The Grumpy Editor's Obviously Incorrect 2006 Predictions [LWN.net]

The Grumpy Editor's Obviously Incorrect 2006 Predictions

Posted Jan 5, 2006 4:36 UTC (Thu) by iabervon (subscriber, #722)
Parent article: The Grumpy Editor's Obviously Incorrect 2006 Predictions

My bet is that proprietary kernel modules will come to a head, and the result will be that they get moved to userspace with clever ways to do a restricted set of kernel-level operations quickly from userspace. Then there will be a lot of contraversy over whether it makes sense for the kernel to have routines specifically to support non-free programs, and whether it makes sense to have a driver in the kernel for the nVidia "does video if you somehow figure out what calls to make" cards, and whether we really just wanted to make sure that the proprietary code wasn't messing up the system. Practical result: high-end video cards work perfectly for Linux, nobody knows what the drivers are doing, but the drivers are sufficiently sandboxed that only RMS actually cares.

Also, people will come up with really odd things you can do with an nVidia card, inspired by stracing the drivers and doing vaguely similar things.

(Log in to post comments)

the non-free kernel module issue will come to a head

Posted Jan 5, 2006 22:08 UTC (Thu) by wilck (subscriber, #29844) [Link]

I wish I could share you optimism. I know this issue is being discussed on LWN over and over, with a strong towards radically anti-prorietary people like Greg KH.

Yet, as I see things currently, "2006 will be the year of the linux desktop" and "the non-free kernel module issue will come to a head" are contradictory predictions to me.

In the worst case, a lot of important hardware will cease to work, or suffer from bad performance. That will scare away a lot of users and leave the open source community with significantly less momentum.

I'd also like to see the day where proprietary modules finally become obsolete, either because the free alternatives catch up or because the hardware vendors understand and release their specs.

I just don't believe that day is near, and I don't share the views of some who think the Linux developer community has the power to impose their will to both a bunch of really heavyweight companies and to the large group of pragmatic users who want their hardware to "just work" with all the features they have paid for.

the non-free kernel module issue will come to a head

Posted Jan 5, 2006 22:12 UTC (Thu) by wilck (subscriber, #29844) [Link]

I know this issue is being discussed on LWN over and over, with a strong towards radically anti-prorietary people like Greg KH.

I meant to say "... with a bias towards ...". Sorry.

Hardware features

Posted Jan 7, 2006 17:00 UTC (Sat) by man_ls (subscriber, #15091) [Link]

I just don't believe that day is near, and I don't share the views of some who think the Linux developer community has the power to impose their will to both a bunch of really heavyweight companies and to the large group of pragmatic users who want their hardware to "just work" with all the features they have paid for.

You are right. One alternative is to pay for less features -- e.g. just use the onboard video cards which come integrated on many motherboards these days. Many people don't really care about bleeding-edge 3d features. The result: less high-end customers for ATI and nVidia.

But if Linux on the desktop has any significance, people will just want their damn 3d things to work right, binary modules or no binary modules. This is true today and hardware vendors know it, or they wouldn't be shipping binary Linux modules to begin with. So they will have no choice but to open their drivers.

Hardware features

Posted Jan 12, 2006 21:54 UTC (Thu) by wilck (subscriber, #29844) [Link]

So they will have no choice but to open their drivers.

That's the part I don't believe. The closed source drivers cost the vendors no more than a few developers. I have no idea if it pays off for them sales-wise, but it's a relatively cheap thing to do. Opening up the source is an entirely different issue. They may have licenced other companies' patented code, and they fear the loss of competitive advantage like hell. Most probably the non-free driver is some sort of cross-platform code, and they are afraid of opening up portions being also part of the Windows driver.

There are other devices out there, like cheap software RAID controllers, which need a proprietary driver top operate their disks in RAID mode. In this case, open-source RAID solutions being readily available, it makes no sense whatsoever to protect the code. Releasing binary drivers and keeping up to date with kernel development and different distributions, OTOH, causes their manufacturers big headaches. Yet they don't even think about opening up the source. And people buy their products (even OEMs who claim to support Linux) because they're a few cents cheaper than others. That's how it goes, I fear.

Still being pessimistic.