Phoronix Test Suite Exploring GPLv2 License

[V10lator]

I always wondered how that worked, as it's "at your option". Whose option? Distributor? End-user? It makes a difference. If the distributor can choose to only consider it GPLv2 at the moment of distribution, then it's obvious why they accept it, as it means they can choose not to follow the patents and anti-tivoisation clauses. If it's end user's, then any of those distributors are effectively giving me licenses to their patents "at my option".

That's exactly what I'm asking myself all the time, especially as I use gentoo, so basically I'm forking the software I install every time I install it (so the "at your option" has to mean me, no?).

On the other side every end-user can ask for the source and compile for himself, choosing the GPL version "at his option" while doing so, so, well, I'm, confused.

//EDIT: A funny imagination I just had: At a court:
Company A: "In our option we choose GPLv3"
Company B: "But in out option, as original creator of the software we choose v2"
Judge: "In my option I choose v4, which doesn't exist. You all loose!"
:P

[deanjo]

I didn't realize until I read your post, but yeah, it becomes circular: to be (legally) able to decrypt it, you must first prove it's under the GPLv3, and if you'd be able to prove that, you wouldn't need to decrypt it to prove it was under the GPLv2 in the other case, so there's no difference.

Not only that but it is like saying "Software Piracy will stop if they stop using DRM". People that steal open code and then cover their tracks don't care what the license says. The only thing a license does is place restrictions on people that are honest to begin with.

[mrugiero]

Or patent protection for users under paragraph 11? Say, you used some mechanism of calling or benchmarking - and you, and all your userbase are sued over patent related to this.

You are wrong here. If the patented code was implemented by Michael, there is no protection until the patent holder distributes his code, as there was no agreement on the patent holder side to allow this license.

GPL 3 doesn't realistically prevent that either. If someone is going to obfuscate the code to hide the source it doesn't matter what license they use.

I didn't realize until I read your post, but yeah, it becomes circular: to be (legally) able to decrypt it, you must first prove it's under the GPLv3, and if you'd be able to prove that, you wouldn't need to decrypt it to prove it was under the GPLv2 in the other case, so there's no difference.

You mean the default "GPLv2 or later" ? :P
I can't see why companies dislike GPLv3 but like "GPLv2 or later", given that "or later" means v3...

Anyway, we should all make small contributions to PTS now, then disallow a license change...

I always wondered how that worked, as it's "at your option". Whose option? Distributor? End-user? It makes a difference. If the distributor can choose to only consider it GPLv2 at the moment of distribution, then it's obvious why they accept it, as it means they can choose not to follow the patents and anti-tivoisation clauses. If it's end user's, then any of those distributors are effectively giving me licenses to their patents "at my option".
I don't really see the point on dual licenses, at least if you allow the user to pick between them under the same conditions. Just use the lowest common denominator (in how restrictive they are), as it's how it will effectively work. If I dual license something as BSD/GPLv3, all of those who are negatively affected (from their point of view, that is) will use it as BSD, and the ones who care about free software will contribute back, which would happen nonetheless with BSD only.
A different story would be in the cases of paid licenses, or in the cases where you care, for example, what they do with the software, for example, imagine you have the following ideology: all of the public administration software should remain open, but otherwise you don't care. Then, you could license your code as BSD for non-public administration applications, and GPL for public administration uses. Although this could be said to restrict the use of the software (although in sum you could use it any way you want, it's just with a different license), infringing both licenses.
Point is, dual licensing makes sense only (or maybe almost only, and I'm missing some cases) if you impose different conditions for the licenses to apply.
One exception might be this, GPLv2+, as I'm not sure exactly how it works.

[V10lator]

How about dual-licensing? That way everyone can choose the license that best fits his needs.

You mean the default "GPLv2 or later" ? :P
I can't see why companies dislike GPLv3 but like "GPLv2 or later", given that "or later" means v3...

Anyway, we should all make small contributions to PTS now, then disallow a license change...

[amehaye]

How about dual-licensing? That way everyone can choose the license that best fits his needs.

[deanjo]

Not giving advantages?
How about paragraph 5&6, that in GPL2 requires you to ship CDs with source code, not acknowledging internet as distribution medium? How about someone, who downloaded PTS from your site asks to ship CDs to China or Kazakhstan?

It does not require you to ship CD's, it requires you to have the source available in machine readable form.

Or tivoisation freedom in paragraph 3? Say, someone forks phoronix on "benchmarking device" that effectively is encrypted, so that you can't prove he is using it with his own modifications.

GPL 3 doesn't realistically prevent that either. If someone is going to obfuscate the code to hide the source it doesn't matter what license they use.

[gQuigs]

Meh..

Please post their actual concerns so we can discuss them. Are they planning on distributing PTS themselves to their customers? Otherwise, they would just be nervous about the Free software world getting some of their patents that apply to the PTS codebase. In which case, we definitely need them to use the GPLv3 ...

I say stay on GPLv3... and hopefully we can make this a learning opportunity for them to become more comfortable with the GPLv3.

[darkbasic]

Please don't.

[brosis]

There's also EUPL if you want something with more options ? though I haven't seen many users.

This license makes only sense for derivative works, such as EU software projects. It is worse than GPL on freedom due to higher compatibility and higher uncertainty (see paragraph 5 of EUPL) if one of licenses conflicts with EUPL. It is more complex, older, while having no advantage over GPL. It also limits legal application to EU countries.

[brosis]

Not giving advantages?
How about paragraph 5&6, that in GPL2 requires you to ship CDs with source code, not acknowledging internet as distribution medium? How about someone, who downloaded PTS from your site asks to ship CDs to China or Kazakhstan?
Or tivoisation freedom in paragraph 3? Say, someone forks phoronix on "benchmarking device" that effectively is encrypted, so that you can't prove he is using it with his own modifications.
Or patent protection for users under paragraph 11? Say, you used some mechanism of calling or benchmarking - and you, and all your userbase are sued over patent related to this.
Or license withdrawal protection which makes CLA unneeded, that is present in GPL3 paragraph 2, but absent in GPL2? (keyword "irrevocable") So that a contributor can't revoke his contribution from you.

What about incompatibility between GPL3 and GPL2, the main concern that prohibited Linux to switch to GPL3 (and it was NOT anti-tivoization)? GPL3 allows to add exceptions into section 7, that GPL2 does not allow, making GPL3 more compatible with other open source licenses.

GPL2 is outdated, stating "We consider switching to GPL2" is like stating "We consider switching to Kernel 2.0". This has no sense, except for explicit compatibility reasons. Its a clear downgrade. IMHO.