[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: How long is it acceptable to leave *undistributable* files in the kernel package?



On Fri, 18 Jun 2004 14:26:05 -0500 Joe Wreschnig wrote:

[...]
> I agree with Michael Poole insofar as this message.

I agree too.

> Here's an attempt
> at an unbiased summary:
> 
> There are four classes of firmware:
> 
> 1. Firmware which no one has any permission to distribute. These have
> to go right away, or be relicensed. Thankfully, there are few of
> these, and the kernel team seems to be willing to help pursue the
> relicensing.

Good to know...

> 
> 2. Firmware which is released under GPL-compatible terms, with source.
> Everyone loves these, they can stay (e.g. the Adaptec drivers), and we
> encourage other manufacturers to do the same.

Yes and I would say that a sort of list of "manufacturers that did the
Right(TM) thing" could potentially encourage other ones to do the same,
perhaps.
Maybe some credit to those manufacturers (and to other hardware
manufacturers that behave in a free-software-friendly way, not
necessarily in a Linux-kernel-specific manner) could be put in a
dedicated section of the Debian web site.
What do you think?

> 
> 3. Firmware which is released under GPL-compatible terms but with no
> source available.
> 
> 4. Firmware which is released under GPL-incompatible terms with no
> source available.
> 
> (There is potentially a fifth class, firmware released under
> GPL-incompatible terms with source available. I don't believe anyone
> has found such a beast yet.)
> 
> The debate is over 3) and 4). Specifically,
> 
> 3a) What is firmware "source"?

The "preferred form for modification", as defined in the GPL.
And I think that *very* few programmers would prefer dealing with those
string encodings rather than some assembly language form (or whatever is
more human-readable)...

> 3b) If the firmware has source, does the DFSG apply to it (i.e. do we
> need it)? Unequivocally yes with the new DFSG, but some of the GRs
> (not voted on yet) may revert sarge to the "old" DFSG, and the RM's
> interpretation of that lets us release such firmware.

Why? 'Cause firmware is not software?

> 3c) If the firmware has source but the source is not available to us,
> is it a GPL violation for us to distribute it?

I think it is.
A violation of the GPL license *applied* to the firmware itself.

> 
> 4a) All the problems above, and,
> 4b) Is there some relationship between firmware and the rest of the
> kernel that makes this amount to a GPL violation? This is related to,
> but not the same as, the answer to 3c).

I think that a kernel that incorporates a firmware is a derivative work
of the firmware. So, yes, I think that distributing a Linux kernel that
embeds a GPL-incompatible firmware is a violation of the GPL license
applied to other parts of the kernel itself.



-- 
             |  GnuPG Key ID = DD6DFCF4 | You're compiling a program
  Francesco  |        Key fingerprint = | and, all of a sudden, boom!
     Poli    | C979 F34B 27CE 5CD8 DC12 |         -- from APT HOWTO,
             | 31B5 78F4 279B DD6D FCF4 |             version 1.8.0

Attachment: pgpRcR6VePu64.pgp
Description: PGP signature


Reply to: