Raul Miller writes: > Another premise which would work better is that firmware is somewhere > between hardware and software and that there are circumstances where it > makes sense to treat firmware as hardware and other circumstances where > it makes sense to treat firmware as software. I feel that this premise > is closer to Brian's point of view than the one you just now expressed.
I think his follow-up post makes it clear that he considers the firmware to be software, with no ifs, ands or buts. Perhaps it makes sense to treat firmware as software and sometimes not; do you have opinions on how we should decide when it is each? My opinion is that if someone wants Debian to distribute the firmware, treat it as software, and apply the DFSG to it; otherwise, treat it as outside the Debian system in the respect that the driver should not be considered to depend on the firmware. I think this is consistent with our practice for other things on the far side of a low-level interface. I do not think applying a very broad idea of dependency is a good idea: it goes beyond what copyright licenses can require, is not likely to lead to more free firmware, and leads to a bigger patchset having to be maintained for the kernel. (As you point out, since firmware can be argued to be software, everyone is likely to think it is on a slippery slope. I would like to avoid arguments based on the idea that a particular conclusion puts us on a slippery slope: we are already on it.) > > Your observation does nothing to convince those who disagree. > > That's always a problem with communications. > > Interested in making communication more possible? Focus on finding > premises you can agree on, and for the words where you can't, focus on > the reasons for picking one definition over an other. I hoped that he would address the differences in our premises once I pointed them out; that was why I mentioned that they differed. Michael Poole