Hacker News Reader

Comment by rogerbraun | Hacker News Reader

Lots of comments are talking about how Apple "doesn't owe [the FSF] anything" and that it's a "business decisison". This is all true, but it misses the point.

The point is that the current trend of companies releasing stuff under non-copyleft licenses and keeping the 'good stuff' in proprietary forks is bad for free (FSF-style) software. That's why the article argues that you should take a good look at what these companies are actually doing beyond press releases and act accordingly.

You might still think that free software is unimportant or that having pseudo-freemium open source release is still better than nothing. But that's your own decision, based on your own values, which might not be the values the FSF thinks of as important.

View on HN

> The point is that the current trend of companies releasing stuff under non-copyleft licenses and keeping the 'good stuff' in proprietary forks is bad for free (FSF-style) software.

Copyleft is meaningless without Copyright. Apple created Swift, and thus owns the Copyright. As the Copyright owner, Apple can licence/release Swift or parts-thereof, as it chooses. Even if they did release Swift under the GPL, nothing would prevent them from also using a non-released version of Swift or from keeping certain parts/libraries in-house.

> You might still think that free software is unimportant or that having pseudo-freemium open source release is still better than nothing.

I think its both insulting and an incredibly narrow-view to call permissively licensed software "pseudo freemium".

>I think its both insulting and an incredibly narrow-view to call permissively licensed software "pseudo freemium".

I don't mean the general case of permissively licensed software, but the way that the article expects Apple to act.

Either you/the author are claiming Apple will break the licence terms they themselves have chosen to release the software under, or you're insulting permissively licensed software.

Which is it?

I'm not trying to insult anyone. The BSD and similar licenses are of course free software licenses, and the FSF does (again, of course) acknowlegde that.

Still, a company like Apple can use these licenses to take outside contributions in their proprietary software. Freemium might not be the right word for that, but I think it's not unreasonable to dislike these (completely legal) practice.

> Copyleft is meaningless without Copyright. Apple created Swift, and thus owns the Copyright. As the Copyright owner, Apple can licence/release Swift or parts-thereof, as it chooses. Even if they did release Swift under the GPL, nothing would prevent them from also using a non-released version of Swift or from keeping certain parts/libraries in-house.

If swift is GPL, Apple cannot keep secret changes it makes on top of third party contributions.

If swift is MIT/ISC, Apple can keep changes made on top of third party contributions secret.

Oh like Oracle cannot make a commercial closed version of MySQL because it is under GPL[0] or QT[1] cannot have both a closed and a GPL version.

There are a lot more examples of this.

On the other hand look at how RMS/FSF talks about LLVM and BSD[2], i am a developer there prefers the MIT license for my software, and for me FSF and GPL are a lot more dangerous then apple and there open source license.

[0] http://www.mysql.com/about/legal/licensing/oem/#4 [1] http://doc.qt.io/qt-5/licensing.html [2] https://lwn.net/Articles/582242/

> Oh like Oracle cannot make a commercial closed version of MySQL because it is under GPL[0] or QT[1] cannot have both a closed and a GPL version.

Oracle can't pick up third party patches and incorporate them into their proprietary version (for example: patches from MariaDB). They can only incorporate their own changes, and changes that other sign over.

As I see it permissively licensed code puts every one on equal terms. We all can make open source forks & we all can make proprietary forks. With copyleft some one is more equal and it is not the users, since we can only make copyleft forks.

Except that lone developers or small teams and large corporations aren't equal. A lone software developer who works on software under a permissive license is not benefitted to the same degree by those supposed "equal terms", because he doesn't have the resources of a larger team or corporation to exploit that "permissiveness".

Or you could contribute back to the copyleft project (without copyright assignment or relicensing grants) and then you're all on equal terms again: nobody can make proprietary forks.

Why should i take the freedom of making proprietary forks away from people?

My goal with open sourcing my software are to make the (software) world just a little better, i do that by writing the best software i can and hope that as many people possible use it, if i write the best ftp library in the world and you need a ftp library for your proprietary program i hope you use mine instead of an inferior implementation.

In my world view i do not loss anything by you making a proprietary forks of my work (i still have my version of my work with my license) but you and your users are getting a better product.

If your priority is to improve the quality of the software that people use, I don't think you should use copyleft.

If you think that quality is important, but that as important as that is that users should have control over the software they use, as part of the control they should have over their own lives, then I think a copyleft license is a good tool to help achieve that.

My comment was in the context of big company X, "releasing something in a permissive license" vs "releasing something in a copyleft license". Either way I prefer to use, support and contribute to permissively licensed projects.

Most corporate copyleft projects only accept contributions with assignation.

If swift is GPL, Apple cannot keep secret changes it makes on top of third party contributions.

AFAIK, even the FSF requires copyright assignment – Apple likely would too, and this means any 'official' work on Swift will can be released by Apple under any license they wish.

Of course, if you don't agree with that, you are free to license your contributions as GPL and release them to the community without assigning copyright. They just won't be part of the official distribution, and Apple can't build on them. That's the best of both worlds, then.

>FAIK, even the FSF requires copyright assignment – Apple likely would too

Assigning copyright to a 501c(3) non-profit is much different than assigning copyright to a for-profit multi-billion dollar corporation.

I've completed the copyright assignment form for some of my contributions to GNU. The FSF's terms are reasonable and the legal text prevents the FSF from being able to make the software non-free in the future. I really dislike copyright assignment in general, but the FSF is in a unique position and I appreciate that they can defend the GPL on my behalf.

I invite you to google "Contributor Licence Agreement". Either way I don't see why we are arguing about it, Swift will be released with a permissive license, which is a good thing :-).

> I invite you to google "Contributor Licence Agreement".

I know what they are, and that doesn't change anything. Apple can't force people to sign a Contributor Agreement. Look at, for example, what happened with MariaDB and MySQL. Oracle can't force the MariaDB devs to sign the agreement, nor would Apple be able to in similar situations.

>pseudo-freemium open source release is still better than nothing

No - it's much better than copyleft - the only point he has is the lack of patent grants - which is achievable without copyleft (like Apache license).

Apache gives you the code and the ability to integrate the code in to your product where the copyleft part might not fit.

Open source isn't some ideological dogma for me or these companies - it's a tool to lower the cost of maintenance/support (eg. community is very good at maintaining ports for niche use cases you don't have the resources to support), gain exposure and provide better tools for developers (it's much easier to work with tools that you have source access to).

If you see open source as that then (A)GPL becomes a niche license that's useful for stuff like giving people free access but knowing that copyleft will unacceptable for most commercial applications so they will need to license a custom version (eg. this is the business model of companies like Qt, Xamarin, etc.)

Being a useful tool for businesses is not the main goal of free (FSF-style) software. The freedom of the user to read the source and modify the software is. That's why they care about the distinction between open source and free software.

You can disagree with this, but I think just dismissing it as ideological does not do it justice.

>The freedom of the user to read the source and modify the software is.

A freedom 99.9999% can't even use (because they can't read code or the code in question). These aren't 70s - code for modern applications is huge/complex and requires familiarity with the field to understand - your average programmer can't use the code even if he/she has access to it.

Instances where code access is actually meaningful is closely related to someones expertise domain and that's also related to business (including "hackers" - as modern hackers are trying to monetize their ideas) and while GPL isn't a blocker for commercial use by default it usually ends up being a no-go in the practical sense.

The other counter-intuitive reality of GPL is that as long as you keep your changes private you aren't required to release the code to the public. This can lead to some ironic cases where for eg. I can modify a GPL project to integrate with my code and produce something valuable - and if I wanted to provide those changes for free to give back to the community - but not release my code - I can't do that - so in the end I get the benefits and can't (for business reasons) give back.

I can understand LGPL - you explicitly ask for changes to the code back so you can benefit from them - this is just a formalization of the implicit assumption that people will push back their changes to simplify their maintenance costs.

Opensource licenses give you all the practical benefits of free software you need without the downsides.

Most people benefit indirectly by using forks created because those rights exits. And you can also pay someone to work on it for you: while this is rarely true for consumer software, we see it all the time in the business world.

The other counter-intuitive reality of GPL is that as long as you keep your changes private you aren't required to release the code to the public.

That's counter-intuitive because, unfortunately, people have the wrong idea about the purpose of the GPL. It's not about giving back, it's about paying-it-forward to the next users. So it makes perfect sense that if you don't have users, the GPL doesn't force you to do anything.

Opensource licenses give you all the practical benefits of free software you need without the downsides.

That's another misconception; the GPL is also an open source license, and the BSD/MIT are free software too.

Almost all open source licenses are free software, and vice-versa.

>Most people benefit indirectly by using forks created because those rights exits. And you can also pay someone to work on it for you: while this is rarely true for consumer software, we see it all the time in the business world.

You're arguing my point here - most people don't care about the code - they care about the derived work - so non-copyleft licenses serve them even better because they allow for more use cases (more forks). Even if my fork is commercial if you're willing to pay for it it means that existing code doesn't serve your purpose - therefore you benefit from that closed source/commercial fork. In most cases the commercial fork would not even exist under copyleft model so you can't argue "they won't benefit from further forks".

>That's another misconception; the GPL is also an open source license, and the BSD/MIT are free software too.

I think it's clear that I meant that "non-copyleft open source licenses give you all the practical benefits of open source without the downsides of copyleft"

Yes, but the point is to promote forks of forks (and so on). If you close it, the branch dies there.

But then we come back to my original point that you're (GPL people) arguing for ideology (idea that being able to crate forks is intrinsically valuable) and i'm arguing that copyleft means less forks in the first place and that a closed source fork is better than no fork.

By the same token, it might also be nice to see FSF advocates taking on board the notion that they have brought some of this on themselves with newer versions/variants of the GPL.

While I consider the GPLv3 family to be shoddy licenses in practice (I wasn't aware that it's possible to build a wall of text that makes seasoned lawyers run away, screaming), I understand the rationale.

First, that there were devices shipping GPLv2 code that couldn't be replaced. Remember that the entire FSF story began with a printer driver that couldn't be replaced. Tivo struck a nerve there.

Second, the SaaS phenomenon which, again, works around the spirit of the GPLv2 for which the (already existing) Affero GPL license was integrated properly through some notion of a modular license (GPLv3 + extensions for the "L" or "A" specializations, instead of three completely distinct texts).

Absolutely, I hope I didn't convey a message that I think the licenses aren't without justification. I do think that they have gone too far though, and I think that is somewhat supported by the (as I see it, at least) lack of interest in v3/AGPL from large companies involved in FOSS.

If anything, from the part of the universe that I observe closest (cloud stuff mainly, especially OpenStack), the exact opposite of what FSF would want, is happening. Projects seem to be choosing far more strongly to go with much more liberal licenses such as the Apache Licence.

>You might still think that free software is unimportant or that having pseudo-freemium open source release is still better than nothing.

It's strange that you call permissive licence "pseudo-freemium" when the GPL is just as often in that way. The GPL is generally seen as a preferable licence to use in that situation since it's more difficult for competitors to make use of the code commercially - which means they are more likely to buy the paid licence. x264 is an example of a project like this. It's GPL licensed specifically so that people who want to use or modify it for use in commercial settings buy the expensive commercial licence (the lead dev considers other people using code he wrote to make money without giving him a cut to be "theft").

> The point is that the current trend of companies releasing stuff under non-copyleft licenses and keeping the 'good stuff' in proprietary forks

What makes you think they'll be doing that? They generally don't with clang/llvm (the only counter-example being that they briefly had a private llvm fork with ARMv8 support, but that was likely just rushing out ARMv8 support when they unexpectedly launched an ARMv8-based phone two years ago).

Reply on news.ycombinator.com