Linux Looks Toward Dropping Very Old WiFi Drivers

[caligula]

i wish the distros would take the lead in disabling drivers. The have a much quicker feedback loop with end users, and then they can give feedback to upstream about which drivers people don't care about. i also think there should be permanent documentation about when drivers were removed from the kernel. some tinkerer is going to have to make educated guesses about what is the latest kernel with a particular driver then just looking up the information.

Distros are the worst. They've accumulated all possible drivers. Usually even the staged ones which are horribly unstable. A custom kernel would only need like 10 megabytes of space.

[caligula]

I am a little hesitant on something like this. Linux has long been known for its very good hardware compatibility, especially on older hardware.

I wonder if making these an optional module would be a better option than getting rid of them altogether.

I wonder who would use a 2 Mbps wifi card. Even 1700 Mbps is a lot slower then gigabit Ethernet. So 2 Mbps would have to be slower than 1 Mbps Ethernet.

[avis]

i wish the distros would take the lead in disabling drivers. The have a much quicker feedback loop with end users, and then they can give feedback to upstream about which drivers people don't care about. i also think there should be permanent documentation about when drivers were removed from the kernel. some tinkerer is going to have to make educated guesses about what is the latest kernel with a particular driver then just looking up the information.

Why would distros disable any drivers at all? They are modules and they are not loaded unless they match your PCI/USB/whatever ID.

And then which exact drivers do you think are worth disabling? Are you ready to maintain all of this? Do you really believe distros have the resources to maintain an arbitrary list of HW to support?

People here seem not to understand Open Source at all. Many believe they are privileged to opine about what distros or Open Source developers must do because Linux fans have chosen Linux. Oh, boy. Even Windows users are a ton more modest despite paying for Windows up to $200.

[TeoLinuX]

Oh no, your 500MHz 24 year old laptop with 128MB of RAM won't be able to access the internet if you update it to the latest Kernel! Any laptop that old, with that little ram will barely be able to function with any kind of remotely modern distribution, let alone do anything on the internet which would bring it to its knees. Even if you managed to somehow hack a modern kernel into a much older and simpler distribution, that is not enough memory or processing power to do anything practical on the internet.

You then might argue that you don't use your 24 year old laptop for browsing. It's being used as some esoteric glorified IoT device. Like an internet print server, or you have a big USB harddrive plugged into it so it functions has a poor man's NAS. My counter to that is for $25, you can get a Raspberry Pi that will greatly out perform it in literally every metric while consuming a fraction of the power, and be supported by Linux for many more years to come.

You then might counter that it could be a very specific network attached computer with some proprietary hardware running some big industrial machine. My response to that, is that there is no reason for such a machine to ever have its kernel updated. Mission critical production computers are not connected to the internet.

Amen to that, bro!!!!

[AmericanLocomotive]

I'm more thinking of using an old laptop as a terminal and ssh in and out of it.

A Raspberry Pi needs a monitor, which if bought new is $100.

But hey, most laptops of that area have a parallel port, so PLIP is an option

So yup, "some esoteric glorified IoT device". You already have a newer computer, use that for your SSH needs. Not some crusty, ancient laptop with a bad battery and a terrible screen with horrible viewing angles. If you want a device with a built in screen that can do SSH, pick up a refurbished chromebook for $80.

There is no valid use case for keeping a 24 year old laptop on the internet, with a modern kernel.

[fitzie]

Why would distros disable any drivers at all? They are modules and they are not loaded unless they match your PCI/USB/whatever ID.

And then which exact drivers do you think are worth disabling? Are you ready to maintain all of this? Do you really believe distros have the resources to maintain an arbitrary list of HW to support?

People here seem not to understand Open Source at all. Many believe they are privileged to opine about what distros or Open Source developers must do because Linux fans have chosen Linux. Oh, boy. Even Windows users are a ton more modest despite paying for Windows up to $200.

distros don't enable every driver, but they certainly do a lot. I compile my own kernel using my distros config, and it takes too long. the benefit would be in compute efficiency, reduced compile times, saving energy, and diskspace, and network bandwidth. also as I stated on the post it would do the work of telling kernel devs what drivers aren't used anymore, which would help them remove unmaintained code, which is a risk and a burden.

I understand that unused modules are mostly harmless, but it's still dead code weighing down an increasingly bloated kernel. but in this case I'm not even arguing for my position, I'm arguing for the distros to help the kernel devs who actually want to remove some drivers, the distros can certainly disable them for their non-lts releases and give them some valuable feedback.

[Eirikr1848]

Nononononononono not the PS3 ethernet driiiiver! Anything but that!!!! *cri*

[elbar]

distros don't enable every driver, but they certainly do a lot. I compile my own kernel using my distros config, and it takes too long. the benefit would be in compute efficiency, reduced compile times, saving energy, and diskspace, and network bandwidth.

And as a benefit of such approach is that when switching cflags there is a high chance that something does not compile or other compilation problem or a problem forcing to use some experimental parts of programs which are not yet prepared for such options but it is only for devs hoping that someone step in to help. And problem also is that when something in that programs changed that another compilation or configuration problem occurs which in other distributions like Ubuntu/Debian/Red Hat and so they() solved for me and I could use computers resources for something they were designed for and not for troubles of computer distribution or something which they currently crying for...

[avis]

distros don't enable every driver, but they certainly do a lot. I compile my own kernel using my distros config, and it takes too long. the benefit would be in compute efficiency, reduced compile times, saving energy, and diskspace, and network bandwidth. also as I stated on the post it would do the work of telling kernel devs what drivers aren't used anymore, which would help them remove unmaintained code, which is a risk and a burden.

I understand that unused modules are mostly harmless, but it's still dead code weighing down an increasingly bloated kernel. but in this case I'm not even arguing for my position, I'm arguing for the distros to help the kernel devs who actually want to remove some drivers, the distros can certainly disable them for their non-lts releases and give them some valuable feedback.

Major popular distros normally enable pretty much all the available drivers, period. I'm too lazy to read the rest of your comment given that you started with a blatantly false statement.

If upstream drops them, fine, distros will follow. Until then no one will take your proposal seriously.

[schmidtbag]

Could be useful for an SD reader (or other type of card reader) if you use that a lot for getting data off something like a camera and your laptop doesn't have one already.

I don't know if I recall seeing a laptop without an SD reader. As far as I'm concerned, if you need an SD slot, you're going to shop for a laptop that has one.

Or for a sound card if your computer doesn't have headphone/microphone jacks or doesn't have particularly good ones.

If you're not an audiophile, Bluetooth is plenty good enough if your computer lacks a 3.5mm jack. If you are an audiophile, you ought to have a decent USB DAC.

But maybe those are easier to just include by default. For some uses a capture card could be handy, and having a PCIe based card in a slot may be better than dealing with a USB device. Could be useful for Ethernet, but I guess if guess the dongle doesn't make that much worse...

If you're talking a radio capture card, sure, that's a good example since you don't necessarily need something integrated. But, if what you're trying to connect to your laptop requires a cord tethered to it, I really don't see the problem in using USB/TB.

For some of these sorts of things things, at least the ones needing high speeds, PCIe cards can be cheaper than external options. And just end up being more reliable. And mechanically secure relative to something hanging off a port. So if a modern version of ExpressCard was standard on laptop, I'd probably choose it over USB for certain things.

True but the convenience/portability factor adds to the cost. So, might as well go all-out with TB rather than continue to upgrade ExpressCard.
Bear in mind, I don't necessarily agree with what I say here: I hate how USB/TB spec has transformed into something so particular that it loses all "universality". I'd rather USB be kept consistently slow and at a lower wattage while using something like ExpressCard for higher-bandwidth applications.