Pragmatism will always gravitate towards the “middle”. But the middle can never be truly found by pragmatism unless it can see either extreme.
Without Stallman, the closed software extreme has a ton of supporters (Microsoft, Apple, etc) but the free software extreme has almost none.
What Stallman does is define the free software extreme which allows the pragmatists to find a much better middle than if he wasn’t pushing that boundary.
I suspect most kernel developers mostly operate with virtual machines in mind, but as a sysadmin I would argue that physical machines also has their use cases. I also suspect that if the linux kernel stop supporting 32-bit machines, others will continue. Are the BSD people thinking about dropping 32-bit support?
"Due to the increased usage of OpenBSD/amd64, as well as the age and practicality of most i386 hardware, only easy and critical security fixes are backported to i386. The project has more important things to focus on." [1]
OpenBSD includes their Xorg port in a base install, and you can run a reasonable graphical desktop on a 32bit machine. Firefox is not packaged for i386 I think. I just use Seamonkey when I test the 32 bit image now and again. There may be other package limitations relevant to server only use. Seems a rational approach to me.
Source: https://liliputing.com/2015/07/install-ubuntu-14-04-lts-on-t...
A friend of mine had one of those laptops and I wanted to install Linux on it. It took me a few weeks for other reasons (those M.2 drives were the worst), but thanks to the work of Ian Morrison, it was possible.
Turns out a bios flash can in some cases enable the 64bit functionality.
Its not a particularly fast system though...
It’s enough for an embedded system, but vastly underpowered for desktop computing or web browsing.
Remove 32-bit x86 support, and it will pin me down to the few 64-bit machines that i have.
Phoronix reported that 5.18 removed the Andes NDS32 CPU arch code. https://www.phoronix.com/scan.php?page=article&item=linux-51...
The reason they do this is because a lot of these systems can break as the internal kAPIs change all the time and the users will not know since the embedded custom distro of software + kernel is not upgraded. Support can be broken for these arches for years until someone gets these systems second hand and posts about them, but may not care to fix it or it is cheaper to replace it.
RMS calls this a major issue, specifically 32-bit support in Ubuntu. I do not agree. My last 32 bit computer died a long time ago after the battery died years ago. It was a Gateway so getting replacement parts would be difficult. I also was never able to get it to run Linux. Aside from my brother-in-law finding his old laptop no one I know uses 32 bit computers (well except for myself with my Pi Zero: ARM v6, which I used once).
It just feels like the wrong battle to pick when Linux's biggest competitor is forcing practically all PCs to use a bootloader signed with their key. Major vendors can choose to refuse to allow users to install their own key. They are already locking CPUs (Ryzens) to their brand of motherboards with a builtin fuse provided by AMD. He of course talks about it-it just feels like the 32 bit part distracts from it.