find them casually)

@Li @fun @elly @cas @postmarketOS
the dark urge to solve Desktop Linux's sorespots by funneling everything through a Haiku compatibility layer

@elly @cas @Li @postmarketOS @fun
as for pmOS: there's nothing wrong with making a distro mainly focusing on hardware support, I just think that if said distro officially supports decade-old portable devices it should take care to provide software that works well on them.

on that note, I suppose postmarketOS could get big in the "desktop" arena by focusing on Chromebook users (including defaults that run well on the really early models).
I hear it already has fairly good support for the ARM ones.

re: the packaging situation, maybe the MUSL2 distros could have a cross-platform package manager of their own (not necessarily as the default, just a shared fallback one).

@elly I started liking Flatpak ever since I installed Steam as a Flatpak on Gentoo :3

@fun @moses_izumi @cas @postmarketOS @Li

@TheEvilSkeleton @fun @moses_izumi @cas @postmarketOS @Li I would use Steam in Flatpak but entire VR stack (monado etc) doesn't play with it that well so I just stuffed it in a docker container

@TheEvilSkeleton @elly @fun @moses_izumi @cas @postmarketOS @Li okay, but when can I install gentoo as a flatpak?

@TheEvilSkeleton

I love flatpak for this and for giving me a very easy way to use and compartentalise all the nonfree programs I (sadly) need while still customising all my free ones with use flags :D

@elly @fun @moses_izumi @cas @postmarketOS @Li

@karolherbst that is a cursed and horrible thought and i may want to hear more.

cclist @TheEvilSkeleton @elly @fun @moses_izumi @cas @postmarketOS @Li

@root @karolherbst we're one step closer https://github.com/axtloss/Linux-flatpak

@elly @fun @moses_izumi @cas @postmarketOS @Li

@cas @postmarketOS

> we just dont have anything to attract the wider desktop community

When Duranium is ready it'll be a *very* compelling option, especially for soon-to-be-former Fedora Silverblue users like me. It's robust and sensible, which is a low bar but every other distro seems to get it wrong in one way or another.

Robust:
- immutable, basically — doesn't randomly break
- minor updates are automatic because I don't care I just want it to work please
- (eventually) major version upgrades are couple of clicks, and actually just work without thousands of questions I don't care about (looking at you, Debian and Ubuntu); Silverblue gets this right

Sensible:
- the user experience is well-curated: plain upstream, and the app grid isn't full of Network Manager Advanced Configuration Properties Utilities, or a launcher for Terminal *and* one for Xterm for some reason (sorry, Debian and openSuse)… it's just apps, like what a human wants
- the app store Just Works for normal people (because it's Flathub); but nerds can use sysexts and coldbrew to do nerd stuff
- crucially, no commercial or AI bullshit (apart from Firefox) — Silverblue has been the least-bad option here, but cautiously; these qualities obviously aren't Red Hat's priority

All because pmOS's whole thing is to make a tool that's useful for normal people, isn't trying to milk them for profit, and actually just bloody works. The project's priorities are genuinely aligned with the user's, and this is rare and valuable!

(Shout out to @elementary which is also very close to being all of these things!)

@greytheearthling @cas @elementary @postmarketOS
so far, what does Duranium have over Silverblue on a technical level?

(not to discount project governance: for example I prefer Antix and MX Linux over Devuan as far as anti-systemd distros go, because those two have a vision beyond "Debian good, systemd bad")

(I've never tried an immutable system, but I briefly considered testing SecureBlue just to see how the INFOSEC sticklers have it)

@moses_izumi @greytheearthling @elementary @postmarketOS the point of duranium isn't in itself to compete with other immutable distros, it is just a natural evolution of the postmarketOS project to take our existing work in hardware enablement, testing, our strong relationships with our upstreams and take the next step in delivering on our mission to build liberatory technology for everyone.

the benefits of an immutable system are obvious, and while we aren't necessarily doing anything revolutionary here, we are bringing together our values in making mobile Linux a truly viable alternative to iOS/Android and implementing them with what we believe to be the modern best practice

on implementation details, there are of course drawbacks to immutable systems with developer experience being a huge one. This is one area where we hope to encourage improvements

@cas @moses_izumi @postmarketOS No, x86 was not dropped because it was broken. It was dropped because there aren't many x86 devices supported at the moment and so we were building packages for an architecture that had no device ports.

It will come back probably when I get around to doing a generic x86 port (which I actually was interested in doing).

> ofc it's a catch-22 a bit but imho our focus and compute is better spent pushing the envelope forward than languishing in obsolete tech (especially when it's clear that the tech doesn't tie into our mission).

the goal of postmarketOS is to extend the life of consumer devices. I really don't see why extending the life of old x86 devices wouldn't count? It's clear that it does tie into our mission, so?

@cas @moses_izumi @postmarketOS like, the reason we dropped armhf also had nothing to do with how old it is. It had entirely everything to do with maintenance of that arch not being very good, and very few devices to justify the trouble.

@moses_izumi @postmarketOS It will.

@elly @moses_izumi @cas @postmarketOS why isn't it our main focus? Our main focus is to extend the longevity of consumer devices, this can be desktops, laptops, phones, smartwatches, TV boxes, anything that can run linux really. Unlike what people say, we are not purely a mobile-focused distro

@cas @moses_izumi @postmarketOS It is important to read the context behind dropping x86: https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/9521

@cas @moses_izumi @greytheearthling @elementary @postmarketOS am I missing something here, I thought that pmOS was geared towards smartphones, why are we talking about a desktop?

@McNeely @moses_izumi @greytheearthling @elementary @postmarketOS phones are pur bread and butter, why wouldn't i want the same OS on my phone and computer though ?

@McNeely @cas @moses_izumi @greytheearthling @elementary @postmarketOS Historically it kind of was. But nowadays pmOS is about increasing the longevity of consumer devices in general, not just phones but also tablets, computers and even TV boxes.

@fun @cas @postmarketOS
Any default window managers in mind?

I don't know what era of 32-bit PC you're targeting, but it'd be sick if your port resulted in pmOS getting a complete desktop experience (complete with control panels and a frontend for APK) that works well on the original Xbox.
maybe even the late-1990s kitchen computers (i.e I-Opener, 3com Audrey, etc.).

@fun @McNeely @cas @elementary @greytheearthling @moses_izumi @postmarketOS but for normal computers, why use pmOS over Alpine?

@noisytoot @fun @greytheearthling @McNeely @elementary @cas @postmarketOS
In theory, pmOS should be able to provide better defaults for desktop/laptop use than Alpine.
Right now it mostly boils down to hosting disk images that already have GUIs installed (also Flatpak, because Alpine's repositories mostly consist of server stuff right now).

@fun @fun @moses_izumi @cas @postmarketOS still very sad about armhf :(. wouldve been happy testing with the pi zero w

tho im looking to see if i can modify pmbootstrap to make pure alpine imgs instead of pmOS imgs and reintroduce armhf that way

@moses_izumi @fun @greytheearthling @McNeely @elementary @cas @postmarketOS alpine has flatpak, I use it on my laptop

@moses_izumi @cas @postmarketOS Any you want: https://wiki.postmarketos.org/wiki/Category:Interface

@Logical_Error @moses_izumi @cas @postmarketOS I wouldn't exclude the possibility of reintroducing armhf again, but that'll need a lot of work to fix it in alpine first

@noisytoot @McNeely @cas @elementary @greytheearthling @moses_izumi @postmarketOS because it's pretty much a pre-configured user-friendly version of Alpine Linux, for these computers.

Plus running the same OS on both your phone and laptop is kewl

@moses_izumi @cas @postmarketOS Alpine seems to mostly target i686 right now (they said i586 but many packages disagree, so it's kind of a mix between i586 and i686 toolchain...)

@noisytoot

If you know the difference between Alpine and postmarketOS, then there's no difference, and maybe even better to use Alpine. Lots of people that have never heard of the projects, though, would have it a lot easier using postmarketOS than Alpine :)

@fun @greytheearthling @moses_izumi @McNeely @elementary @cas @postmarketOS

@fun @cas @postmarketOS
Always good to have more options for Pentium MMX-spec machines, though I assume most of the testing and usage will happen on early/mid-2000s laptops.

@fun @cas @postmarketOS
on that note: I don't know how common 2-in-1 laptops are in the community pmOS, but I personally see them as the ideal test platform and the key to the project's future.

- Components already have mainline support and the bootloader is just standard BIOS or UEFI, so you can focus all your effort on the user-facing software
- The multitude of built-in input devices (keyboard, trackpad, touchscreen, maybe even trackpoint) makes it easy to test all the different GUIs
- The smaller/cheaper models (i.e school Chromebooks) are usually stuck with a few dozen gigabytes of soldered eMMC: offering a good reason to rely less on Flatpak or offer an intuitive way of installing stuff to external storage

@fun @moses_izumi @cas @postmarketOS what’s broken? one thing i found: gnome is available for armhf, but gnome-shell isnt, so it’s unusable (i think). is there more that u know of?

i do have sway with gpu acceleration working :D. custom script tho, im going to see if i can bring it over to pmbootstrap :)

@moses_izumi @cas @postmarketOS we already support any x86_64 computer as well as a good number of chromebooks.

@pabloyoyoista @noisytoot @greytheearthling @moses_izumi @McNeely @elementary @cas @postmarketOS I heard people saying that pmOS is to Alpine what Ubuntu is to Debian (in a good way)

and I guess you could say that

@fun @postmarketOS I *was* planning on porting snaps 😉

@pabloyoyoista @noisytoot @fun @greytheearthling @moses_izumi @McNeely @elementary @cas @postmarketOS happily running alpine on my laptop and pmos on my phone

i find breaking changes to happen more often on pmOS than on alpine. for alpine i dont think twice when updating daily on edge or not updating for weeks at a time

@justsoup @postmarketOS apk add !snap

I have somewhere unfinished snapcraft builds for Alpine that I could try to find and publish

@justsoup @fun @postmarketOS

@pj @justsoup @fun @postmarketOS oh come the fuck on, first systemd now this? ;_;

@domi @pj @justsoup @fun uhhh we aren't actually gonna ship snaps lmao(?)

@cas @domi @pj @fun Yeah, I was joking when I posted that. Should have been clearer. I *could* theoretically (since the only thing we were missing was systemd), but why would I want to deal with Canonical-ware...?

@justsoup @cas @pj @fun it is genuinely not obvious when you joke about things like that, especially since pmOS has already adopted a bunch of things that I initially interpreted as a bad joke

@domi @pj @justsoup @cas On the remaining Ubuntu installs I adminstrate I explicitly blacklist snap, so implementing snap in postmarketOS will have to be done over my dead body.

@domi @cas @justsoup @pj also, we still have openrc if you don't like systemd :>

@fun @domi the amount of people very unhappy about that and rather force upstream alpine into systemd as well is still upsetting me. I went from liking pmOS and working on stupid ports occasionally to having lost all motivation because of this shit :/

@famfo @fun @domi who is forcing systemd into Alpine and where exactly?

@pj @domi @famfo No one is forcing upstream alpine into systemd?

@pj @domi @famfo at least none that I'm aware of? And at least downstream we do keep systemd packaging wholly separate to OpenRC, to the point where systemd stuff is in an extra-repos/systemd repo

@domi @famfo @pj but there are definitely still people interested in pmOS/OpenRC (including myself) so it will stay around