[cc-community] Open Hardware Licence
Greg London
email at greglondon.com
Sun May 10 19:55:44 EDT 2009
>> GNU-GPL won't provide copyleft protection
>> if the source code ends up in an ASIC.
>
> You really don't seem keen to acknowledge
> the fact that chip layouts are copyrightable works.
So, I realize that if anything disagrees with what you say,
(like if I quote FSF's stand on not using their license for hardware),
you'll say "But you should talk to a lawyer", so, I realize
on some level that explaining this to you is pointless,
so, really, I'm only sayign this to the folks who happen to
be reading along: you are wrong.
http://en.wikipedia.org/wiki/Semiconductor_Chip_Protection_Act_of_1984#A_sui_generis_law
::Although the U.S. SCPA is codified in title 17
::(copyrights), the SCPA is not a copyright or
::patent law. Rather, it is a sui generis law resembling
::a utility model law or Gebrauchsmuster. It has
::some aspects of copyright law, some aspects of
::patent law, and in some ways it is completely different
::from either.
I await your reply where you dismiss being shown wrong
and instead cop back to the standard "there is no case law,
you need to talk to a lawyer" nonsense.
For the rest of the folks reading along,
GNU-GPL triggers its copyleft aspect only
if you create a derivative and only if you
distribute that derivative, and the copyleft
aspect only applies to that distributed
derivative. This is because FSF have an
particular political motive that private
mods must never be forced to be copyleft.
So, it only triggers on a derivative that
is distributed.
If an ASIC vendor creates a derivative,
AND KEEPS IT PRIVATE, and only sells a
physical manifestation in the form of
a silcon ASIC, then GNU-GPL will NOT trigger
it's copyleft aspects to the silcon.
Only a fricken moron would suggest that
a silicon ASIC is a copyright derivative
that has anything whatsoever to do with
copyright law.
The MASK-related laws has NOTHING to do with
derivatives. It ONLY is used to protect the
physical layout of a chip so that other vendors
don't simply copy the physical layout and duplicate
an exact replica of a chip. Because it's a physical
and purely functional layout, normally copying a
physical object in that manner wouldn't be
susceptible to copyright law. The mask law stuff
says you can't simply cut and paste the physical
layout of an ASIC and make cheap knock offs.
You have to create your own functional source code,
translate it to gates, and run it through your own
place and route system.
The short of it, Alan, is you are once again wrong.
A physical ASIC is not a copyright derivative
of source code. Selling a physical ASIC is
not distributing a derivative of the source code.
Therefore GNU-GPL will NOT protect the source code.
The MASK is registered with the copyright office.
But it is NOT considered a "derivative" of the
source code, and selling a physical ASIC is not
considered "distributing" that "derivative"
So GNU-GPL will not trigger its copyleft protection.
>> That's the simple fact. It isn't faith.
>
> If there is no caselaw it is not a fact.
Argument from ignorance. Stop blathering.
There is nothing that would indicate GNU-GPL
would protect hardware source code that got
put into an ASIC. Nothing.
Advising someone to adopt GNU-GPL to protect
their open hardware project simply because
no one has come out and explained it to you
in terms you can understand is IRRESPONSIBLE.
>> And it was confirmed to me by FSF via email
>> a year or two ago when I was looking into
>> an open hardware license.
>
> I'm not convinced the FSF is right.
God forbid you admit you've made a mistake.
> It would take a lawyer to find out
GNU-GPL won't protect hardware.
You are wrong to say that it's up in the air,
or that we don't know, or that there is not
case law so it could change.
Wrong.
And finding someone who wants some form of
strong copyleft protection on their hardware
and telling them to use GNU-GPL based on an
argument from ignorance is setting them up
for failure.
> however, and a lawyer is where people should
> go for advice not this list.
If you think the only point of this list is to
tell anyone who asks any question to go ask a
lawyer, then you should get off this list.
> Having talked to several it appears a
> lot more murky than you make out.
Just because you're confused doesn't mean
everyone else is.
> The other alternative for copyleft
> type behaviour on novel hardware is
> patents - because they can be made
> to work like the GPL (in fact they
> reach further than copyright licences
> can and into areas that the FSF can
> only wish they could do in software copyright).
This is just a useless waste of space.
This doesn't help anyone. Of all the
stuff in an ASIC, there might not be
anything that is patentable, or there
might be a few things that are patentable,
the rest can be reversed engineered by
anyone. Patents cost money, copyright is cheap.
No open hardware group can possibly sustain
using patent law for protection.
>> Gee, Wally, I've only been dealing with the issues around
>
>and now we get to insults.
Wow, you think we're just getting to the insults JUST NOW???
You've been insulting, condescending, and dismissive
from my first post.
> Then your legal advice differs from the advice
> I've been given - but that would have been
> mostly about EU law.
Wow. That's sounded a lot like a politician saying
something like "mistakes were made" rather than
"I was wrong".
> I've spent a lot of time banging my head against
> this *exact* brick wall because I have the same
> problem with etched brass works. The artwork is
> copyrighted. If that artwork then produces a set
> of pieces for a functional object (eg a frame or
> even something complex like compensated trucks)
> those pieces so produced are not copyrighted and
Same with open hardware
> even an APSL type licence won't stop someone
> taking the design if provided under a copyleft
> style licence, modifying it and selling parts
> based upon the modified design providing they
> don't publish the modified design - because
> they don't "Externally Deploy [their] Modifications".
With open chip design, it's pretty difficult to modify
something at the functional level. You'd have to do
an ECO or something, and that ain't easy, and it ain't
something that most engineers would want to do, and
it ain't something that you can do really complex
modifications with.
With a chip design, just by virtue of the complexity
of the design, most modifications will have to start
with the source code.
So if someone is doing an open hardware project
that might end up in an ASIC, something like the
APSL would be sufficient for most uses.
> They apparently externally deploy a work created
> with the modified work - which is different.
I told the hardware group that was asking for help
that they would probably have a weak defense trying
to protect something in their design that someone
else modified at a functional boundary. It could be
possible to build the Open Core as something called
a Hard Macro, where it is laid out as a 2-D chunk of
the ASIC, and then if the vendor connects anything
to that hard macro, it may qualify as a functional
connection rather than as a copyright derivative.
But, if a group designs an open processor core,
and they release it under something like APSL
which triggers on externally deploying the mods,
then the core is going to be so complex that
most mods will be made to the source code, rather
than at a functional/gate level, so they should
get all the mods that anyone makes. If a vendor
treats teh processor as a hard macro, then they
might be able to plug that processor into other
logic in the chip, and that other logic will probably
not be protected by any license. But at least
they got better protection than GNU-GPL, which
will do squat in an ASIC.
If your brass works are modified in the source code
then they'll have to make their mods public.
If they create a new piece that is separate,
and start from their own source code, and add
that piece in with the others at a functional
level, then they won't have to release their
source code. But, you can always reverse engineer
their new functionality. And if your license has
a patent-protection clause, then they can't use
your stuff, add an extra piece, and patent it to
prevent you from reverse engineering an open version
of their part.
Which is why open hardware needs a copyleft license
that doesn't require distribution of a derivative
to trigger, but rather some sort of external deployment,
and needs some kind of strong patent protection to
make sure that anything the vendor does can be
reverse engineered by the open group. The stuff
you can patent in an ASIC is crazy, so open hardware
really needs this kind of protection.
More information about the cc-community
mailing list