Re: [gNewSense-users] firmware in cx88-blackbird.c

[Sam Geeraerts]

Peter and Jesse wrote:
> ubuntu/media/cx88/cx88-blackbird.c does not seem to have any large hex
> tables, but has functions like blackbird_load_firmware. I identified it
> as non-free, but maybe someone else should double-check.
> By the way, maybe someone who knows more about this could update the
> wiki site. The KFV site still implies that all we are doing is checking
> for the license. What exactly are we looking for when we are looking for
> firmware and blobs? We've covered this somewhat in our lists, but it
> would be helpful if they were all compiled on the wiki.
>  Peter Stevenson

I agree that it would be nice to have some rules of thumb for 
distinguishing non-free firmware/blobs. The problem is that a lot of 
times it's kind of a you-know-it-when-you-see-it thing. You can't really 
put a hard limit on how large a table of hex values should be to mark it 
as non-free.

There's also the problem of 'expert knowledge'. A table of hex values 
could seem unreadable in the eyes of a layperson, but the same table 
might make a lot of sense to an expert. [1] is not the best illustration 
of my point, but you get the idea. There's a huge thread somewhere on 
the fedora-devel mailing list where Alexandre Oliva lobbies for a 
strictly libre Fedora kernel. There, Alan Cox (among others) argues that

a) some hex data is common knowledge the field
b) some of it is the result of trial and error tweaking because that's 
the only way to develop for that kind of chip or hardware.

The trouble with (a) is that we probably don't have any experts working 
on gNewSense, so some data is probably incorrectly marked as non-free.

For (b) we need to decide wether this is an acceptable method to produce 
libre software. If so then we also need a way to recognize it.

More practically, what we could do (and should have done more formally 
during KFV) is make a clear note about why a certain piece of data or 
file is marked (or not marked) as a non-free blob. Those notes can be 
summarized in a table or list in the wiki. Every time a new (suspected) 
blob is found the list should be checked and if necessary extended.

[1] http://lists.gnu.org/archive/html/gnewsense-users/2008-06/msg00020.html