Re: Knuth statement on renaming cm files and Licence violation.
Thomas Bushnell, BSG <tb@becket.net> writes:
> Russ Allbery <rra@stanford.edu> writes:
>> Even if Debian is not violating the intended license directly, to base
>> a stance on the viewpoint that the license is legally uninforcable and
>> therefore irrelevant seems rather disconcerting.
> Are you worried that we would be violating it indirectly?
I'm worried about the basis for decision-making that would be used, in a
broader sense, and what potential slippery slopes it may lead into. I
understand that legal exactness has to be used as a basis for legal
decision-making. It just seems to me like when it comes to a question of
policy, using exact legal readings as a way of working around a
problematic licensing clause whose intent doesn't coincide with the exact
legal reading is a bad approach.
That's most of what I was trying to say. The questions of whether the
license are acceptable are a separate issue to me; the main point that I
feel strongly about is to evaluate the license on its own terms, rather
than on the basis of whether it's worded so as to give people a legal out.
> Note that DFSG-4 does not allow requiring he changing of *file names*,
> but rather only the name of the *work*, which is a different matter--and
> essentially so--since the name of the work is not a functional element.
Yes. This is the fundamental question, namely whether the DFSG only
allows requirements of changing names as a way to handle what are in
essence marketing issues (making sure the user knows they have a modified
package), or whether it's permissable to protect APIs this way.
(In essence, that's what we're talking about with TeX fonts and with LaTeX
packages. The name of the package and the name of the font are an element
in the TeX API, and if you change them, you have to change the document to
reflect that change, exactly as if you changed an API in a library.)
Knuth clearly is attempting to protect an API. The whole reason why he
feels so strongly about this is because he wants to be able to take a TeX
document and process it with a TeX system 50 years from now and get the
exact same result down to the word wrapping, and in order to do that, he
has to be sure that when one uses the standard API for the TeX system, the
results will be identical as when one used that standard API twenty years
earlier on a completely different system.
My personal belief, for whatever it's worth, is that this is so borderline
as to almost require the nature of the language be taken into account; if
changing the calls to fit a new API requires a lot of work, that's
different than changing a line in the prelude of a document. I can both
understand Knuth's concern and understand the concern from the direction
of supporting modification and redistribution, and it's hard to find the
right balance, particularly when we're talking about reproducibility of
written work rather than software. Written works don't change in the same
ways and for the same reasons as software packages.
But my basic point here is not to so much get into all of that as it is to
say that I think that fundamental question needs to be addressed according
to the intentions of the original author, regardless of the precise
wording of the license.
> You seem to be suddenly entering this as if you were totally unaware of
> the preceding controversy. I would suggest you go to the mailing list
> archives and review the discussion.
I'm aware of the basic issues and have followed this exact argument a few
times through from different angles; what you're probably picking up on is
that I didn't follow the immediately previous Debian discussion before
the recent threads.
--
Russ Allbery (rra@stanford.edu) <http://www.eyrie.org/~eagle/>
Reply to: