kernel-libre (hopefully 100% Free) for Fedora 8 and rawhide

[Anders Karlsson]

* Alexandre Oliva <aoliva at redhat.com> [20080325 07:07]:
> On Mar 23, 2008, Anders Karlsson <anders at trudheim.co.uk> wrote:
> 
> > * Alexandre Oliva <aoliva at redhat.com> [20080323 20:52]:
> >> On Mar 23, 2008, Chris Snook <csnook at redhat.com> wrote:
> >> > Pardon my ignorance, but I honestly don't see a risk in shipping
> >> > *sources* which contain hex-coded firmware blobs that have been
> >> > licensed for distribution,
> >> 
> >> This is not about the risk.  This is about not distributing non-Free
> >> software.  For me, it's not a matter of licensing, not a matter of
> >> license compatibility.  It's a matter of not supporting the
> >> distribution of non-Free Software, no matter how hidden it is, or how
> >> important it is for some.
> 
> > Right, so what is stopping you from maintaining your own branch of
> > Fedora, own repo's and own infrastructure for the work that you are
> > doing?
> 
> Nothing AFAIK.  I guess I could add these bits as a separate package
> to CVS (which is precisely what I had in mind), given approval for
> that.

I think others have already said that separate package is not the way
to go. Modification to existing kernel spec and then passing the right
switch when building to generate your kernel is probably the better
way.

> > I just do not understand the need to cripple Fedora
> 
> I think you misunderstand.  Nobody's talking about crippling Fedora.
> Nobody's talking about crippling Fedora's kernel.  I'm talking about
> adding an *alternate* kernel package set, such that people can create
> Fedora spins that are 100% Free Software and still call it Fedora, and
> such that people who would rather run a 100% Free kernel have an
> option to do it.

Then that is fair. No further objections your honour. :)

> > If this really was so important, should this work not be done at the
> > upstream level, with kernel.org, rather than on distribution level.
> 
> Considering how much effort gNewSense, BLAG, Dynebolic et al put into
> it, I don't get the impression that upstream is interested in that.

With persuasive arguments enough, they'd listen. Problem is, only
presenting an argument against, for example, non-free firmware,
without presenting (multiple) alternative ways to do it - that also
can be put to the vendors creating the hardware that use the firmware
- will most likely be written off as a non-starter.

Linux (kernel and distributions alike) are still in a phase where they
are getting big, but are no way near big enough to start dictating
terms to vendors such as what they should and should not do with
firmware as an example.

Picking the timing to do these things are crucial, and doing to early
is likely to backfire. I appreciate your goal, but I think you may be
premature in going for it.

> > I am really glad to hear that. I still think that this effort you
> > describe, if this is that important, should be done upstream or by
> > cloning the distribution.
> 
> Wouldn't you consider it silly to clone an entire distribution,
> replace all of its Fedora naming and themes (because the Fedora
> trademark guidelines wouldn't allow such a modified distro to be
> called Fedora any more), just because of a wish for an additional
> kernel that's right in line with Fedora's stated goals?

That's because I had the impression you intended to have the kernel
replaced with your one. As you have cleared that up, this point is now
moot.

> > Utopia sounds good, but the path there is not short, nor a straight
> > line. While I agree with you that 100% Free (as in speech) should be
> > the goal, having a gun-foot moment now is counter-productive.
> 
> There's a big misunderstanding here.  There's no gun-foot moment, I'm
> talking just about an additional package, that I'm willing to maintain
> myself.  We add new packages every day.  Sure, not so big ones, not so
> active ones, but still, that's mostly my problem, no?  If you don't
> want to use it, carry out your life as usual.  If you care about
> Fedora's stated mission of building an OS entirely out of Free
> Software, this helps achieve this goal, and considering upstream's
> stance, that's the best we can aim at.

I agree with you. :)

Shake?

> > Wireless firmware is closed source at the moment for very good
> > reasons.
> 
> s/good reasons/lame excuses/

Well, different points of view and all that. As a choice between
having a native driver at all or having to run something like
ndiswrapper with -ick- Windows driver code - I'd pick the native
driver.

> > The trade-off then was open driver - closed firmware (so that power to
> > the transmitter was controlled in firmware). Looking at this thread,
> > that seems to still be the case.
> 
> That firmware is not part of the kernel at all.  It's completely
> unaffected.

True. Point conceeded.

> > If you were to take this 'fight' seriously, then you should convince
> > the FCC (and similar organisations worldwide) that the control on RF
> > transmitters and airwaves should be relaxed to the point that
> > companies can release fully open firmware (as no-one will care if a
> > tweaked firmware is used to boost signal strength at the likely cost
> > of interference with other peoples equipment around them).
> 
> There are people working on that.  Meanwhile, I'm working on something
> that I *can* fix right away.  Why wait?
> 
> > As you see, "choice" have repercussions. The question is, how far do
> > you want to take *your* "choice" at the expense of others?
> 
> There's a misunderstanding here.  How could my choice of maintaining
> an additional package (that per chance happens to be a 100% Free
> kernel) harm anyone whatsoever?

Yes, and as you've clarified this and the misunderstanding is no more,
this became moot.

Thank you for the response, it has been enlightening. :)

/Anders