[Gnewsense-dev] Re: Libre wifi for Ben Nanonote

[Daniel Clark]

Have people thought about creating a daughtercard / extended enclosure
that would allow ExpressCard or MiniPCI Express cards to be used in
the Ben Nanonote?

Or that would implement a fully functional USB port, perhaps with an
extended enclosure that would fit a few dongles on a hub (like the
always innovating touch book)?

I'm sticking future research on this up on libreplanet:
http://groups.fsf.org/wiki/Hardware/Freest/Nanonote

I'll move over to the Qi Hardware wiki once anything concrete is
achieved, and perhaps before; I just noticed that the new account
creation was on the bottom right-hand side of some page, instead of
the usual mediawiki top right-hand side.

(For those interested, my answer to the political debate are below;
please note that even when I was employed by the FSF, I did not speak
for the FSF; and specifically the below are my personal thoughts.)

On Mon, Jan 18, 2010 at 8:48 PM, Wolfgang Spraul <address@hidden> wrote:
> Daniel,
>
>> The http://www.cardaccess-inc.com/products/index.php?a=wlan_burt is an
>> SD sized Atheros AR6001GZ chipset device (exactly the same chipset as
>> the openmoko freerunner as far as I can tell), so perhaps that + the
>> microSD to SD adapter mentioned earlier is an option for people who
>> refuse to use any hardware that requires binary blobs.
>
> He he. Yes, for people who (please don't take this personal) refuse to use
> non-free software unless it is hidden, the AR6001 might be an option.
> Let's just be very clear. The AR6001 has the same binary blob as the 6002,
> only that it is stored in flash. Flash is more expensive and needs more
> power than RAM, so they switched to RAM.
>
> But please - for you or any other free software supporter out there, don't
> follow this wrong AR6001 path. First of all the binary blob also exists, just
> hidden in flash. The flash update utility is very guarded by Atheros.
> But even worse - the firmware in the chip has several versions, during my
> time at Openmoko we encountered 1.1, 1.3 and 2.0. The GPL driver and binary
> firmware are sending binary data structures back and forth. At Atheros, the
> sources for _BOTH_ the firmware and GPL driver are kept together in one
> revision control system. The build process produces both the binary firmware
> for the chip, as well as the corresponding GPL driver. There is no continuity
> between the GPL driver sources produced to match the different firmware
> versions, 1.1, 1.3 and 2.0, or any future ones they may have done since.
> Actually, releases are made for each customer! The trunk in the revision
> control system is just the 'unstable' branch basically, and once they have
> a customer, they allocate a certain amount of testing/bugfixing/polishing
> resources to that customer, then they branch in the revision control system,
> and work towards a release for that customer. Of course in this process the
> binary firmware blob and GPL driver are always seen as a pair. There is no
> independent development of a GPL driver. So actually you have as many GPL
> drivers as you have Linux customers.
>
> My take on this: Let's stop thinking that it's a good idea to hide non-free
> software and call it hardware. We must not be afraid to realize that there
> is a lot of non-free software running on these chips. We need to understand
> this software better, and write free replacements. Then make our own
> copyleft chips.
>
> Once we have a totally free chip (of course including the firmware and driver
> under GPL), I know you will support it :-)

I think we are mostly in agreement (c.f. [1]), and are working towards
the same end goals.

I respect your more functional / hardware-side views, but from a
software point of view it is very clear that including a binary blob
as part of linux, or any other software that purports to be licensed
under the GPL, is in violation of the GPL, since it is not the
"preferred form of the work for making modifications to it [the
software]".

The people who legally control linux choose to include non-free
software in code they overall choose to call GPLed, and of course they
do not sue themselves. People who truly care about the GPL and
software freedom should not accept this; they use the linux-libre [2]
fork of linux, where due diligence is done to remove GPL incompatible
code, and code that is pointed out to be in violation of the GPL is
actually removed quickly.

If you use linux-libre with the wifi hardware we have been discussing
other than the AR6001, you will of course find that it does not work
at all; so from the point of view of someone who is unwilling to
compromise on this issue, it is actually worse than nothing existing,
as people's time and effort are being squandered on something that
will never be usable by them / GPLed software, instead of working on
trying to get free hardware (whatever your personal definition of that
may be, as long as it is usable with GPLed software) to exist / work
well.

I think the difference here is that when you install software that is
licensed under the GPL, you expect that software to actually abide by
the GPL; that is, you expect you will be able to modify it using the
same tools and methods that the author did, and you expect those tools
to be free as in freedom and free as in beer; eg for there to be no
large monetary or political cost to you doing so, the only barriers to
changing or improving the software are your personal time, knowledge,
or ability to get other people to apply their time and knowledge.

When you buy hardware, you have no such implied (or legally spelled
out by the vendor's choice of a licence such as the GPL) expectation,
and most likely even if all of the software used to create the
hardware is available, you do not have the often very expensive tools
and machines needed to actually create the piece of hardware from
scratch.

I hope that we eventually get to a world where people will come to
expect that at least some hardware in every category is available with
the same freedoms that come with free software, and I really respect
qi's work towards getting the world to that point; I am just trying to
make the current version of that work as useful as possible for people
who are unwilling to compromise on software freedom while on that
path.

[1] [Openec] Nonfree EC issues pop up again - Lemote YeeLoong netbook
http://lists.laptop.org/pipermail/openec/2009-December/000131.html
"This situation, and the solution of just never modifying the EC code *on
a specific machine* once the user has it, is obviously suboptimal."
Note: We did find someone willing to work on this; drop by
#gnewsense-dev on freenode IRC if you are interested; I should send a
follow-up to this mailing list :-)

[2] Linux-libre project
http://linux-libre.fsfla.org/
"Linux, the kernel developed and distributed by Linus Torvalds et al,
contains non-Free Software, i.e., software that does not respect your
essential freedoms, and it induces you to install additional non-Free
Software that it doesn't contain."

-- 
Daniel JB Clark | http://pobox.com/~dclark