Making the GPL more scary

That is exactly the point, with GCC the community would have enjoyed those improvements.

With LLVM, Sony gets to cut down their compiler research budget, and the community sees whatever Sony feels like giving back, minor improvements without the juicy optimizer sauce that actually matters.

Personally I don't care, given I use mostly commercial stuff nowadays, but I do see the return of shareware long term.

Then MIT/BSD license advocates can consider how much those licenses are actually worth to the overall community.

> It seems like it is to me, it's just extremely protective of the source code owner's rights.

It is effectively impossible to follow the conditions of the license -- according to the interpretation by Bradley Kuhn and others this would require you to provide the source code for Linux under this license if you run MongoDB on Linux. Linux is GPLv2 and has thousands of copyright owners -- it is not possible for any individual to relicense it to any other license.

If you have an "open source" license (which violates the one of the entries in the OSI definition -- "License Must Not Restrict Other Software") which cannot be followed by anyone then it's obviously not an FOSS software. It doesn't matter that in theory it would be FOSS, what matters is whether it is FOSS in practice.

LWN isn't anti-GPL. I don't know what made you think this -- the title is obviously a reference to people thinking the GPL is "scary" and that the SSPL takes the GPL and actually makes it scary.

Weird thought, but I think that with this license MongoDB has managed to attack the OSI without attacking the FSF. To me, this is perfectly acceptable Free Software but so strict that it basically reads "do not use this software commercially, it requires you to open every piece of software that any of your employees have ever seen or heard of, including software you do not own. Purchase a license from us." In other words, commercial usage is allowed in the post-revolution Free world that Free Software hopes to bring about, where people have access to the contents of the computers that they interact with, and the freedom to mess with the ones they own.

The Open Source concept, however, is not philosophically based; its primary goal is to allow businesses to collaborate on building blocks, but not to prevent them from taking one of those blocks and carving it into something that they're not willing to share. Since their primary goal is to help business, they can't help but disapprove; since they don't explicitly state this goal, they have no nominal basis to reject it without rejecting Free Software in general.

I predict some eventual announcement that the license is "effectively closed source" and instead of accompanying that announcement with a specific supporting argument, they'll accompany it with a questionable written history of the OSI. They'll disapprove, or approve with asterisk.

edit: and, of course, what kemitchell said.

Individuals interpretations of this are unlikely to matter much until such a clause is tested in court. The interpretation that the license is as broad as Mr Kuhn claims may be true, but it may not be.

Personally I think that both Mongo and Redis doing this is great as it shows the danger posed when a company can just relicense a project at will, which IMHO is the greater issue here. If Mongo felt the need they could amend their license again if it was tested in court and the court ruled against them. Individuals and companies should be aware that with some of these license agreements they could change at any time.

Ultimately governance of a project matters. We have seen somewhat of a split in the Cassandra community between Apache Cassandra and DataStax. That may ultimately prove to be beneficial as a new set of organizations has started to contribute. Having an organization like Apache to provide oversight and help draw a firm distinction between an open source project and a vendors proprietary offering was crucial there.