We may potentially be in for some big changes in Fedora 44, with plans in place being discussed to drop 32-bit multilib / i686 packages.
I hope they remember what happened when Canonical planned the same for Ubuntu, when Valve jumped in to note they would no longer recommend Ubuntu and then Canonical backtracked on it. Linux distro developers may not like it, but Steam is huge and requires 32-bit to work properly for the client and for Proton / Wine.
This is only a proposed change for Fedora 44 onwards, it has not yet been approved. They're having a vote on the Fedora Forum where the change is being discussed. Even if it's approved, Steam wouldn't be the only problem, there's no doubt various other apps, tools and games that would break with such a huge change. At least with the Wine 9.0 release, WoW64 saw a huge improvement for running 32bit games on a full 64bit system.
The Fedora developers do at least note in the proposal that Wine and Steam are part of the dependencies on this, so they are thinking it through and aware of the issues.
Changes like this might push Valve to move a little quicker to adapt Steam to full 64bit, but I wouldn't expect anything ready any time soon. Especially when you look at Valve's own stats where Fedora doesn't even register in the top 11 distributions used on Steam. Although, we don't know what distros make up the 7% for the Steam Flatpak - but that's not supported by Valve anyway.
What are your thoughts?
Update 12:52 UTC - in response to some replies on their forum, the developer who proposed the change made it clear that eventually it will have to happen:
This is why it’s a proposal, and why I filed it more than 6 months earlier than strictly required.
But just to clarify - we will need to drop support for 32-bit x86 at some point. It’s dead, and more and more software just doesn’t support being built and / or run in 32-bit environments at all.
Yes, some things will stop working. But I hope that we can provide solutions and / or workarounds for most use cases.
And it’s better to start planning for the removal of i686 packages now than when (insert foundational package here - for example, CPython) stops supporting 32-bit architectures and we need to scramble to adapt.
But there is seriously no good reason why 32 bit applications should still be a thing in 2025
They shouldn't be made, or they should cease to exist?
Flatpak Steam has a whole bunch of its own issues and is not supported by Valve, it's not a direct replacement.
The RPM package is not supported either.
Valve made some non-official contributions to the Flatpak package and nothing for any other than the official Deb for Ubuntu and Steam client implemented in SteamOS.
no good reason why 32 bit applications should still be a thing in 2025Who says? The followers of a progress-first ideology that was founded on maximizing corporate profits at the expense of all other benefits? All that has done is to create a throwaway culture that fails to value anything from the past.
This is not supposed to be what drives Linux development. Torvalds has always said that what makes Linux stand out is it's backwards compatibility. Why should Linux ape the mentality of the for-profit sector?
Valve should go full 64bit, and drop/integrate the 32bit requirement/legacy.You may feel this is a good idea but that does not mean that Valve agrees. There could be a very good reason they have not done this (don't ask me what, I have no idea myself).
They shouldn't be made, or they should cease to exist?
Other than unmaintained legacy software, they should not exist anymore. There is simply no good excuse for not having ported your application to 64 bits, if you're still working on it in 2025.
Other than unmaintained legacy software. . . Of which there is probably quite a bit. Specifically relevant to this forum, nearly all older games are "unmaintained legacy software".