"Open Source" and "Free Software" are almost legally identical - the only difference being that the OSI's definition of "open" was based on the Debian Free Software Guidelines which are very slightly less restrictive[0] than the Free Software Definition the FSF uses. If a license passes FSD it will pass DFSG/OSI; and almost always a license that passes DFSG/OSI will pass FSD.
The primary difference between "Open Source" and "Free Software" is political. The FSF was explicitly a political organization to promote Stallman's idea of software freedom[1] while the OSI was created specifically to make an "apolitical" business case for it. This means a lot for advocacy and ideology but not so much for the actual rules of what you can copy and when.
Also, Apache license is both OSI and FSF approved. The main problem is that you cannot combine GPLv2 and Apache software. This is only a problem if you don't want to take GPLv3's TiVo clause and/or blocked v3 upgrades.
[0] If I remember correctly, there's one license that requires modifications be distributed in patch form, which satisfies DFSG/OSI but not FSD.
[1] To the point of writing a documentation license, GFDL, that prohibits you from removing Stallman's rants on free software, which ironically makes it non-Free.
I thought it was the Artistic License which required modifications to be distributed in patch form? The FSF does seem to have considered the original Artistic License to be non-free.
I think there was some disagreement about some version of the Apache License too, but I think it was about whether it was GPL-compatible, not about whether it was free.
This is going to be super important, as soon as we have automated digital fabricators that can fabricate silicon chips competitive with TSMC's and Samsung's. Until then, you're kind of stuck with what TSMC and Samsung and Intel decided to fab for you, AMD PSP and all; you can't just recompile your CPU to remove the backdoor or add half-precision matrix multiplies, or recompile your DRAM with a large CAM to see if it helps with database queries. You can do a lot of experimentation with FPGAs but you'll get 10x worse performance and 100x worse price than an ASIC.
It's 02027 and you don't like your CPU reporting you to the US Trade Representative for copyright infringement because you have a file consisting of "1" (https://news.ycombinator.com/item?id=30060405), or to the Chinese Communist Party because you have a copy of the Quran? Not sure that Rage Against The Machine album always had lyrics urging you to get out there and buy Sony products? Suck it up, peon, that's just the way things are.
I agree that manufacturing is a real problem. FOSS processors can get adoption but until we open up the manufacturing pipeline they won't provide end users with the benefits we get from FOSS software.
12 comments
[ 3.8 ms ] story [ 37.5 ms ] threadThere is still no commercially usable free-software-like PDK
"Apache License, Version 2.0 (#apache2) This is a free software license, compatible with version 3 of the GNU GPL."
The primary difference between "Open Source" and "Free Software" is political. The FSF was explicitly a political organization to promote Stallman's idea of software freedom[1] while the OSI was created specifically to make an "apolitical" business case for it. This means a lot for advocacy and ideology but not so much for the actual rules of what you can copy and when.
Also, Apache license is both OSI and FSF approved. The main problem is that you cannot combine GPLv2 and Apache software. This is only a problem if you don't want to take GPLv3's TiVo clause and/or blocked v3 upgrades.
[0] If I remember correctly, there's one license that requires modifications be distributed in patch form, which satisfies DFSG/OSI but not FSD.
[1] To the point of writing a documentation license, GFDL, that prohibits you from removing Stallman's rants on free software, which ironically makes it non-Free.
I think there was some disagreement about some version of the Apache License too, but I think it was about whether it was GPL-compatible, not about whether it was free.
It's 02027 and you don't like your CPU reporting you to the US Trade Representative for copyright infringement because you have a file consisting of "1" (https://news.ycombinator.com/item?id=30060405), or to the Chinese Communist Party because you have a copy of the Quran? Not sure that Rage Against The Machine album always had lyrics urging you to get out there and buy Sony products? Suck it up, peon, that's just the way things are.
Software is one thing. Anyone can sit down and write software.
Chip production though? Unless you look at the most simple of designs, you need tool chains worth millions and/or billions of dollars.
The software and design rules alone is proprietary, costs obscene amounts of money, and their manufacturers have decades of experience.
Yes, I know about the ONE FOSS CPU design there is.
But realistically, there’s no way to compete against the players already in the industry. Especially if you come with a FOSS perspective.
https://tomverbeure.github.io/rtl/2018/12/06/The-VexRiscV-CP...
https://www.oracle.com/servers/technologies/opensparc-overvi...
https://opencores.org/projects?expanded=Processor
I agree that manufacturing is a real problem. FOSS processors can get adoption but until we open up the manufacturing pipeline they won't provide end users with the benefits we get from FOSS software.