Binary-only modules not to be banned - this time

Posted Dec 14, 2006 15:42 UTC (Thu) by rvfh (guest, #31018)
Parent article: Binary-only modules not to be banned - this time

Pfeww! For a while I thought my freedom to use WHATEVER was gone. No more 3D on my desktop would really bother me! And no, I haven't got the skills to work on 'nouveau' (but I'll switch with great joy when they get something kinda working).

Binary-only modules not to be banned - this time

Posted Dec 14, 2006 15:48 UTC (Thu) by alexbk (subscriber, #37839) [Link] (9 responses)

Surely you'd be able to revert the change, since all the source is out there, or persuade your vendor to do it, so your freedom was never gone.

Binary-only modules not to be banned - this time

Posted Dec 14, 2006 21:36 UTC (Thu) by huaz (guest, #10168) [Link] (8 responses)

So you are saying only people who can rebuild their own kernels would get the freedom back.

Binary-only modules not to be banned - this time

Posted Dec 14, 2006 22:02 UTC (Thu) by Arker (guest, #14205) [Link] (7 responses)

In a limited sense, that's true, and in that sense, it's true regardless. Freedom of the Press belongs to those that own one.

Of course you don't have to compile it yourself to take advantage of the fact that each distributor compile's it with the patches they believe are most valuable for their customers, just as you don't have to own your own press to benefit from the freedom of those who do to publish without censorship.

Binary-only modules not to be banned - this time

Posted Dec 14, 2006 22:37 UTC (Thu) by drag (guest, #31333) [Link] (6 responses)

I don't remember anybody ever going down with their ship in battle saying:

"Give me convenience or give me death!"

Although I dont' beleive they should use technical measures against binary-only drivers, especially since it's obviously going to cause problems for legal use.

But remember that freedom isn't free, and there is no reason why it has to be convient also.

I wouldn't see anything wrong with having a patch that outputed a explanation in dmesg about how if they got this driver pre-compiled from somebody else that this is indeed a violation of the linux developer's copyright and they would appreaciate it if you tell them to stop doing that.

Binary-only modules not to be banned - this time

Posted Dec 15, 2006 3:14 UTC (Fri) by Arker (guest, #14205) [Link] (5 responses)

I'm not sure how that would cause problems for any legal use. Any user or distributor is free to strip that code out, after all. It would make it harder for people to run tainted kernels without being aware of the problem, however.

Binary-only modules not to be banned - this time

Posted Dec 15, 2006 3:32 UTC (Fri) by drag (guest, #31333) [Link] (4 responses)

Well the awareness is the issue for me also.

A lot of people are going to run drivers that they don't understand that nobody can fix them or troubleshoot them if something goes wrong.

I am thinking of new users and such. Somebody new to Linux and isn't very technical could end up buying a machine or obtaining a peice of hardware that has a propriatory driver and they don't understand the problems you can run into with something like that.

Say somebody buys a Redhat server for home or something like that and something goes wrong as he is getting file system corruption.

He goes to Redhat and it turns out for the SATA RAID he is stuck using a propriatory kernel module.

How does something like that get resolved? Redhat can't fix it for him even if he has a contract with Redhat.

If that person had bought hardware whose maker supported Linux properly then it could get resolved and he wouldn't have to worry about Redhat corrupting his data anymore. Or if Redhat didn't work out he could of gone to somebody else and had it fixed.

Surely if he known ahead of time about it he would of gotten a real sata raid controller or used MD, or gone with a different vendor.

But how can this sort of thing be avoided?

Binary modules for hardware

Posted Dec 15, 2006 7:40 UTC (Fri) by man_ls (guest, #15091) [Link] (3 responses)

Having a piece of hardware which can only be controlled by a proprietary module is not much different than having a piece of hardware which has no Linux support at all. Does that situation have a solution?

In fact it is a very limited subset of supported hardware, and in fact it has been argued that proprietary modules cannot be called "Linux support" properly. Finally, from what I have heard I don't think that Red Hat would be very happy to let you use a proprietary module today.

So you would just say: "Sorry, that piece of hardware is not supported; buy another SATA RAID kit" and be done with it.

Binary modules for hardware

Posted Dec 15, 2006 9:38 UTC (Fri) by Arker (guest, #14205) [Link] (2 responses)

The trouble is that, although it's an obvious lie to anyone that understands the situation, people *do* sell unsupported software as supported. Something like this could help to discourage that. Right now, you can go out and buy a machine using new ATI or Nvidia video cards that cannot be supported under linux, in machines sold specifically for linux use. You and I may know enough to check the specs and notice the lie, but the more new users come into the community the more of us are vulnerable to such tactics. Anything that makes that line between truly supportable under linux and might work with linux if you corrupt your kernel with an opaque blob clearer is a good thing IMOP.

Binary modules for hardware

Posted Dec 15, 2006 11:06 UTC (Fri) by drag (guest, #31333) [Link] (1 responses)

Ya.

Plus the goal is to avoid having people in that situation in the first place.

It's difficult as you will have something like a Adaptic 'fakeraid' controller that actually advertises Linux support and you'll have people in the Windows world that swear up and down how wonderfull it is.

So a carefull person, even if they are relatively knowledgable about most things but a bit naive in Linux will go out and buy it thinking they made a wise choice, which under normal circumstances they did.

Maybe I am being overly dramatic becuase I dislike the idea of having a open source os depend on closed source software for even basic functionality, but it can be a difficult problem it seems to me.

Binary modules for hardware

Posted Dec 15, 2006 12:10 UTC (Fri) by Arker (guest, #14205) [Link]

I've seen this happen way too often, and I've seen too many people come away thinking Linux is crappy because of it. It's much better to just be able to say 'that device isn't supported' than to have a situation where you *might* be able to make it work, but only if you stay within the narrow confines that the manufacturer produces a blob for, and only at the cost of throwing nearly all the natural advantages of using free software down the drain. And it's particularly horrid when you have to explain to a customer that a particular piece of hardware they already paid good money for really isn't supported under linux, despite the manufacturers claims to the contrary.