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

Re: Non-LLM example where we do not in practice use original training data



Simon Josefsson <simon@josefsson.org> writes:

> Are you suggesting that the DCO is a license text that has to be part of
> the licensing information for a piece of work, and mentioned in
> debian/copyright?

It struck me as a similar sort of thing on first glance, so I used your
comment to make a broader point, but it sounds like this is a more
complicated situation that I don't know anything about. I haven't done any
research here, so I have no opinion about whether the DCO should be
included in Debian packages or not.

My general point is that we will need to have some exceptions for various
reasons, and I'd rather document them explicitly in the DFSG rather than
having a set of well-understood exceptions within Debian that aren't
recorded where people would expect that information to be. I think that's
in general alignment with your point.

I will make the general comment that I think it's reasonable to care more
about code or data that is integral to the functionality of something we
package and less about ancillary files that aren't particularly important
to the normal functioning of the package (such as files that exist only in
source packages and don't contribute to the binary package). Bad licensing
for the latter is still a bug, to be clear, but the centrality of the code
or data to functionality does affect my opinion about the severity of the
bug (unless, of course, we would get into legal trouble for distributing
it at all).

Upstreams put all sorts of weird things in source packages, so there will
be a steady stream of bugs about files with odd licensing. We should
strive to fix them all, but I would prioritize ones that affect the code
that users run or the documentation that they read. Those are more central
to what we're trying to accomplish as a project. That's why I'm
particularly interested in the source for AI models and less worried about
whether we manage to find and excise every poorly-licensed RFC in a Debian
source package, although I agree that the presence of the latter is a bug
and we should fix those bugs (but perhaps less urgently than fixing bugs
where the code itself is not free).

-- 
Russ Allbery (rra@debian.org)              <https://www.eyrie.org/~eagle/>


Reply to: