Numerous Linux/X11 Display Drivers Can No Longer Even Properly Build

[cb88]

There needs to be far more aggression in taking out old code. What is the use case for a 20-30yo machine running a modern stack?

No binary Linux distro (as far as I’m aware) supports any of these cards anymore, so you’re now apparently supporting people who either use a gentoo or slack derivative (or even LFS) and at the same time need a totally modern stack.

To what end... the code is unmaintained, that means it isn't a burden... often these types of problem are also good for new developers to poke at to get experience. Chill.

In any case other Unix distrobutions have no issue supporting such old drivers... and even do so on a vast variety of computer architectures. The only reason this code stopped working is code breaking changes nobody needed... aka churn which is the real problem on Linux these days.

What you call modern I call bloated... even Linus himself admits that there is several percent slowdown in the kernel per year that he has noticed.

[Raka555]

The "I have it, it turns on, and it must work!" crowd, which seems to be largely European, wants it working. Many of them have hardware from the late 90's/early 2000's that has no legit value outside of retro computing, specifically gaming. They won't upgrade, citing e-waste, poverty, marxist bullshit, or any number of other reasons. I myself won't run Linux on anything older than hardware from 2006 and even then, it has to be a multi-core, large amount of RAM system.

Pitty I can't give you a thumbs down ...

[Raka555]

To what end... the code is unmaintained, that means it isn't a burden... often these types of problem are also good for new developers to poke at to get experience. Chill.

In any case other Unix distrobutions have no issue supporting such old drivers... and even do so on a vast variety of computer architectures. The only reason this code stopped working is code breaking changes nobody needed... aka churn which is the real problem on Linux these days.

What you call modern I call bloated... even Linus himself admits that there is several percent slowdown in the kernel per year that he has noticed.

Standing ovation for you !

[Raka555]

A friend of mine lost his job about a year ago due to covid. (But he made enough to survive for a long time)
He plays games day in and day out now.
He never owned any thing better than an 9800GT. (The exact model in the picture)
500.jpg
He plays on linux and is having a great time.
He only has to play StarCraft II on Windows as we are unable to get it working under wine.

[dpeterc]

People should be able to run modern Linux on 30 years old computers if they choose. Just be responsible in your choices and pick more mainstream hardware where there are still users around, probably still developers around who know the code base, and the initial code was better developed longterm/in the outset. It also requires the user to be responsible to run a light weight GUI or else expect to use command line only.

I maintain a lot of code I started writing 30 years ago, and it requires constant attention, to keep it in shape.
What was good practice 30 years ago is no longer valid nowadays. Some functions get deprecated. Everything is getting more restrictive.
So it is not developer's fault from 30 years ago if they could not predict the future. High quality portable software is a moving target.

Look at the updates being made to the code for floppy drives. If it's mainstream enough and there already was a great foundation to stand on then age should not be a limiting factor.

Situation with floppies is different, since you can still buy NEW external USB floppies and they are used in industrial setups. So developers can and should test in on a modern computer.

We can't expect developers to keep an old i386 with TsengLabs ET4000 in order to try if there aren't any regressions.

[sl1pkn07]

A friend of mine lost his job about a year ago due to covid. (But he made enough to survive for a long time)
He plays games day in and day out now.
He never owned any thing better than an 9800GT. (The exact model in the picture)
500.jpg
He plays on linux and is having a great time.
He only has to play StarCraft II on Windows as we are unable to get it working under wine.

i have exact the same card. good choice in earl days when can mix it with AMD cards with Nvidia for use in games with physx support (the famous Hybrid PhysX Mod).
and some late adquire a second unit for get SLI xd


good times

in linux is impossible do that

greetings

[jabl]

Feels like a wonderful opportunity for some spring cleaning! If it's been broken since 2012 (or the likes of it), archive and let somebody un-archive if they want.

If it's not built by default and build failures aren't considered blockers for improving common code (as can be seen by several of these drivers failing to build due to using some common X code that has long since been removed), then evidently few of the developers will care about removing them.

It's a bit different in the Linux kernel where improvements to common code like API changes is expected to fixup all the code that uses it. In such a case removing some old and crufty code that nobody uses brings a clear maintainability benefit.

I thought issues like these were found daily by building the kernel with all configuration flags enabled?

Well, this is X.org and not the kernel. And since these seem to be old school user mode setting drivers without a corresponding kernel driver, kernel build testing is unlikely to catch any of these issues.

[Azpegath]

Well, this is X.org and not the kernel. And since these seem to be old school user mode setting drivers without a corresponding kernel driver, kernel build testing is unlikely to catch any of these issues.

Wow, my bad, for some reason I had the idea the article was referring to kernel drivers. Thanks for the correction.

[torsionbar28]

they can no longer even build with with modern toolchains / software components. Given the lack of bug reports around such issues, there are very likely few users trying some of these vintage hardware combinations.

I don't think this is a valid conclusion. End users typically install pre-compiled binaries as distributed by their favorite Linux distro. They are not compiling from source, especially not on ancient hardware where a single kernel build could take all day. I would imagine that distro maintainers are the ones who would encounter the fails-to-build problem, not the end users.

That said, folks running vintage hardware, whether for some business reason or just vintage enthusiasts at home, are typically doing so specifically to run vintage software. The number of users who want to run the latest code on 20+ year old hardware has to be extremely small.

I've got some vintage machines in the basement, 486 era, Pentium, AMD K6-II and K6-III, etc. All of them are used with vintage software. I did try to find a modern distro that would run on a K6-III+ with 256 MB of RAM and found it nearly impossible. The kernel has been too large to fit on a floppy disc for years now, so you at least need something new enough to boot from CD, which means very late 90's and newer. Plus, even the "lightweight" distros designed for older hardware mostly require 512 MB or 1 GB of memory as a minimum, which wasn't really a thing until the Athlon/P3 era circa ~2000.

tldr; running a modern distro on 20 year old hardware is not really a viable option, and there's no good reason to attempt it.

[sinepgib]

I don't think this is a valid conclusion. End users typically install pre-compiled binaries as distributed by their favorite Linux distro. They are not compiling from source, especially not on ancient hardware where a single kernel build could take all day. I would imagine that distro maintainers are the ones who would encounter the fails-to-build problem, not the end users.

And it follows such distro maintainers would report broken builds, no?