OSNews

The article questioned the quality of linux drivers, not the quantity, and that's a valid point. I think the author did a decent job of clarifying the point with his examples.

Aside from the author's point about drivers not being able to support the full functionality of hardware, there is also the issue of drivers that become stale in the kernel because they have no maintainers and suffer regressions when other subsystems change in the kernel, in part due to the "we can change the internal API's at will" mentality of the devs. This is a point that Andrew Morton himself brought up, that many of the devs work on newer and sexier projects rather than the grunt work of maintaining the older stuff and the number of bugs and regressions is increasing because of that.

I commend the kernel devs and respect the work they have done with providing drivers, particularly when they require reverse-engineering, and I'm certainly grateful for the ability I have to run linux on most systems I install it on. That level of development support is something that is so far above my own ability to contribute towards that I won't dare judge. I also understand the challenges they have with hardware vendors not releasing specs, or not even having access to the hardware that users are requesting support for. Their challenges are going to become even more apparent as development efforts move towards things like power efficiency, which is difficult to implement without vendor support. So don't take this as condemnation of the effort and work the devs have done to this point, that's not my intent.

But pretending the problem doesn't exist doesn't make it go away. Yes, linux may support more hardware than Windows does, but Windows (and OSX) generally has better support for hardware than linux does. There is a difference. It's not criticism of linux or praise of Microsoft/Apple, just a reality that we need to figure out how to best address.