… ZFS … Oracle …

… I see licenses as more about protecting the dev from trouble (legal or financial) than to really dictate terms to users. …

ZFS aside​

Oracle has a less than stellar reputation with regard to audits and scope.

March 2022:


Second top-level comment under a 2019 post about the Oracle VM VirtualBox Extension Pack:

We pay for the pack and still get threats from them. It's a different a-hole from oracle every time and once we tell them our purchase info they leave us alone for a couple months.

2022 follow-up to that comment: <https://old.reddit.com/r/sysadmin/comments/d1ttzp/-/i3yhbwx/?context=6>.

 
Yeah, if Oracle starts behaving like that when Java is involved, I see that as a reason to migrate to Rust...

I started this thread as mostly a way to vent about people who talk about licensing without understanding what it really is about. But I'm seeing this conversation evolve to include examples of where users (private and commercial) would disagree with license terms - and I think that's a good thing for the conversation.

My thinking (in response to grahamperrin 's post) goes, Oracle did try to draw some conclusions after watching the MPAA (2004) and SCO/UNIX (2003-07) debacles. Oracle does have an interest in monetizing their 2009 acquisition of Sun and Sun's technology... It's called Return On Investment. But it sounds like they're trying to enforce commercial Java licenses. If a commercial outfit A doesn't want to pay for Oracle's Java EE (Enterprise Edition), it's a much easier business decision to use something else (like Rust) to implement a solution that is a decent moneymaker for A. Commercial outfit A has its own interests and investments to protect, and if Oracle wants more money than A is willing to pony up for convenience of being able to use Java, then yeah, we all gotta decide if using Java is worth the trouble and the money. And that's called 'Free Market Economy'. Yeah, FME is a bit murky and weak on software, but it's difficult to argue that you deserve money for your solution when viable alternatives do exist.
 
So, why do people talk like the software license matters, when it really doesn't (for most people) ? After all, how many people are actually in a position to even care?
You are asking that question from the wrong entry point, namely most people. Most people just care about the product, not so much the license.

The right entry point to ask that question for is the software developer. And for developers licenses are important, even more so for open source developers.

And here it depends on the preferences of the OSS developer, and how he views the world, namely:

* you are happy if your project can be used everywhere, but nobody legally has to give back aside mentioning the project is used somewhere, because you do believe in good behaviour and just want to encourage others to do so? Use BSD or something similar.
* you think corporations are evil, and only will give back if there is a legal lever which requires it? Use GPL.

Another OSS developer maybe wants to prohibit that his work can be used in military or intelligence technology, because the idea that his piece of software could drive an ICBM or be used to spy on people really drives him nuts, and really wants to forbid that.

This is what software licenses are about.
 
It's deeper than that. As a file system, ZFS really wants to be in the kernel (or at least have extensive hooks there). In the case of Linux, that means that the code HAS TO BE licensed under GPL version 2. Why is this? Because Linus has said so. The kernel components of ZFS are licensed under the CDDL, which is not compatible with the GPL. From a technical viewpoint, it is obviously possible to make ZFS compile and run (even extremely well) under Linux. The few Linux distributions that ship ZFS are perfectly legal; they distribute ZFS as a patch to the kernel. But Linus will NOT UNDER ANY CIRCUMSTANCE allow that to be distributed with he standard kernel, and all the big distributions are unwilling to ignore his wishes. So the basic reason here is that Linus is the malevolent tyrant for life, and a jerk.
You got it totally wrong here. If somebody, like e.g. RedHat, would want to create a Linux kernel with ZFS in it, they could do it technically. They would just fork the kernel and start working.

The problem though is when RedHat would start shipping such a RHEL kernel commercially with ZFS as kernel component enabled what Oracle would do. Most likely cease and desist subpoenas, if not more.

And that's the reason why nobody dares to do this. It is not because of would Linus Torvalds says, it is because of what Oracle would do. Nobody wants to try this out for good enough reasons.

We got some insight about this when Oracle sued Google for using Java's API without paying licensing fees. Oracle definitely would sue the company shipping such a modified kernel.

Given that it is totally understandable that Linus Torvalds does not embrace having ZFS in the mainline kernel at the moment, because he does not want to be at the receiving end of Oracle's lawyers. If anybody thinks else, they are free to try it out.

And he made it quite clear as well:

Other people think it can be ok to merge ZFS code into the kernel and that the module interface makes it ok, and that’s their decision. But considering Oracle’s litigious nature, and the questions over licensing, there’s no way I can feel safe in ever doing so.
And I’m not at all interested in some “ZFS shim layer” thing either that some people seem to think would isolate the two projects. That adds no value to our side, and given Oracle’s interface copyright suits (see Java), I don’t think it’s any real licensing win either.

And honestly, there is no way I can merge any of the ZFS efforts until I get an official letter from Oracle that is signed by their main legal counsel or preferably by Larry Ellison himself that says that yes, it’s ok to do so and treat the end result as GPL’d.

If somebody adds a kernel module like ZFS, they are on their own. I can’t maintain it, and I can not be bound by other peoples kernel changes.
 
It is said that licenses are there to protect the rights of the creator. While theoretically true, this is not what happens in practise. In practise, licenses are assets used in the powerplay of the mighty. Creators have no benefit from that.

This is not about ZFS. If ZFS were in the Linux kernel, FreeBSD would be forgotten by now. We're actually lucky, because this is something we didn't do, it just happened to us.
And this again has nothing to do with licenses, only with side effects from the powerplay of the mighty.

Berkeley has a do-whatever-you-want license. As a consequence, in the times before linux got so extremely popular, practically every router on the Internet had Berkeley code in the box. Even commercial closed-software unix brands happened to have the same bugs as we fixed in our kernel. Did we get any benefit from that? No. Because it is allowed to do that, and one cannot call it stealing.

We, Berkeley, have a complete and well fitting OS that runs perfectly fine. Linix, OTOH, have just a kernel and a bunch of assorted stuff. Nevertheless Linux got extremely popular. If there were no Linux, would our OS be similarly popular? No, because shops would just cannibalize it and sell it under their own name. With Linux they cannot do that, they must provide it as Linux. And as a consequence, everybody today knows Linux.
 
Wrong, there's a long list of court trials where the license was put to the test - mostly the GPL of course, because the BSD license rarely has room for any trial at all. And often the plaintiff won the case. And the part in it which protects the rights of the creator most in any license is where it states NO LIABILITY.

 
You are asking that question from the wrong entry point, namely most people. Most people just care about the product, not so much the license.

The right entry point to ask that question for is the software developer. And for developers licenses are important, even more so for open source developers.
Exactly, licenses matter to developers, rather than the vast majority of users. Earlier in the conversation, it was pointed out that you first have to have something that is worth protecting with a license. It's just that there's a lot of people who seem to want to start a software project, and they want to talk about licenses before doing anything else, even having a working demo. The very concept of licenses does command a lot of money, we see that in the news all the time. Money is of interest, sure, but lining up license details is pointless until you have something running.
 
Other people think it can be ok to merge ZFS code into the kernel and that the module interface makes it ok, and that’s their decision. But considering Oracle’s litigious nature, and the questions over licensing, there’s no way I can feel safe in ever doing so.
And I’m not at all interested in some “ZFS shim layer” thing either that some people seem to think would isolate the two projects. That adds no value to our side, and given Oracle’s interface copyright suits (see Java), I don’t think it’s any real licensing win either.

And honestly, there is no way I can merge any of the ZFS efforts until I get an official letter from Oracle that is signed by their main legal counsel or preferably by Larry Ellison himself that says that yes, it’s ok to do so and treat the end result as GPL’d.

If somebody adds a kernel module like ZFS, they are on their own. I can’t maintain it, and I can not be bound by other peoples kernel changes.
That's basically what I said in post #6...
 
Wrong, there's a long list of court trials where the license was put to the test - mostly the GPL of course, because the BSD license rarely has room for any trial at all. And often the plaintiff won the case. And the part in it which protects the rights of the creator most in any license is where it states NO LIABILITY.

That wikipedia link told me that there is a name for the beast that gets people to want to line up license details before they even begin to matter - Copyright Trolling. CT in this case is basically strategic litigation

That same link also mentioned a trade secrets case where the court ruled that GPL shouldn't have been applied to the source code.

Well, this is kind of devolving into a 'Legal research' thread... While I think it's interesting and useful in its own right, I was hoping to keep the scope to 'Playbook for getting a software project off the ground'... As in, maybe start by having a tool that actually works and has the features you want first, and worry about licensing later? A lot of people seem to have that idea just completely backwards, at least in part because Copyright Trolling is that loud and nasty. 😩
 
Eventually licensing a software is a legal decision, and often goes beyond what the author really wants.

Here some insight: https://writefreesoftware.org/learn/participate/choose-a-license/

Personally i divided licenses in two categories: the ones that protect the end-users and the ones that protects the developers. I am not an expert in anything of these topics, I use a simplistic method: all the licenses involved in the Linux kernel development are meant to protect the developers which most of the time sign a CLA to drop their copyright for the company they are working for; this imply that the majority of the licenses discarded by the Linux kernel are unavoidably pro end-users.

The GPL3 is better to avoid the corporate manipulation than any GLP < 3; MIT/BSD for me are even better because force the corporations to be really altruistic when they contribute upstream and this is the reason why is mostly avoided for a collaborative endeavor.

Some GPL supporters often criticize the MIT/BSD license affirming that, with exclusively them available, the opensource / free software as we intend it today would never existed.

I think the BSD like licenses born when computers were more important than the software who ran on them, that was the time to selling computers not software; when the software became rapidly the real thing then it born a different awareness around it, and therefore the idea of the four freedom evangelized by RMS which are perfectly respected by BSD/MIT, actually they offer a fifth freedom that the GPL won't ever allow.

Since I belong to the ones who use the free software I tend for MIT/BSD and GPL3 like licenses, also I don't dislike the viral aspect of the GPL I believe that is a good thing. I consider that TechBros Corps really distorted and abused the software licensing, the evidence is the constant infighting between corporations and their predatory behaviors that eventually create all these deformations like "Server Side Public License".

All this headache about the GPL is among the reasons that pushed me away from Linux and let me discover the BSD universe.
 
The GPL3 is better to avoid the corporate manipulation than any GLP < 3; MIT/BSD for me are even better because force the corporations to be really altruistic when they contribute upstream and this is the reason why is mostly avoided for a collaborative endeavor.
The GPLv3 was amongst other things designed to avoid Tivoization in the future. What this is: while the software is open source, only digitally signed packages are able to run on the hardware.

In other words: you own the software, but not your hardware. GPLv3 has passages in it to prevent this. Linus Torvalds by the way was of the opinion that tivoization should be possible.
 
That is pretty rich given that the GPL was proposed as the license that was working and no longer is - but any new license to appear now to solve the perceived problem would add incompatible clauses so that you can't mix it with the GPL.
Actually, this is one of my frustrations with licenses in general - There's plenty of complexity already, there's plenty of trolls who don't like how I use my copy of the software on frankly MY friggin' hardware - at some point I just stop caring about licensing intricacies, I just want my device to work and be useful to me, please???

it's bad enough that there's no real escape from technical dependency hell (Some of it real, some of it just bullshit fabricated on a whim), licensing intricacies and issues only makes it worse.
 
What are the largest projects that use GPL-3 (not GPL in general)? I can think of bash, gimp and ansible.
EDIT: cracauer added rsync to that list, thanks.

I think the Linux kernel and the GNU userland use GPL-2, right? Corrections and additions please.
 
What are the largest projects that use GPL-3 (not GPL in general)?

I don't know about largest in general, but in GitHub (sorted by number of stars):
That's for GPL-3.0.



In contrast (don't shoot the messenger):

Don't Say “Licensed under GNU GPL 2”! - GNU Project - Free Software Foundation

<https://github.com/search?q=license:GPL-3.0-or-later&type=repositories&s=stars> finds nothing, <https://github.com/search?q=license:GPL-3.0-only&type=repositories&s=stars> finds nothing.
 
Here's a nightmare scenario made possible by the GPLs.

LGPL, GPL, GPLv2 and GPLv3 all have a clause that permits relicensing the work under a newer version of the GPL. Think well into the future long after some of us are gone from this world. The ideals spouted by RMS are forgotten. What survives is a kind of brand loyalty where the Linux brand is treated like Gucci or Fanta. Imagine the FSF gets co-opted by corporate interests or decides - the best way to beat the corporations - is to become one. They roll out GPL version X which is basically a pay-to-play license that guarantees an unlimited revenue stream. This war chest is used to destroy anyone that dares challenge them in civil court.

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or at your option) any later version.

Unlikely - but stranger things have happened. Maybe mwl will adopt this plot in his next techno-thriller.
 
For what I understood it allows the freedom of the BSDs together with the virality of the GPLs, and is that kind of license that is prone to protect the users rather than the authors.
EUPL says it's supposed to be like LGPL. GPL allows use with it through dynamic linking, if the GPL code makes the exception for EUPL. Perhaps that's considered part of a larger work. GPL has problems with compatibility. Other instances depend on how FSF wants to interpret its GPL licenses for compatibility, and what it states is a good exception.

For now, EUPL looks like a good LGPL alternative, which LGPL allows dynamic linking. Dynamic linking is good enough. EUPL allows static linking with other licenses which allows it, which is better. LGPL2 and permissive licenses have more default widespread use than EUPL, because they by default have compatibility with GPL2 and GPL3. EUPL doesn't have default compatibility, and has to be allowed by the authors.

EUPL makes clear that a restrictive constraint is different than a requirement, which is why GPL is allowed to absorb BSD licensed code, because maintaining the BSD license disclaimer is a requirement, rather than a use restriction.


I was arguing for a license similar to LGPL to protect code, while allowing any other use of linked code. Whether or not that be compatible with GPL, due to their unnecessary restrictions. If EUPL allows GPL to use it through dynamic linking too, that's even better. I was for gradually replacing the GPL, by using better licenses.

I was also arguing for a license that keeps its hold on all derivatives forever, but allows freedom through linking, so that no matter how many forks there are, preserving anyone will preserve the spirit of lost code. Also, to prevent code from getting locked away from the code under this license. We could recreate lost code in the spirit of code which was forgotten or lost.

Here's a nightmare scenario made possible by the GPLs.

LGPL, GPL, GPLv2 and GPLv3 all have a clause that permits relicensing the work under a newer version of the GPL. Think well into the future long after some of us are gone from this world. The ideals spouted by RMS are forgotten. What survives is a kind of brand loyalty where the Linux brand is treated like Gucci or Fanta. Imagine the FSF gets co-opted by corporate interests
LGPL2 offers the most compatibility of the GPL's with other code. LGPL3 lacks compatibility with use with GPL. In essence, the problem you described allows LGPL and permissive licenses to get locked away into that ecosystem of license. It likely won't happen to that extreme, but I like it so that code is permanently kept from any such scenario.

As long as any copy exists under the more permissive license, than GPL or worse, the nightmare scenario, that copy will last as long as the project. The type of license I argue for, which may not currently exist, is meant to be eternal, no matter how many projects die out, as long as one copy remains.


I was also wanting something less viral than GPL, which had clear defined limits, especially on dependencies of it, had freedom to use any opensource library, or even as directory wide weak copyleft.

The Linux kernel for instance, isn't so bad for programs which use it, because it has a defined limit to how far the licenses reaches, at the API, other programs tend to forget this, and assume other software can be run with it. Where the Linux kernel is bad, is that it forces other drivers to be under LGPL2.


Also, Thread whats-a-permissive-license-and-why-should-i-care.92068/. This relies on spreading permissive license ideology and relies on good will. It's great for original works, but its limit is at forks which choose to regress back to overly viral or restrictive licensing. I'm for something more that indefinitely locks in freedom of use for all iterations and that indefinitely prevents code from getting locked into something extremely viral or theoretically worse.
 
I think forced sharing is downright anti-social and overbearing; no matter the flavor. I'd never want to put that restriction on another fellow developer. The appeal of the BSD license, to me at least, is that the no-strings attached premise, and it keeps everyone honest. Should someone close off or relicense code in another license; you can sort of discern their intentions. Keep it copy-centered and everyone wins IMO.
 
I think forced sharing is downright anti-social and overbearing; no matter the flavor. I'd never want to put that restriction on another fellow developer. The appeal of the BSD license, to me at least, is that the no-strings attached premise, and it keeps everyone honest. Should someone close off or relicense code in another license; you can sort of discern their intentions. Keep it copy-centered and everyone wins IMO.
It's a paradox. No one should do it, yet the license allows it. Arguing for a permissive license is also inadvertently arguing for allowing the rights for it to be locked away for antisocial, overbearing or other purposes.

The best argument for permissive licensing is for libraries, and for other widespread use which can be used by GPL2 and GPL3. To a lesser extent LGPL2 can be argued for in this way.

I'm for, preserving all iterations, allowing all use with opensource and proprietary, keeping a separation, and limiting the viralness of that code, and locking it in for all iterations is how to do such. It's the best balance, which isn't a paradox. This way, companies can keep their proprietary code, thus using the opensource through linking.

When there's no reason for it, it should be prevented from being kept out of being locked away from use. The main benefit for permissive licensing is allowing compatibility for dependencies and libraries, and rebranding to another license which preserves freedoms in a nonviral way.
 
It's a paradox. No one should do it, yet the license allows it. Arguing for a permissive license is also inadvertently arguing for allowing the rights for it to be locked away for antisocial, overbearing or other purposes.

Nope, I don't buy it. What should or shouldn't be done is irrelevant; so I don’t think it’s inadvertent when the premise of the BSD license is entirely different. This is Stallman-speak. The freedom to choose what one can do with their derivative work is more "fair/moral" than being coerced into sharing. Has the GPL ever even been successfully tried in court anyway? If the point of open source is “freedom”; there no free-er license than the BSD license. Developers write code; there's little incentive for open source when ones behavior is dictated IMO. WIth the GPL, you're trading liberty for the possibility of preventing commercial exploitation.. not my cup of tea.

All else is just politics.
 
Back
Top