3Dfx Voodoo 1 & 2 Glide Linux Driver Retired, Other X.Org Code Officially Retired

[ClosedSource]

Hey! I wrote that Glide driver! :-)

That was a lot of fun. More of a "can it be done?" type of thing than really useful since it was so slow without any 2D acceleration. But it let me have two screens with the Voodoo driving the second screen which was cool.

Sad to see it go, but it's been irrelevant for forever by now so I'm not surprised.... :-)

-- Henrik

Awesome. Thank you for your contributions to the drivers for this legendary hardware!

[Adarion]

I guess xf86-input-wacom was simply forgotten in the list of maintained ones?

[ClosedSource]

They will stay in the git history forever, but what is the point of keeping them in the main branch if they are unmaintained and not functional for years?

I feel this is the result of the physiological effect that the fragile modular nature of Linux has on people. If you look at Windows, it's just Windows. There are no "packages" inside the core system. While overly bloated so it can cater for every customer running any application written between the 15th century BC and today, no one was ever exposed just how bloated it is. The entire source tree of Windows exceeds a 100 GB in size but it is actually very modular. A lot of it is old C code that is barely maintained to compile with a C++ compiler. You'll see a lot of MACRO_FUNCTIONS_THAT_LOOK_LIKE_THIS which would make today's C++ developers laugh. However, this is being hidden from users who are led to believe that it is actually "one module". They need not worry about bloat. After all, everything is developed under one umbrella and it's the Windows team's responsibility to maintain the entire thing. Ignorance is a bliss sometimes and this certainly fits the Windows user base.

However, the situation is entirely different under Linux. Everything is written by a different team and often by single individuals. Things are scattered all over the globe. The bus factor is scary and doesn't inspire confidence. Someone somewhere is thanklessly volunteering to maintain a library that the entire Linux ecosystem depends on. This is very bad for large vendors such as Canonical, Red Hat, Debian, and Oracle. They simply can't maintain tens of thousands of packages of software developed by random people and loosely tied communities. What happens when your customer complains about a bug and you can't get hold of the upstream developer? So, the paranoia that this sort of system causes started affecting package maintainers. Something hasn't received a commit on GitHub in the last two years? OK, let's ax the package. The package may still work just fine and it would be a wiser choice to just keep it till someone reports a bug.
As the situation progressed, large vendors started contributing to some libraries their audiences rely on, but still not with a sustainable amount of man power. Something doesn't work and isn't absolutely necessary? OK, let's rip it out instead of just disabling it.
This sadly propagated to users. They could see the maintenance burden this fragmentation causes and developed a hidden distrust in their distributions which they hide behind fanboyism. This is human nature. Sometimes we compliment people to avoid facing ourselves and admitting how little we think of them. This leads to an unhealthy obsession with ripping out things even when there is zero benefit from removing. People remove unneeded packages even though there is no "registry" on Linux and everyone has enough disk space. Then they cite nonsensical reasons for doing so. They essentially forgot where their paranoia came from and its detrimental effect on the Linux ecosystem. Some even go beyond that and make self entitled comment on how developers should be allocating their time resources since they are unconsciously scared the project may otherwise fail.
Sadly when someone tries to alleviate the situation by merging things into one umbrella such as when systemd encompassed init + system logging + udevd, they panic even though 1 package is safer and easier to maintain that 3x1 = 3 packages. Nothing ever makes them happy.

Objectively though, no one seems to be striking a balance between risking the occasional false advertising (the Microsoft way) and the overzealous trimming (the Linux way). Neither are doing it right but the latter is simply unhealthy.

[kpedersen]

Objectively though, no one seems to be striking a balance between risking the occasional false advertising (the Microsoft way) and the overzealous trimming (the Linux way). Neither are doing it right but the latter is simply unhealthy.

An interesting reflection. This part in particular. Some random points:

I feel in many ways that Microsoft Windows actually has pretty weak hardware support for certain types of older stuff. I.e the older Intel GMA GPUs are completely unsupported and even the vendor drivers from intel only support Windows 7. Older wifi cards, older usb serial, etc are also no longer available. So the key word is "old" that Microsoft doesn't do great with. It might be a certain element of planned obsolescence (i.e Intel pays them to not support it well)

Linux does tend to do a little better (probably the best all things considered) but it sometimes appears that they are dropping support quickly. In reality some of this niche stuff should not have been there in the first place. But it was just fun and the developer could do it, so why not? I guess.

I feel the BSDs have quite a decent niche. They are sometimes slower at uptake for drivers but hardware tends to be in there for life. FreeBSD has a patchwork quilt of compatibility layers (i.e LinuxKPI). This is quite cool but isn't really leveraged for ancient driver maintenance.

In some ways I wish drivers were more modular. On Linux, it is a modular monolith. A driver can't easily be grabbed and merged with a different era kernel. But this kind of digital preservation is a difficult problem to solve. It sometimes isn't the driver that has rotten but all the infrastructure / boilerplate it requires.

[royce]

i've just seen a movie on youtube with someone that made one of these cards to work on w10

And it worked pretty well, Quake 2 was running smooth asf

[Teggs]

Ah, yes. About the time video cards decided you just weren't cool if you weren't sporting a fan.

[stiiixy]

Ah, yes. About the time video cards decided you just weren't cool if you weren't sporting a fan.

Yeah, I had top replace my really nice Sapphire 9600XT (mid-range) with a passive cooler because I didn't want the fan breaking when I needed it most, which is always how it works. I think my nVidia GeForce 1050 actually had it as well. Pretty hard now I believe to go to passive cooling, at least with higher end? I can imagine the sheer weight of a passive cooler if the active one's plus card are already 3-slots in the high-end.

I've not bought a VGA since Ryzen came out. Just have no need currently.

[Delgarde]

Ah, yes. About the time video cards decided you just weren't cool if you weren't sporting a fan.

Yeah, they were the hot new thing...