Having heard about this issue from the usual places, and having read
Branden's rather frightening plunge into the rat's nest of licensing
cruft (BR: you're a braver man than I), I set about trying to figure out
exactly what the problem is. So, I headed over to xfree86.org and
tracked down the new license:
http://www.xfree86.org/legal/licenses.html
Then I popped up the venerable Debian Social Contract,
http://www.debian.org/social_contract.html
and set about trying to find the problem for myself.
Now, there may be deep and subtle issues here that I don't grasp, but
this license doesn't appear to clash with the Debian Free Software
Guidelines. I've read the nine points of the guidelines, and the four
points of the new Xfree86 license, and I simply don't see it.
It's clear enough that the BSD community and the FSF don't see
eye-to-eye on what, exactly, Free means. I think it's fair to say that
because both licenses stipulate that the author retains ownership, it
boils down to a matter taste (this isn't to say that there arn't such
things as "good taste" and "bad taste"). That aside, the BSD license and
the GPL both meet the requirements of the DFSG.
Now, looking at the sample BSD style license cited on the DFSG page,
http://www.debian.org/misc/bsd.license
and looking at the XFree86 1.1 license, I don't see a material
difference. The license clearly isn't GPLish, which no doubt annoys the
FSF. But it _is_ BSDish, which, as I pointed out, is perfectly
acceptable as Free Software, according to the DFSG.
It appears that they want the copyright notice in the documentation of
binary redistributions. Now, I have the latest X packages installed,
which I understand to be "binary distributions." I note that there is a
file /usr/share/doc/xfree86-common/copyright that, insofar as I
understand the situation, addresses exactly this issue.
So, I'm stumped. We wouldn't be having this discussion if there weren't
a problem, but I'll be damned if I can see what it is. As Branden has
shown, the XFree86 Project is very inconsistent in it's handling of the
license situation, and their state of their codebase speaks to that.
It might be illustrative to relate a short anecdote. My friend Ian
Langworth is the author of a program called cadubi. Ian wrote this
little utility when he was very young, and as he puts it, "didn't know
anything." He assembled a license for it by wandering around BBSs and
web sites and plopping together bits of boilerplate that "sounded good."
Legally speaking, it was gibberish.
Years later, someone decided to upload it to Debian, and there was an
argument about the license. From what he tells me, it ended up being
placed in non-free, and everyone was very sour about it. Oddly enough,
no one thought to email him, that author. Years after that, he became a
Debian user. On a lark, he looked for cadubi in the archives, and there
it was, in non-free. So, he ripped out his gibberish license and put it
under the Artistic license, and all was well.
The reason I'm bringing this up is that it is very easy as maintainers
and packagers to forget that the authors are regular folks too. If there
is a problem with an ostensibly Free Software license, it is probably
not intentional. In Ian's case, the whole issue could have been avoided
if someone had simply fired off an email to him saying, "Hey! We want to
upload cadubi, but your license doesn't make any sense. Could you please
pick one of these licenses that we _do_ understand?"
Clearly, XFree86 is a bit more complicated. Nevertheless, shouldn't we
be talking about how to work with the XFree86 Project to resolve the
issues (whatever they are), instead of talking about forking the whole
project? Or is forking/re-implementing/replacing XFree86 the hot new
thing?
Russell
