Slashdot Mirror
Linux Supports More Devices Than Any Other OS
Linux Blog recommends an interview up on the O'Reilly site with Greg Kroah-Hartman, long-time Linux kernel hacker and the current Linux kernel maintainer for the USB driver core. He updates the free Linux driver program announced almost two years ago, which has really caught traction now with more than 300 developers volunteering. The interviewer begins by asking about Kroah-Hartman's claim that the Linux kernel now supports more devices than any other operating system ever has. "[One factor is] the ease of writing drivers; Linux drivers are at normally one-third smaller than Windows drivers or other operating system drivers. We have all the examples there, so it's trivial to write a new one if you have new hardware, usually because you can copy the code and go. We maintain them... forever, so the old ones don't disappear and we run on every single processor out there. I mean Linux is 80% of the world's top 500 super computers right now and we're also the number one embedded operating system today. We've got both sides of the market because it's — yeah it's pretty amazing. I don't know why, but we're doing something right."
My Swedish vibrator still doesn't have Linux drivers
Note: Human operator not included.
I will bet you just about any amount of money that the standard kernel for Vista doesn't detect that card. Yes, Windows has third-party drivers, but Windows relies on third-party drivers for everything, Linux does not.
Taxation is legalized theft, no more, no less.
Upgraded my Ubuntu server from Feisty (7.04) to Hardy (8.04). The path to Hardy includes Gutsy (7.10). The series of apt-get dist-upgrades went well...then I tried to run apache2. Error:
symbol lookup error: /usr/lib/libxml2.so.2: undefined symbol: gzopen64
I googled...turns out it doesn't remove an old libz file...certain things still refer to it. /usr/local/libz.so.1.2.3.3 is the right one, while the links in /usr/local/lib/ point to /usr/local/lib/libz.so.1.2.3 which is the wrong one. Copy the former into the latter, redo the links, everything's hunky dory.
I think the difference here between Windows and Linux is that I wouldn't have upgraded Windows...I would have reinstalled (going from 2000 to 2003, for example).
I know the data can be very misleading. The average Windows Machine has about 87 devices on it, the majority of which are supported by class drivers (hard drives, chipsets, processors, etc).
In the Windows world, almost all display devices are covered by VGA.sys - so the device has a driver, but is the user experience good?
Also, what is considered a unique device? Most hard drives have a unique identification string, but they are all supported by a null driver. By just supporting a generic hard drive, you have covered close to 40% of the unique devices on the market. If you bias this towards market presence, this gets even more ambiguous. An extremely high percentage of the popular devices on the market are chipset devices - boring things that you just expect to work.
The information about device support metrics only becomes interesting when it applies to less popular devices. Peripherals like printers, scanners, networking, display devices running on full functionality are really the only thing worth measuring - and this is very difficult to do.
I saw this packaging problem completely stop what would have been a large enterprise rollout of Linux desktops. It caused a couple of jobs to be eliminated as well. Say what you want, but if a datacenter manager cannot specify a component by some sort of product ID and acquire a compatible component reliably, it won't happen.
I've reported this story many times, and have had people tell me that desktops "shouldn't" need wireless adapters. But the problem persists: What PCI wireless adapter do you specify, if your job depends on it?
The equipment you're working with probably comes from companies like Barco, Agfa, Siemens, ... am I right ? The ones I saw in that field all ran proprietary software directly on the hardware or on a very thin proprietary OS. Which is why this equipment is so $-intensive (that, and medical research generally pays whatever bill you present them with).
"Violence is the last refuge of the competent, and, generally, the first refuge of the incompetent" - Thing_1
I've seen CE in robotics and lab equipment (oscilloscopes, vector analysers, EMC measurement, ...). I've yet to encounter Linux in this world.
It has always amazed me how much test equipment manufactures have embraced windows. Even HP(Agilent) switched their logic analyzers from HP/UX to windows some time ago.
SONET testers are about the only exceptions that I am aware of.
It's not "linux supports more devices, period", it's "linux supports more devices out of the box".
Of course, if your device doesn't work immediately in linux, you're SOL in most cases.
I ahve yet to install a version of windoes that didn't require immediate driver updates.
The Kruger Dunning explains most post on
But, honestly, I'd rather the Linux "basic" driver to the third-party crapware that you have to install to get some printers working. Things that are so slow to do some things make the device (or Windows) totally unusable because of the slowness. It would be one thing if all the drivers were standardized and worked seamlessly but it seems like every device requires yet another crapware extension to use the software.
Taxation is legalized theft, no more, no less.
I say! Hallo over there.
Could some of you fine upstanding penguins please find it in your pint-size reptilian hearts to migrate over here to Van Daemon's Land this season, and help our poor bewildered little FreeBSD creature rebuild his USB nest?
This is no joke, penguin people. Seriously, I need to keep a Kubuntu machine handy just to read the SD cards from my Canon. That simple task crashes FreeBSD. Regularly, reliably crashes it.
I will probably be hunted down and speared with a tiny fork for this. But I think we need some penguin DNA over here, because no one has been able to properly deal with this for the past six years or more.
There's a recent article at Linux.com about the ancient FreeBSD kernel panic involved in this, that has now even tripped up the PC-BSD project. http://www.linux.com/feature/149224
And now, I must scurry hurry to hide from the fork prongs!
Sincerely - a frightened daemon captive
That's true, but Linux isn't where it could - or should - be. There are many antique 3rd part drivers for Linux for embedded devices and busses (COMEDI doesn't get updated often, DDC's Linux drivers for $1000+ aviation buses haven't been updated in years, VME drivers are equally badly maintained), where comparable drivers for Windows are nice, shiny and up-to-date... even though you know damn well that's not where the market is. It seems to me that some hardware vendors release Windows drivers because they know it'll look good to PHBs, but neglect Linux drivers because they don't give a rat's ass whether the hardware is usable or not after it's out the door.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
Bloggers are just getting in line with professional journalists. When was the last time you saw a reference on a paper news? Or even on a web news?
Rethinking email
You've hit the point where I'm not sure Greg Kroah-Hartman is in touch with reality. Now I very much value his efforts and this doesn't change that, but if you RTFA (I know, I know) he says that every type of device is supported and there are only two classes of devices that are problems. He mentions webcams and wireless. He says webcam suport has recently been greatly improved and about wireless: About a year ago wireless wasn't doing so well. We got a bunch of people working on that now and everything is supported now. The one hold-out is Broadcom but even they have Linux drivers, they're just closed source.
That would be major news to me. Where are these Broadcom drivers? And of course that doesn't fit with the "everything works now" that he is saying. Specifically the BCM4328 isn't usable without ndiswrapper.
So back to the main point, he misses the major class of devices that aren't well supported on Linux: printers. I just spent a lot of time researching and finally settled on an HP Officejet J4580 that I could go buy at any of the major stores and it is perfectly supported in the just released Linux distributions such as Ubuntu Intrepid. And that's all the functions as it is a multifunction. But you certainly cannot just buy any printer out there and have it be supported. I did hours of research to find the ones that are and that met my needs and are available in stores.
So again I greatly appreciate the work he and other driver developers do, but it does no one any good and probably damages Linux to act as if it is something it's not. It's better to be realistic and work from there.