Personally, I think making it a requirement to tie the hardware and software together would be a mistake. With a little luck, good hardware will get itself attached to a lot of software and vice versa. Having each side maintain a UsedBy list would seem a heavy burden on the original designers. On the other hand, I would hope that in the spirit of Open Whatever, it becomes common practice for designers to maintain Uses lists to address your need.

Chris - AE6VK

These days, hardware is software. It's difficult for me to see that VHDL is different from C in a substantive way.

What may differ is where you can make the license terms "stick". Distribution is sufficient to activate the GPL, although today with application service providers that may no longer be true. This license adds a step, manufacturing the hardware, where the license terms should explicitly activate.

This may be a realm where patent is more important than copyright, perhaps because the creation of schematics (rather than VHDL) are not necessarily fixation of an uncopyrightable idea into a copyrightable form, or perhaps because that copyright is not reliably transmitted to the actual implementation. I'll leave it for John to explain more.

Bruce

• As Bruce has pointed out, HDL didn't seem that far removed from software, and
• Avoiding license proliferation.  There didn't seem to be anything that could be added by another license, so the "GPL or BSD" recommendation was made to avoid license proliferation while giving a choice between copyleft and non-copyleft.

I'm not at all familiar with RTL, so I appreciate you bringing it up.  Could you provide a pointer to more information about it?

The OHL language is not designed to be exclusive to Gerber, PDF, etc. -- they are used mainly as examples of open file formats.  I think that Documentation in RTL format would certainly be covered by the OHL as currently drafted.  But if it makes sense to specifically reference RTL format files, I have no problem with doing that.

Thanks for bringing this up!

John Ackermann

I going to assume by RTL you mean Register Transfer Level. If not, the following will make no sense.

Also, these are my private opinions and I am not a lawyer, so again, they may make no sense.

My admittedly feeble understanding of this is that RTL is a name given to the abstract mental process of designing logic hardware at the level of data transfers among registers. Concrete artifacts of this process step would be things like Verilog or VHDL files. As I suggested in an earlier posting, I think these qualify as software and fall under the OSL of your choice. The OHL is meant to cover artifacts like Gerbers that are sent to  a manufactuing house along with a BoM so they can be made into a piece of physical hardware you can hold in your hand. In the examples I know of and the free tools I am aware of, these latter artifacts are the product of a separate but related mental design process.

My mental model has so far been of a physical board that contains connectors, some specialized chips and some form of FPGA/CPLD to make them all play nicely together. In that model, the Gerbers for the board would fall under the OHL, but the RTL Verilog for the FPGA would fall under an OSL.

I don't happen to know of an example myself, but I suppose I could imagine a case where the Verilog represents full behavior of the board and there is a tool somewhere that compiles the Verilog into a 74xxx realization of the logic, then lays out a PC directly. (Although I don't know how you would specify the connectors to use in Verilog.) In that case, the Gerbers would be a derived work from the Verilog [legal landmine here] and the OSL would actually apply. In that case, I wouldn't think you could, for example, apply the non-commercial version of the OHL to the Gerbers if the Verilog came under GPL.

In my base mental model, neither the software description nor the hardware description is sufficient in itself to produce a functional product. Since paths from the two discriptions to the final product are usually very different and mostly independent, I guess I don't find it surprising that different forms of legal protection should apply.

The OHL is intended to cover the creation and use of hardware; it doesn't address software/firmware at all.  So, to the extent you need software to use the instantiation, no we don't cover that, and my own view is that we shouldn't -- we wanted to keep this license focused on the unique issues of hardware and leave software to the many licenses already out there.

But the question of whether the Documentation has to be complete enough for someone else to actually *make* Products is addressed in Section 4(d) which says that if you make Products, your Documentation package has to include all the files reasonably necessary to allow others to make those same Products. 

I was trying to make a distinction between someone who makes a "theoretical" modification without actually reducing it to practice, who I didn't want to burden with producing production files, and someone who actually makes Products.  In making Products,  you have presumably created those files, and we require that they be included in the package.  It's an attempt to adopt the "preferred form" concept used in the GPL.

John Ackermann

The GPL operates purely as a copyright license.

Pure copyright doesn't work very well for protecting hardware designs because copyright can only protect the expression of an idea, and not the idea itself.  So, the copyright in a schematic diagram, or even a PCB layout, is "thin" in that it is highly functional and there are only a limited number of ways in which the components can interconnect.  In other words, someone can take my schematic and use the idea -- that U1 pin 1 connects to U2 pin 3 -- to quite easily recreate that schematic without infringing my copyright.  And, once I have a schematic it's not that big a deal these days to recreate the PCB and get around any copyright in the original artwork.

The OHL uses the mutual grant of patent immunities as the legal "consideration" necessary to create a binding contract, and it operates on a contract rather than license theory.  While this isn't perfect, we believe that it provides a more enforceable way to control how Documentation is used than a pure copyright model could.

By the way, we use an "immunity from suit" rather than a patent license because the immunity is like a quitclaim deed for property -- it says "to the extent I have any rights, I won't use them against you" without making any claim that I actually have those rights.

John Ackermann

As I understand it, hardware can be covered by copyright or patents.  Trademark law doesn't seem to be applicable to a piece of hardware.  Patents can cover the ideas behind a design.  Copyright can cover things like schematics, circuit board layouts, chip layouts, FPGA bitstreams, HDL source files, ...

I just took a look at the link you included for circuit board layouts.  I was intrigued because I haven't seen anything specifically referring to PCB designs, other than the bare fact that they are copyrightable.  (When I started this project, I did some research of the Westlaw database and found no US cases that talked about, for example, whether Gerber files are a derivative work of the schematic from which they are derived, or whether the PCB itself is a derivative of either the schematic or the Gerbers.  That's one of the reasons we veered away from a pure copyright approach.)

The link you provided actually talks about what we in the US call "mask works" which is a form of protection for integrated circuits; it doesn't relate to PC layouts at all.  About 20 years ago this was a big thing; the idea was that you could protect the "masks" that were used to create each layer of a chip.  It turned out that for various reasons (including the march of technology), this was much less useful than it seemed, and today mask work protection is pretty much a dead letter.

John

This raises a good fundamental question that may be a little different than the ones you asked, and I don't have an immediate answer to it.

I would phrase the question as: "Does the patent immunity granted by a user extend to modifications created at a later date by someone else?"

Clearly, the immunity you grant should cover the Product based on the Documentation in your hands, and the modifications you make, at the time.  But if someone later makes a modification that infringes your patent that wasn't before relevant, should the immunity extend to that as well?

This is a policy, rather than a drafting, question, and I would like to hear other opinions about it. 

Any takers?

John

This is a reply to your more specific questions about assignment to your employer.

This is another interesting point.  Because you are not granting a license, but merely making a promise not to sue, it's not at all clear that the promise would somehow transfer to your employer by the mere fact of your employment.  In other words, your individual commitment not to sue would remain, but by hiring you, your employer would probably not be bound to the same promise.

However, if you sold, assigned, or otherwise transferred your patentable invention, either to your employer or to someone else, we would want the immunity to travel with the transfer.  The language currently doesn't address that, and it probably should.

Thanks for bringing this up!

John