OSNews

Well, I would say, how many devices you support, does matter to a good extend but as the article notes, for a large majority, the drivers do work very well.

For gpsca webcam driver, I wrote to Greg KH and Hans from Fedora started working on it as well, so not only is the driver in the latest upstream kernel, in Fedora 10, we also include a libv4l library that supports all the webcams using their own custom formats as well.

http://lwn.net/Articles/287910/
http://fedoraproject.org/wiki/Features/BetterWebcamSupport

Still doesn't take away some of the pain points of proprietary drivers (Nvidia is the last major offender there) and proprietary firmware ( Broadcom is a pain because they are not even redistributable) and many others but overall the situation is getting very much better. It is interesting to see the industry turn around.

There's support and there's support. My laptop for example is technically fully supported, with open source drivers even, for all its features. Basically it is possible to get everything to work, that however doesn't mean it's easy to get working.

To get my fully supported laptop fully working I had to edit several files in /etc, download and compile the latest version of some driver that hadn't made it into my distros repository, cut and paste several scripts from several different web sites, change some of those script to fit my configuration and then edit a couple of more files, based on information from yet another third party site. Once that was done everything worked, more or less.

Compare this to Windows. I install Windows, go to the laptop manufacturers website, download the relevant packages for my laptop, double click those packages, hit Continue a couple of times, reboot and everything works.

Sure, many vendors do exact that. Some vendors like Intel even get their drivers merged before they even release their hardware. What doesn't work is proprietary drivers. Whether the vendors make them or not is mostly irrelevant. Just because people know to make hardware, doesn't make them experts in being driver developers. Often, the opposite.

What do you mean by that? A large portion of the drivers included in Linux is developed by the respective hardware manufacturers, most of them are even maintained directly by them, too.

Perhaps you just lack the knowledge about the Linux driver situation. Why don't you read the Article / listen to the interview then?

The problem with that (I read that as allow separate, binary manufacturer drivers) is that it starts to destroy the integrity of the system, lots of unexplainable problems start to occur, and quite frankly, the vast majority of manufacturer written drivers are crap and there is no guarantee whatsoever how long they will keep those drivers available. Once a driver is in the Linux kernel it stays there until pretty much no one uses it any more.

Edited 2008-11-02 21:42 UTC

Actually, it's about how many currently relevant devices you don't support, don't support well, or make the user jump through hoops to make work. (Inspiring speeches about "Freedom" notwithstanding.)

Ten "attaboys" is worth one "Oh, shit!". People these days expect their devices to just work without a lot of excuses.

Edited 2008-11-02 21:45 UTC

Wow, they sure are going to be disapointed with Windows then

What are you trying to imply by that?

It might be claimed that Windows better supports devices for PC class hardware. But on the other hand, I have seen a report showing a large number of NT kernel crashes attributed to drivers. So without some reasonable evidence, I don't think such a thing can just be asserted as true.

As for how easy to get that support working, do you mean from a user's perspective, or driver / architecture port writer? If it is from a user's perspective, then again, I would like to see some quantifiable evidence. I know that it is easier for me to get Linux drivers working than Windows, because I'm more comfortable in that environment.

From the other perspective.. again, evidence.