Re: Bug#239952: kernel-source-2.6.4: qla2xxx contains non-free
J.D. Hood wrote:
>Some people have suggested that this interpretation would have dire
>consequences but I don't see how it would. Most authors who license
>their works under the GPL also provide a form of the work in a high
>level language and thereby implicitly designate that form as the
>preferred form for making modifications.
This interpretation (ie, the author gets to decide what the preferred
form for modification is) /does/ have bad consequences. If I compile a
binary and then statically link in, say, GNU readline, I shouldn't be
able to get away with claiming that I only need to provide the source to
readline since the binary code is my preferred form of modification.
>Basically my argument is that authors should be able to use the GPL
>while remaining free to choose which forms of their program that
>they publish.
In itself, this is arguably not a problem. However, in the case of the
kernel, you're linking your code against code written by a large set of
other copyright holders. If I supplied a binary blob in a C wrapper and
claimed that this was my preferred form for modification, there'd be no
realistic way that the other copyright holders could agree to allow that
into the kernel. Your argument creates a class of GPLed code that can't
be linked against any other GPLed code, and it's hard to see how we
could realistically claim that that would be Free.
--
Matthew Garrett | mjg59-chiark.mail.debian.devel@srcf.ucam.org
Reply to: