Re: more evil firmwares found
Joey Hess wrote:
> Nathanael Nerode wrote:
>> For the 2.6 kernel (only), here's a simple-minded system.
>>
>> Add to the driver floppies a 'non-free firmware' floppy. The non-free
>> firmware floppy contains files which are copied
>
> So you're going to work on implementing this?
As soon as there are enough drivers in the kernel with loadable firmware
support to make it worthwhile, sure. I was going to do that first, though,
on the grounds that it has immediate benefits beyond the installer. I did
already spend the time to learn how to convert a driver to use userland
firmware loading, even though it's not useful for any hardware I use
personally.
>> (I am assuming that the installer can boot, set up its ramdisk,
>> communicate with the console, and read from a floppy before any of these
>> drivers are
>> necessary. If it can't, you have much bigger problems.)
>
> Many of d-i's target systems do not have floppy drives.
Fine. A non-free CD will do just as well (so then you need to read from CD,
but you already did need to). If you're installing from another OS, a
non-free something-image will do as well. If those can't be managed
either, then they can only be installed with non-free variant installers.
These systems should be quite a small minority, especially given that the
vast majority of Debian users are using i386-architecture machines.
> I don't consider an installer usable unless I can use it on whatever
> hardware comes along.
Interesting definition (certainly not the one I use). In that case, the
installer already has failed, boot-floppies has failed, and you will never
have an installer which will not "fail".
>If I have to reach for Red Hat or Knoppix half
> the time to get a system installed, I will just swap the Debian CD out
> of my working set. Then the Debian installer will have failed.
OK, so this is a slightly weaker definition (still not the one I use); you
want it to work on more than half of some sort of random sample of systems.
>> I actually agree that as long as sarge is releasing "soon" -- and is
>> already going to be full of non-free stuff -- there's no point in dealing
>> with this for sarge, and it should instead be dealt with for sarge+1; the
>> transition
>> pains should be spread out, not stuffed all right before release. (Of
>> course, it's obnoxious that nobody "noticed" these freeness problems
>> until now....)
>
> This is not the first time this issue has been raised on this list, and
I guess nobody bothered to deal with it before. :-P
> it has been raised on linux-kernel before too.
They came to a different conclusion because they were starting from
different premises (no DFSG, for one thing).
> Various people decided to
> make it a big deal *now*, and I find that rather annoying. Especially
> since some of them have made a big deal about our slow release cycles in
> the past.
Well, I wasn't one of those, anyway.
At this point, given the incredible resistance to removing non-free stuff
from Debian, I'm wondering if the Social Contract should be amended to say
"Debian will remain 99% Free Software", just for the sake of honesty. (Or
maybe "50% Free Software", if non-free documentation stays in.) Then 1% of
'main' could perfectly well be non-free, even deliberately. It would no
longer be lying, or a betrayal of Debian's users, to intentionally leave
non-free-software in 'main' -- as far as I'm concerned it is both
currently. :-P
--
Make sure your vote will count.
http://www.verifiedvoting.org/
Reply to: