Alternative to GPL license

[ijor]

I am ready to release the sources of my 68000 cycle accurate core. I just need to decide which open source license to use. I admit I am not precisely a GPL fan. And I would like to hear comments about other possible alternatives.

My main concern, or at least my main doubts, are about allow commercial usage. It is perfectly ok for me if individuals or a small company sells a product that include my core, even at a reasonable profit. But I'm not so sure about the hypothetical case that a corporation would use it as an essential part of a commercial product.

I'm not a lawyer, but I guess there is no legal way to distinguish between a similar usage by the "good guys" and by the "bad guys". So probably this is not much more than thinking in loud voice.

[Sorgelig]

In your license you can write that commercial usage is allowed only with your written agreement. You cannot distinguish in license if it is small commercial by hobbyist or large corporation as there is no strong legal division between them.

[troed]

I think the closest you can come is to use Creative Commons NC (non-commercial) SA (share-alike). After that you indicate on your website that hobbyists can contact you for a commercial license (you write to them that it's ok in an email).

[alexh]

I thought that LGPL was the licence to use if you were considering future commercial licensing?

I thought the problem with GPL was that it requires that applications which use the open source have to release some / all of the source to the overlying program? As a result companies go out of their way not to use it, or use it as an addition so they don't have to show the rest of their work.

If you're interested we have a program here at work which converts Verilog/VHDL into non-human readable code which remains formally identical to the original and synthesises to the same code. We use it from time to time when giving examples to EDA companies.

[Sorgelig]

If you're interested we have a program here at work which converts Verilog/VHDL into non-human readable code which remains formally identical to the original and synthesises to the same code. We use it from time to time when giving examples to EDA companies.

how it's related to topic?
There was no question "how to obfuscate the code".
Otherwise there are many way to do it. As far as i know, it's possible to release netlist version which also will be identical and synthesize-able but hardly understandable by human.

In general, i don't see much commercial value for big companies from retro cores. Big companies can easily buy many different IPs including MC68000 which is available to purchase quite some time. It's not big money for big company. So all these cores for MiST, MiSTer, etc are not worth for big companies.
Even those companies who release their minis don't look at FPGA or ASIC anymore. ARM SoC can do this on much cheaper level.

So, just release the source without worry. It's really interesting only for hobbyists, those freaks who still trying to recreate 20-30 years old machines (mostly interesting for 40+ years old people) to preserve them, nothing more.

[BlankVector]

I am ready to release the sources of my 68000 cycle accurate core.

Very good

But I'm not so sure about the hypothetical case that a corporation would use it as an essential part of a commercial product.

Regarding to GPL: indeed, there is no way to differentiate "good guys" and by the "bad guys". Rules are the same for all users, either individuals or big companies. They all will be allowed to redistribute your work, modified or unmodified, provided that they also provide sources if someone asks. There is no question about money in GPL. People may sell your work with big benefits (without asking you), but other people may also give your work for free. Usually, it is considered that people will not buy GPL software as it can be obtained for free elsewhere.

Note that you are the copyright holder of your core. So you define your own rules. If you release your core under GPL, people will have to respect your rules : they may redistribute your work, but if they make modifications, they must also release their modifications under GPL. Then you can get them back.
But as copyright holder, you can also provide the very same core to other people under a restrictive, commercial licence (dual licensing). Typically, you ask them money to give them the right to distribute your binaries with a restrictive, non-GPL licence.

I understand that your concern is to avoid the situation where a big company steals your work and make big money without giving cash back to you. If you release your core as GPL, you will not be able to prevent that situation (GPL doesn't care about money). But your core, including eventual modifications, will stay GPL for anyone except you. While you can still do what you want (restrictive licences, for special customers) with your own code.

I can't tell much about restrictive licences, which seems to be what you are looking for. But keep dual-licensing in mind.

[ijor]

I think the closest you can come is to use Creative Commons NC (non-commercial) SA (share-alike). After that you indicate on your website that hobbyists can contact you for a commercial license (you write to them that it's ok in an email).

Yes, I know I can use this strategy. But this is not really what I wanted.

Big companies can easily buy many different IPs including MC68000 which is available to purchase quite some time. It's not big money for big company.

But is it cycle accurate?

Even those companies who release their minis don't look at FPGA or ASIC anymore. ARM SoC can do this on much cheaper level.

That is very true. Btw, the new Sony PS Classic is based on a public open source software emulator!

Usually, it is considered that people will not buy GPL software as it can be obtained for free elsewhere.

I think you are thinking in terms of software. This is not software. People would probably not buy GPL software, that's very true. But people might certainly buy hardware based on GPL code.

I thought that LGPL was the licence to use if you were considering future commercial licensing?

I think you are right. And I am coming back and considering just using GPL. If you look at many of the free FPGA cores available, you will see some mix of GPL and non-GPL code. This is not allowed by the strict letter of the license. But nobody care. Except that a corporation might certainly care. So using GPL might actually stop the "bad guys", or at least make things more difficult to them.

Btw, it seems that HDL code is not as protected by copyright as software is. Say, if some company wants to put my core on an ASIC and start selling 68K derivatives, I have no way to prevent it. That is not governed by copyright.

[Sorgelig]

Btw, it seems that HDL code is not as protected by copyright as software is. Say, if some company wants to put my core on an ASIC and start selling 68K derivatives, I have no way to prevent it. That is not governed by copyright.

It's not true. As an author of code you have rights for this. If you can prove someone is using your work commercially without your permission then you may sue them. But it's hrd to sue because it will be almost impossible to prove they've used your code.
And where you've got that HDL is not protected as software? Strictly speaking HDL is the same source code compiled to binary the same way as traditional software for barebone MCU. Will this binary used as microcode for CPU or connect the logic in configurable array - it's already doesn't matter.
Opencores website has a lot of cores licensed as GPL and other software licenses. So, it's protected the same way as any source code.

[jotego]

Many interesting comments in this thread. Just some notes...

I do not agree to that distinction of good and bad guys. In general, I do not agree to the world view of large companies being evil and small ones being angels. If I agreed to it, I wouldn't be working on implementations of chips from large companies such as Yamaha. And, I wouldn't be working for a large company either.

I do not think either that there is much commercial value to our work. A microcontroller can do the job, and companies do alter the original source code of their games when they re-sale it for modern platforms. I remember one compilation from Konami where they introduced a wider play field to Nemesis so it fits modern screens without altering the aspect ratio. They also added options to the Nemesis game menu in order to go back to the emulator menu. What I mean with this, is that they have resources to alter the original game in order to fix things so it works better inside the emulator and with modern hardware. So for them, fixing a weird dependency on clock accuracy is acceptable; thus, cycle clock accuracy is not a requirement.

As for me, I think the purpose of an open source license about retro stuff should be to promote more work done in the scene. So a license that requires publishing the core source is a must.

[ijor]

And where you've got that HDL is not protected as software? Strictly speaking HDL is the same source code compiled to binary the same way as traditional software for barebone MCU. Will this binary used as microcode for CPU or connect the logic in configurable array - it's already doesn't matter. Opencores website has a lot of cores licensed as GPL and other software licenses. So, it's protected the same way as any source code.

I am not a lawyer and I can't argue about this too much. But do some googling and you will find that most law experts agree with what I said. This is an extract from a page linked at opencores:

Software licenses (free or otherwise) are based almost entirely on copyright law. Almost any conceivable use of an electronic file is considered copying, at least under U.S. law, meaning that a would-be user needs to (a) have a license and (b) comply with its terms in order to be legal. While this is still true for the HDL files describing a design, the acts of manufacturing a device based on the design, or using such a device, are not subject to copyright.

[Sorgelig]

While this is still true for the HDL files describing a design, the acts of manufacturing a device based on the design, or using such a device, are not subject to copyright.

So i don't see much contradict here. It's like endless patents war between Apple and Samsung. They don't have rights to devices of each other while they still sue each other for infringements of specific rights.
You don't have right for produced device based on your HDL, but still (in theory) can sue the company if it is used your HDL.

[ijor]

So i don't see much contradict here. It's like endless patents war between Apple and Samsung. They don't have rights to devices of each other while they still sue each other for infringements of specific rights. You don't have right for produced device based on your HDL, but still (in theory) can sue the company if it is used your HDL.

No. It is perfectly legal for a company to use your HDL code for manufacturing an ASIC device without your permission. You can't sue them. You can't because there is no copyright infringement whatsoever. Copyright laws simply don't apply in that case. That's why most hardware is protected by patents. Patents do cover hardware, copyright not that much. But patents are not granted automatically as copyright. Your must pay for being granted a patent.

As you are saying, the Apple vs Samsung, as most other similar cases, it is mostly about patents, not about copyright.

It is similar to copyrighted schematics. The drawing is protected, the actual circuit is not. That means that somebody can't publish your schematics without your permission, but he can legally make a circuit based on your schematic, and even sell it without your permission. Again, unless protected by a patent.

[Sorgelig]

Analogy with schematic doesn't apply for HDL. For HDL you still need to use the source code which is part of software, or copyrighted text if you like this term more.

[jotego]

While this is still true for the HDL files describing a design, the acts of manufacturing a device based on the design, or using such a device, are not subject to copyright.

So i don't see much contradict here. It's like endless patents war between Apple and Samsung. They don't have rights to devices of each other while they still sue each other for infringements of specific rights.
You don't have right for produced device based on your HDL, but still (in theory) can sue the company if it is used your HDL.

Add a back door. Implement a forbidden instruction to do something you can verify. That is a way to prove that they have got your code. I know that it is open source and they could remove it, but they'll probably not check for it.

[Sorgelig]

Anyway, for me it's just for fun. And like i've said all this retro has no commercial value besides just some very low quantities and mostly as a side projects.

[ijor]

As for me, I think the purpose of an open source license about retro stuff should be to promote more work done in the scene. So a license that requires publishing the core source is a must.

I agree of course.

To be honest, I'm afraid that the license has a limited protection in cases like this. The value on my core is on the research and reverse engineering work. The actual volume of the code, the total number of code lines, is relatively small. Once you have a source to study and compare, it is not too difficult to perform a "legal" re implementation if you want. Re implementing from source, even using clean room, requires a small fraction of the work that involves reverse engineering from pure hardware.

There are quite some examples of this in the open source emulation world.

[ijor]

I finally released the source under the GPL license: viewtopic.php?f=28&t=34730

I decided it wasn't worth the complication to write a special license or use a more restricted one.

[Locutus73]

Just a curiosity of mine: isn’t it legal to open source a project through a strict copy-left license such as GPL3 and then relicense the same project to a company with a specific payed license for a commercial product? It’s my understanding that, if you’re the original author of a whole project, I mean not using other works with a viral license such as GPL3 which binds you, then you are free to release the same project many times with different licenses.

Many thanks in advance.
Best regards.

Locutus73

[AnthonyJ]

It’s my understanding that, if you’re the original author of a whole project, I mean not using other works with a viral license such as GPL3 which binds you, then you are free to release the same project many times with different licenses.

Yes, that is correct, authors can release works under multiple licences as they wish (so long as they have the rights to everything...).

However the GPL3 licence (by design) doesn't prevent a company for example building a replica ST, putting ijor's core in it (for which the source would be available), and then selling thousands of them without giving ijor any money or even acknowledgement that it is fundamental to their product (other than having to distribute the copyright statements in the legal text that nobody reads). With building it into an ASIC rather than using an FPGA, it's even worse - as they're simply using it as instructions of how how to build a chip (which they're entitled to do). The thing they're distributing (the ASIC) is no longer ijor's copyrighted work so they can distribute (sell) it without having to pay attention to GPL at all. That theoretical scenario is what ijor said he was concerned about.

[ijor]

However the GPL3 licence (by design) doesn't prevent a company for example building a replica ST, putting ijor's core in it (for which the source would be available), and then selling thousands of them without giving ijor any money or even acknowledgement that it is fundamental to their product (other than having to distribute the copyright statements in the legal text that nobody reads).

Exactly. I understand something like this happens with the Sony PS classic. Except that in that case is a software emulator, not an FPGA core. The matter seems to be a bit gray regarding if the GPL license can be applied to an FPGA core, at least on a commercial product. So chances that a company won't take the risk.

With building it into an ASIC rather than using an FPGA, it's even worse - as they're simply using it as instructions of how how to build a chip (which they're entitled to do)...

Right, a circuit can't be copyrighted at all. But even for a FPGA core. There are legal workarounds. The real value of my work is not on the code. It is on the research and the reverse engineering. The research took years, writing the code took a few months. This is not something like, say, Linux kernel with millions lines of source code. If somebody wants, he can use the result of the research and produce his own implementation. The research is now public and free, I can't claim any copyright (and that's one of the reasons I decided it was not worth the complications of a more restricted license).

Furthermore, it is possible to use my core to debug a separate implementation. You can run both "your" core and mine and record any discrepancies. This is a fabulous, invaluable, debug technique. This is actually what Mike from Fpgaarcade is doing. But Mike is honest, is not hiding it, and his core would be open source as well. So I don't mind at all on his case, quite the contrary.

But not everybody is honest as Mike. I wouldn't be surprised if some cycle accurate closed source implementation would appear in the near future though.