Debian appears to have dropped 32 bit kernel builds

[seb275]

"Debian will not be building 32 bit x86 kernels in Debian 13 and onward"

As a antiX-23.1_386-base user (on Asus eeePC Atom) I feel sad to hear that :-(

[Adrian]

You can get pick a 64-bit capable PC on eBay for ~$35

I think it's time for people who have devices older than 20 years to seek to update them, if nothing else they will get some great benefits for a very modest investment.

[j2mcgreg]

You can get pick a 64-bit capable PC on eBay for ~$35

I think it's time for people who have devices older than 20 years to seek to update them, if nothing else they will get some great benefits for a very modest investment.

+1

[FullScale4Me]

In my lifetime I've seen many examples of when something was about to go away a group of new volunteers appeared out of nowhere.

Good everyday example - historic structure/artifact preservation. Let us hope those capable see the call for help.

[AVLinux]

In my lifetime I've seen many examples of when something was about to go away a group of new volunteers appeared out of nowhere.

Good everyday example - historic structure/artifact preservation. Let us hope those capable see the call for help.

What such volunteers have expected of them could make a huge difference, if people expect a new 32bit Kernel for every new Kernel release then that is a fools errand, nobody has 32bit hardware that will benefit from that and it seems some people just mindlessly update their kernel for no reason all the time without having any clue what has changed and often lately they break more than they fix... (suspend? bluetooth?)... It's not like my ancient Dell Inspiron or those eeePC's will suddenly stop working without Kernel updates. Now if people would be content with a few kernel updates during the lifetime of a future 32bit Debian stable release and of course due diligence with malware security kernel updates then that may be a tenable solution. Compiling Kernels is not terribly difficult and requires few dependencies. I'm NOT saying that the MX developers need to do this necessarily, not at all. But perhaps if they decide not to continue with 32bit there with be somewhere in the Debisphere will provide 32bit Kernels, in the meantime I would be surprised if antiX doesn't continue with them since one of their mandates is keeping old iron alive..

[FullScale4Me]

Umm, perhaps I should have said Security Updates only for 4.x and 5.x kernels?

[AVLinux]

Umm, perhaps I should have said Security Updates only for 4.x and 5.x kernels?

Sorry, not sure I follow..?

I was talking about moving forward from the point where Debian stopped, so kernel builds newer than 6.11, my suggestion was not necessarily every post 6.11 release though.

[lars_the_bear]

You can get pick a 64-bit capable PC on eBay for ~$35

Not all computers are PCs. There's a heap of embedded and industrial equipment based on 32-bit CPUs. To be fair, these tend to be ARM-based, and I think these architectures remain reasonably well-supported in the Linux kernel -- at least for now. Still, it would bug me if the kernel itself lost 32-bit support, and the dropping of support by distribution maintainers seems to me to make this more likely.

I guess whether this happens depends on whether there's enough interest among people with the skills to keep the older architectures alive.

While I understand the economic and time pressures, it annoys me that "buy new stuff" is such a common response to technical problems.

BR, Lars.

[DukeComposed]

these architectures remain reasonably well-supported in the Linux kernel -- at least for now. Still, it would bug me if the kernel itself lost 32-bit support, and the dropping of support by distribution maintainers seems to me to make this more likely.

While I understand the economic and time pressures, it annoys me that "buy new stuff" is such a common response to technical problems.

It's more like "buy old stuff and insist it's just as good". "Time pressures" is putting it mildly. The i386 architecture dates back to 1985. The 16-bit 8086 processor goes back to 1978.

64-bit compatible replacements for x86 started showing up in 1999[0], so folks have had about 26 years to get ready for sunsetting their 32-bit x86s. For comparison, the 16-bit PDP-11 had a respectable manufacturing run from 1970 until 1997. There's no doubt that the x86 has earned a place in the annals of computing history.

But that is what it is now. It is history. There are still PDP hobbyists and communities that keep the heyday of the PDP-11 alive in their hearts and I admire that. Expecting legacy architectures to stay current and maintained along with an ever-growing number of newer, faster, more feature-rich processors is a little like getting upset that your auto mechanic doesn't keep parts for your Ford Model T in stock anymore.

I don't see 32-bit x86 support leaving the Linux kernel anytime soon, but there are plenty of other architectures that have been removed over the years, including some 32-bit SPARC chips, older x86 processors (including the i386), and the 64-bit Itanium.

You can still run Linux on these machines. You just can't pester the kernel devs anymore when something goes wrong. If there's interest, there will always be a community of 32-bit enthusiasts who I'm sure will maintain forks of the old kernels and continue writing patches for them as needed. That's the great thing about open source software. You're free to keep your thing going if you really have a drive to do so.

[0] While 64-bit processors have been around since at least the DEC Alpha in 1992, these were not x86-compatible.

[lars_the_bear]

Expecting legacy architectures to stay current and maintained along with an ever-growing number of newer, faster, more feature-rich processors is a little like getting upset that your auto mechanic doesn't keep parts for your Ford Model T in stock anymore.

Sure. I have the same kind of problem in my professional life: I'm still maintaining ten-year-old versions of my software; it's a burdensome expense, which inevitably gets passed along to the customer.

And yet...

Who benefits from this relentless pace of technological development? In about 1990 I owned a CX5M computer with (IIRC) 16kB RAM, which I used for typesetting music notation. The application came on a 32kB ROM cartridge. These days I use Sibelius, which is a 40Gb install and struggles with 32Gb RAM. Is it better? Sure. Is it two million times better? I'm not sure about that.

Where is that two-million-fold increase in resources going? Not much of it into better applications, I suspect. I can't help thinking that economic pressures drove us in the wrong direction over the last thirty years or so.

I venture to suggest that nobody who is running Linux on a desktop computer is doing so to be one of the crowd. It isn't necessary for Linux to be just like everybody else, or to respond to economic pressures in the same way as everybody else. We all have our reasons for using Linux, and I suspect there are many different reasons.

I fully understand why Fedora/Ubuntu/SuSE don't want support 32-bit stuff. But distributions that are more built for enthusiasts than business could do things a bit differently. Keeping old hardware alive isn't just a thing to do because it's fun -- it's a way to raise your middle finger at the desperate, needless waste and profligacy that characterizes the modern software industry. A pointless gesture, perhaps, but one that it pleases me to make.

BR, Lars.