Re: [gNewSense-users] KFV: Detecting sourceless firmware

[Sam Geeraerts]

Brian Brazil wrote:
> On Tue, Apr 22, 2008 at 11:08 AM, Markus Laire <address@hidden> wrote:
>
> > Brian Brazil wrote:
> >
> > > On Tue, Apr 22, 2008 at 10:33 AM, Markus Laire <address@hidden<mailto:
> > > address@hidden>> wrote:
> > > >   Also, do you mean by "you should be looking at the entire file in
> > > >   any case" only checking the whole file for sourceless blobs, or also
> > > >   something else (e.g. whether there are additional license details
> > > >   somewhere else than in first comment)?
> > >
> > > Yes, also checking for other license details. I'd be very suprised if
> > > there's something non-free (by Linux's definition) in the license header,
> > > but caveats in the actual code might have snuck in.
> > That is nice ideal, but I doubt whether this project will (for the
> > forseeable future) have enough manpower and will to actually read through
> > everything contained in Linux Kernel (>260MB at the moment).
>
> That doesn't mean we shouldn't do it though. Heurstics can only get you so
> far, reading through everything is the only way to be sure.
>
> Brian

If heuristics and helper scripts can get us 80% (or more) of the way 
pretty fast then it's probably better to go by that way first (as we're 
doing now). But we should strive to achieve the goal that has been set, 
even if it's ambitious.

What we then need to consider is: do we hold off the release until we 
have a 100% for sure completely libre distro or do we just finish our 
current work ("glancing over" the licenses and blobs of our current 
kernel and then doing the same for the diff with Ubuntu's release 
kernel) and do an early release with updates afterwards.

The problem with the former is that it's going to take a long time until 
the release. That means users and (potential) contributors may loose 
interest eventually which will slow the work down even more. On the 
other hand, it could also be an incentive to start helping out, although 
I think that's less likely.

Doing an early release does not deliver on our promise and I think that 
should be made very clear then. But the release notes should also 
include that a lot of work has already been done and that the team is 
strongly committed to fulfill that promise of complete and guaranteed 
freedom in the short term. I don't think an early release should be done 
for reasons of convenience (having newer software, postponing the 
thorough check indefinitely), but in the interest of gaining more 
momentum. If we release soon after Ubuntu, we can kind of ride the wave 
of it's media attention and possibly attract more contributors. For 
example, if we can release before (or on) the 3rd of May I intend to 
pimp gNewSense (but not as a speaker) at the Belgian release party 
(http://hardy.releaseparty.be/).

On a related note, I've been following the "Rethinking Gobuntu" thread 
on the Gobuntu mailing list. It sadly seems to have died, but I think 
Mark Shuttleworth does have a point. We should try to think of ways to 
cooperate closer with Ubuntu, not only to make our work easier but also 
to attract more attention from members of the Ubuntu community who would 
like to contribute to gNewSense. But that discussion is probably best 
left for a later (quieter) time.