Re: hello

[graham@belegost.mit.edu]

> 
> > I liked the background to the fcpu license you mailed to
> > the f-cpu group a while back. Could I pin it up on the
> > hardlicense web page as a starting point for discussion?
> get comfy and do whatever can advance the debate.
> criticize and do whatever you want, but remember the willingly
> utopic idea that it holds.
> 
> of course i don't think that everybody will conform to the
> rules, because they are guidelines in the spirit. the goal is to
> prevent and avoid the problems that may arise in an industry
> that is different from the software industry. in the last mail to
> Rares Marian, i explained the latest ideas.
> 
> > And if so, can I remove the references to 'intellectual
> > property' ;-) ? [see http://www.gnu.org/philosophy/words-to-avoid.html
> > for why]
> 
> i'll have to have a look. what do you want to replace it with ?
> if you modify the text, just include your own comments so we know
> who does/thinks what.
I don't know what to replace it with. 'IP' is one of those nasty terms
that is convenient because it avoids having to think about the quite
different things it covers - but it smuggles in a belief that ideas
can be 'property' - that once you've had an idea no-one else is
allowed to have it, which I think is quite contradictory with your
generally utopian approach (and the FSF's approach too). In any
case, you cannot claim ownership of 'IP' in the sense it is used in EDA
using only copyright law - if there is an original IDEA in your
design, you could only protect that idea using patent law. The
gpl and any other license you create are based on copyright, which
only covers form.

<snip>
> 
> For the F-CPU licence, i think that my goal is to have a new concept.
> it's not a freeware, not public domain, it's more "hardcore" than the GPL.
> it's a stronger licence than the GPL because of some of the additions
> i want to include (including the obligation to divulgue the source at
> no cost and no delay through the Web, anonymously for the retriever).
IMO this is going to cause problems. With the gpl you only have to
divulge the source when you distribute the software, not when you
are still working on developing it. If you're anything like me, while
you're working on any kind of design its a total mess of bits and pieces
till near the end. So you're half way through developing the
'improved FCPU' and someone says to you 'I demand by my rights under
the FCPU license that you stop what you are doing, tidy up all your
files, tarball them for me, and please document what changes you have
made'. Will you do it? No, you'd say 'go away till I have my design
ready for distribution, when I will package it up for you'. Which
is what the gpl does anyway.

> But, more fundamentally, with the GPL the copyright belongs to the
> author and (iirc) to the people who added/modified it. for the F-CPU,
> it's a community, so the copyright belongs to this community. 
How is that different from the gpl? When a big piece of software
has been developed by many people, the fact that many people 'own'
the copyright effectively means that no-one owns it - it is not
possible for anyone to get all the people who have worked on Linux
together to agree to change the license. The gpl is a practical
way of achieving what you want to achieve in this case.

>The problem
> is that it has no legal existence : it is ideally every engineer that the
> Earth holds. i think that i have reached a point of no-return : the f-cpu
> belongs to everybody and nobody. the point i want to make is : because it
> belongs to everybody too, you have no right to hide it, to make your own
> version without releasing it freely at no cost (the same way you obtained it).

> Of course, the "parallel market" of the money makers with CD distros is
> valid, as long as this free alternative is provided. otherwise it wouldn't
> be fair. As a parent, even if you're "only" a father (because the mother
> has had all the pain of giving birth) you have one half of your genes in
> your child's genes. When a nurse comes to care for them, it's a natural
> right to check if your babies are ok. i did the same thing with the f-cpu's
> "software" : the goal is to prevent anybody/companies to pollute the "freedom"
> spirit. I have seen GPL infringements through "reinterpretation" of the GPL.
Any reference to this? 

> with the disclosure paragraph, we can keep companies from hiding "features",
> or including features that break the community's efforts. Intel's and M$'s
> efforts in that domain (obscure data formats and hidden "features") are a threat
> not only to the consumer's rights but also to our freedom and our society's
> integrity (ever heard about the commercial use of the Neurotic Spook Agency ?)
No, what do you mean?
> 
> this engineer dream has taken me far from my initial, simple goal...
> 
> Nicolas Matringe asked on the f-cpu ml :
> > I don't understand why source code (VHDL or Verilog) can't be protected
> > by existing public licenses (GPL or whatever). It's only source code, like C.
> > HDL is only a *functionnal* description. Different synthesis tools won't
> > produce exactly the same hardware from the same source code (that's why
> > I am currently running the same design through 2 synthesizers to see which is the best).
> > I don't think EDIF netlists (synthesis output format) can be protected
> > by existing licenses but HDL source code should.
> i'm not a specialist, anyway. he made a point but i don't know which.
>
HDL source code, whether behavioural or structural, can be protected
by the gpl. The European Space Agency, who have good lawyers, do this;
Richard Stallman and the FSF accept that this is fine. The Free Hardware
Foundry do the same with non-synthesizable HDL; again, the FSF backs
them.

The problem IMO comes from the fact that where software comes in
only two forms - source and executable, where the executable can
be created AUTOMATICALLY from the source, hardware designs have multiple
'layers' which may need human intervention at each stage. If Nicholas
just takes his netlist and uses it as input to another stage, then
no-one needs to have the netlist to recreate the hardware - only
the source. But what if he needs to manually tweak the netlist?
Then his netlist also needs to be covered by the license, because without
his tweaks, the design cannot be implemented. The same goes with
placement and routing files.
A second difference from the gpl comes with the method of distribution
of the final product. OK, Yann's approach avoids this problem  (but IMO
creates worse ones). The problem is this: for software, if I distribute
the executable, then under the gpl I must distribute the source, too.
The obvious parallel for hardware would be: if you distribute the
gerber files/FPGA bitstream/GDS files etc then you must distribute the
HDL code/schematics too. If that is all you want then the gpl is
pretty close too it. But a second case for hardware is when a
manufacturer takes your design and uses it to make a chip, without
ever distributing the design in any form. In this case I think
the license should also say 'do this, and you must make the design
available'. But its something that isn't in the gpl, and also has legal
problems (remembering that manufacturers can fight with patents but
we can  only use copyright law).

> 
> last question : can i get a tarball (gziped please) of the whole mailing list archive ?
>
Done that. It's in http://opencollector.org/Backend/hardlicense-discuss.tar.gz

Graham (who is now going to bed :-) 
> 
> WHYGEE
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> the F-CPU: http://www.mime.univ-paris8.fr/~whygee/f-cpu.html
> SHARCPAGE: http://www.mime.univ-paris8.fr/~whygee/sharcpage.html
>