Hacker News Reader

Comment by AnkhMorporkian | Hacker News Reader

I'm sorry, I'm not seeing it. One person involved with the Linux Foundation wrote an article that used some terms this author and some others didn't like, and after some outcry the Linux Foundation removed it. Somehow this makes the entire organization a non-friend of Desktop Linux and the GPL?

It's just such a massive leap of logic. The GPL is in the most restrictive class of open-source licenses for companies, and it is viral by nature. Should those terms be sugar-coated? I didn't read an ounce of contempt in the quoted section.

View on HN

> One person involved with the Linux Foundation

...there are a lot of people involved with the Linux Foundation, what didn't help in this case is that the one person involved with the Linux Foundation who came up with this piece of Ballmer-era nostalgia is the Linux Foundation's Senior Director of Open Source Consulting.

It's a pretty major PR blunder. It rubs people the wrong way because it basically repeats the anti-Linux arguments which used to be promoted by a lot of companies that were heavily-invested in proprietary software -- some of which are Linux Foundation members today.

The terms aren't exactly innovative. 20 years ago or so they were used liberally (and, I think, coined) by a particularly disliked company from Redmond, and their marketing material (thinly-veiled as whitepapers, sometimes from partner companies) read pretty much like this guy's piece.

This is basically a blast from the past, except 20 years ago we'd see it on Microsoft's website under some Get the Facts stamp, not on the Linux Foundation's website.

I apologize for the late reply, I have been out of state making a pitch, and enjoying the scenery after it was accepted.

I can see your viewpoint. I do remember the anti-linux Microsoft days, and I did not enjoy them as a company back then. However, I do think viral is an apt description of the license, despite the objections in this thread. As I mentioned before I left, I think viral is a good description of GPL. Viral is a feature, not a consequence. If you choose GPL, it's because you want that viral nature. If you're a company, you'll have to weigh that license against your business concerns.

Zealots may say something like "every piece of code should be open source in all circumstances", but pragmatists will realize that some code just doesn't make sense as open source, and making all derivatives of that open source.

So, long story short, someone who is in charge of open source consulting should understand better than anyone else that the GPL isn't appropriate for all companies. If the changed code is patented, GPL isn't appropriate in any circumstance, like it or not.

If you want to combine viral open-source code with toxic proprietary code, sure. Proprietary code is toxic by nature, so why should we not use that term?

> I didn't read an ounce of contempt

I disagree. Calling a grant of permissions "restrictive", or calling it "viral" has as much contempt as calling a offering of proprietary licensed software toxic. Its insult flinging, plain and simple.

I'm not sure I want to wade into this religious war, but you can surely agree that the GPL is more restrictive (as in, it adds restrictions) than say the BSD or MIT licenses.

When you rent a hotel room you get permission to enter and sleep there, but are also restricted to not destroy the room. However the restriction already existed before you paid, so the change between the owner and you only exist in added permission. No one is allowed to destroy property of someone else so those restrictions are not additional part of the transaction.

GPL do not add restrictions. It is purely permissions given over what already exist. In theory one could rewrite the GPL license text to be lines of "I grant you permission to do X", and it would still have the same effect.

So, to bridge the analogy, what would be the "before paying" state of a GPL'd work? The GPL does not, to my knowledge, clearly define this, and there's disagreement among the experts.

> what would be the "before paying" state of a GPL'd work

Whatever the law is for copyrighted but unlicensed code.

What I'm asking is, at what time is someone bound by the terms of the GPL? See for example the IPC loophole, dynamic linking, etc.

Any time that you want to use the software and you're not the copyright owner. The terms of the GPL grant you permission to use it under some circumstances that you wouldn't have otherwise.

It is true that sometimes it is not clear when you are satisfying the conditions in the GPL that allow you to use the software, since there is no case law. But the constraints that forbid you to use the code all come from copyright law, not the GPL itself.

You are bound mostly when you distribute GPL covered works or derivative works. The licence is a-priori unconcerned whether such derivative work involve IPC or dynamic linking or anything else. There is no trivial relationship between the technical kind of SW interaction and if it constitutes a derivative work or not. Sometimes projects clarify their interpretation, e.g. Linux telling (with much reason!) that merely using standard syscalls does not per-se constitute a derivative work - (and note that this is a quite stronger form of interaction than most IPC, however this does not intrinsically prevents some IPC usages to be considered as derivative, maybe especially if such IPC are added after the fact in a vain attempt to circumvent the GPL with a meaningless technical trick)

Let me give an example of how to rewrite part of the copyleft concepts as pure permission rather then term and conditions.

"I here grant permission to distribute source code of the unmodified complete work"

"I also grant permission to distribute a binary version together with the corresponding preferred source code of the unmodified complete work"

"I also grant permission to distribute a compilation of a modified version of the source code with the source code of all dynamic linked software that it depends on"

Of course, to match a license like GPL you would need to make precise wording to get the permission to mean exactly one and only exactly one permitted action (lawyers, interpretation and loopholes), but they are just plain permissions. You can use one, two, or ignore any number of them and mix and match. Since they are isolated permission they hold no state and there is no "I agree to all terms".

This is also the main argument behind the concept of software license not being contracts. A contract goes beyond permissions and binds two parties together in one document, while permissions just exist regardless if anyone accepts it.

The GPL protects freedoms of users by placing restrictions on those that would do harm---developers (by ensuring that the code remains free) and distributors (by ensuring that the user has access to the source code).

If someone is hostile to restrictions on those two groups, then copyleft is undesirable, at the expense of the user, who has fewer guarantees of freedom.

The preferred term nowadays is "protective". You are given some constraints, but those are intended to protect the ability of people to keep using software from the project in the long term for any purpose.

If the net effect is to increase the overall freedom that people have for using the project, that is not necessarily more restrictive, other than in a local way. Metaphorically speaking, being able to sell yourself into slavery is not something that increases your freedom, and forbidding it isn't "more restrictive".

When you use the word restrictive without any other context to describe copyleft licenses, as you just have, it only serves to intentionally make the license look bad/unappealing, which is disingenuous at best. It's only restrictive in the sense that it restricts others from profiting off the hard work of the author without having to contribute anything back in return. For the author, the word restrictive is about the last thing I'd call it.

All I hear when I see people describing copyleft licenses as restrictive or viral is self-entitled whining that people can't do what they want with other people's work.

Edit: To be clear, I think the parent made their point in good faith and this is not meant as an attack on them.

It's it less restrictive in some ways, and more restrictive in others.

GPL offers protections for the users, so it prevents developers restricting the users freedom, GPL prevents restricting users so ONLY developers that want to restrict users freedom feel restricted. It is like in society where your freedoms are limited to prevent removing the freedom from others, like you are not free to kill people because you teke the freedom to live from them. So killing is illegal is protecting freedom not removing freedom.

> If you want to combine viral open-source code with toxic proprietary code, sure.

Even the Linux kernel qualifies as toxic proprietary code for the simple reason that GPLv3 is not backwards compatible. If anything the GPL is toxic to any other open source license including itself. Others at least try to play well with each other.

What is the point of GPLv3 if it is not allowed to have any difference from GPLv2?

That would be my question? It introduces an incompatible split with existing code that cannot be re-licensed by hand wave of the FSF for purely dogmatic reasons.

> Even the Linux kernel qualifies as toxic proprietary code for the simple reason that GPLv3 is not backwards compatible.

That is not true. Being "GPL compatible" is not the golden rule by which software is considered Free Software; granting the "4 user freedoms" is.[1] Since the GPLv2 grants those freedoms, it is Free Software.

[1] https://www.gnu.org/philosophy/free-sw.html.en

Now mix GPLv3 code with the Linux kernel without violating an FSF license. Of course that you cannot release the result just means that there are also zero users, so the freedoms of those zero users are preserved.

Edit: I am not arguing from an ideological standpoint, just from the resulting compatibility and usability issues. The GPL 2 and 3 were both written to protect the four freedoms, that they are incompatible and that we now have to invest time and effort work around that even for free software is just ugly.

Yes, but that's also true for any non-GPL compatible license, not just the GPLv3. And it's a result of Linus Torwalds not using the GPLv2 as intended by removing the "any later version" clause.

Well to be fair to the author the writer of the article has the job title:

Sr. Director Open Source Consulting Services

And they're posting on the Linux Foundation's site so that's a bit more than just some random person writing an article for the site.

Also the term "viral" comes off as an insult to the licence, I think the term GNU Public Virus used to be a common descriptor.

I think that's why the term Copyleft was introduced.

I don't see 'viral' as an insult, I see it as a feature. I'm certainly pro-GPL, but I don't use it in too many of my projects. When I do use it, it's because I want that viral nature. When I don't, it's because I want people to have the freedom to modify it on their terms. However, pretending that every project should be licensed under copyleft licenses or that there's no room for criticism of the GPL smacks of zealotry.

As for the author, I don't have the full article, but he isn't listed on the 'Board Member', 'Leadership', or 'Technical Advisory Board' sections on the Linux Foundation webpage. Everyone there seems to have pretty damn fancy titles. I have no idea of his actual position in the hierarchy.

It has a history of being used pejoratively. https://en.wikipedia.org/wiki/Viral_license

Many terms do. Socialists and communists still call themselves those terms despite McCarthyism and the red scare, the term queer is still used despite the fact that it was used as an insult for a very long time, the n-word's reappropriation despite its horrible history, and too many other examples for me to list.

Perhaps it's time for the FSF and its ilk to wear 'viral' as a badge of honor, instead of playing the semantics game.

Using the term "viral" is a pejorative term used explicitly to make the GPL seem dangerous. It's propaganda by the anti-copyleft crowd. Asking a representative of a foundation that represents the most important copylefted software project in history to stop using anti-copyleft propaganda terms isn't asking them to "sugar coat" their words.

Nonsense. It is an adjective that correctly describes the behaviour. If you want a non-viral GPL, then use the MPL2. It is exactly like the GPL, except it has a file-level firewall so that other files that it interacts with don't need to have the same licence, something the GPL is intolerant of. In every other sense the MPL2 provides all the protections the GPL does, it just doesn't consume the rest of the project you integrate it with.

It DOESN'T "consume" the rest of your project. Your code does not have to use the GPL as its license!

Unless the GPL code matches the Compiler or OS exceptions, yes, you do have to relicense as GPL. This is what GPL detractors refer to when they use the word "viral".

You don't have to relicense your code as GPL. You are confusing the copyright applied to your own code vs. distribution of the whole program.

Technically you are correct, and I guess that's the best kind of correct. However the GPL does apply to non-GPL code in the case of distribution, which is what I was referring to.

Perhaps with the advent of cloud computing and fast networks the more sticky problems of GPL code in proprietary systems become less of an issue. That said, I've already seen an uptick in AGPL code in some of the fields I'm interested in.

If you're going to say that the GPL is viral because it applies to derivative works, then almost any license is viral because the license of any work follows it around. That is the nature of licenses.

Anyone who expects to copy-paste other people's code into their own code and change the license is just fighting the legal system. You're still going to have to follow the terms of the license, whether that means sharing changes or including attributions.

The problem is that the GPL considers adding a single GPL method to a large non-GPL project as being a derivative work of the GPL method.

Moreover, it considers the interactions between the method and other code a derivative of the GPL, even if only the implementation was under GPL but the header/API was provided under a more liberal licence.

No. Copyright makes adding a single method to a large project a derivative work of the method.

Reply on news.ycombinator.com