3 ms·

> I dug up MongoDB's old blog post on why they chose the AGPL. Link me to it? I think I know which one you have in mind, but I'm not quite sure.

eesmith 7y ago
Thank you for asking me to double-check. I was wrong. I confused MongoDB with Redis. The earliest MongoDB AGPL post in Archive.org is https://web.archive.org/web/20090515015736/http://blog.mongodb.org/post/103832439/the-agpl https://web.archive.org/web/20090515015736/http://blog.mongo... . It clearly is concerned with software freedom and doesn't mention funding. (Though it does comment 'the goal is that you have to contribute those modifications back to the community.' But AGPL doesn't require contributions 'back to the community', only to the downstream users, yes?) The blog post I was thinking of, for Redis, is archived at https://web.archive.org/web/20180825041319/https://redislabs.com/blog/why-redis-labs-modules-are-agpl/ https://web.archive.org/web/20180825041319/https://redislabs... titled "Why Redis Labs’ Modules are AGPL". It makes the odd statement "Therefore, we wanted to make sure no one else can resell them for commercial purposes." I say 'odd' since anyone could resell them for commercial purposes under the AGPL, yes? It also has "In other words, if you modify a Redis Labs Module source code, the goal is that you have to contribute those modifications back to the community." So .. does AGPL really require contributions of modifications back to the community, as both Redis and Mongo wrote? If so, that's quite at odds with the GPL.
- kemitchell 7y ago
  Folks almost universally understand licenses by reputation, rather than by reading them. That's particularly understandable when it comes to AGPL, which is terrible to read. By reputation, AGPL is the "free for free software" or "you must share back" license, even though that's not what it says. The gap between reputation and reality afflicts both sides, developers and users. So MongoDB had to spend a lot of time sending users letters reassuring them that they didn't have to release their app code, even though they're commercial software companies making commercial apps with a (then) AGPL database. And so Redis Labs, on the developer side, held AGPL out as essentially noncommercial, and probably got away with it most of the time, since many of their potential users and customers misunderstood AGPL the same way they did. Some confusion about licenses like AGPL can be exploited opportunistically, and is. But much of that confusion arises quite without self-serving FUD, due to interplay between industry realities and license terms. I blogged about some of those recently, too: https://writing.kemitchell.com/2019/06/15/Industry-Environment.html https://writing.kemitchell.com/2019/06/15/Industry-Environme... You're right that nothing in any GPL license requires handing patches back to developers as such. But in practice, they often function that way. By giving users source code and a GPL license, you're giving users permission to hand that source back to the developers. Often enough, that's exactly what happens, especially now that the Web makes it so easy to do. There's still some risk in confusing the practical effect of a license with what it requires. But very well known developers and projects, like Linux, get away with it, at least to some extent: https://youtu.be/1Mg5_gxNXTo?t=47m20s https://youtu.be/1Mg5_gxNXTo?t=47m20s Linus is wrong in that video, when he says that GPL means "I give you source code, you give me your changes back, we’re even." But being wrong isn't costing him too much. Others using the same license are't so fortunate. For more on patches back and the GPLs: https://writing.kemitchell.com/2018/08/28/Unhappy-Coincidences.html#software-freedom-doesnt-mean-patches-back https://writing.kemitchell.com/2018/08/28/Unhappy-Coincidenc...

eesmith 7y ago

Thank you for asking me to double-check. I was wrong. I confused MongoDB with Redis. The earliest MongoDB AGPL post in Archive.org is https://web.archive.org/web/20090515015736/http://blog.mongodb.org/post/103832439/the-agpl https://web.archive.org/web/20090515015736/http://blog.mongo... . It clearly is concerned with software freedom and doesn't mention funding. (Though it does comment 'the goal is that you have to contribute those modifications back to the community.' But AGPL doesn't require contributions 'back to the community', only to the downstream users, yes?) The blog post I was thinking of, for Redis, is archived at https://web.archive.org/web/20180825041319/https://redislabs.com/blog/why-redis-labs-modules-are-agpl/ https://web.archive.org/web/20180825041319/https://redislabs... titled "Why Redis Labs’ Modules are AGPL". It makes the odd statement "Therefore, we wanted to make sure no one else can resell them for commercial purposes." I say 'odd' since anyone could resell them for commercial purposes under the AGPL, yes? It also has "In other words, if you modify a Redis Labs Module source code, the goal is that you have to contribute those modifications back to the community." So .. does AGPL really require contributions of modifications back to the community, as both Redis and Mongo wrote? If so, that's quite at odds with the GPL.

kemitchell 7y ago
Folks almost universally understand licenses by reputation, rather than by reading them. That's particularly understandable when it comes to AGPL, which is terrible to read. By reputation, AGPL is the "free for free software" or "you must share back" license, even though that's not what it says. The gap between reputation and reality afflicts both sides, developers and users. So MongoDB had to spend a lot of time sending users letters reassuring them that they didn't have to release their app code, even though they're commercial software companies making commercial apps with a (then) AGPL database. And so Redis Labs, on the developer side, held AGPL out as essentially noncommercial, and probably got away with it most of the time, since many of their potential users and customers misunderstood AGPL the same way they did. Some confusion about licenses like AGPL can be exploited opportunistically, and is. But much of that confusion arises quite without self-serving FUD, due to interplay between industry realities and license terms. I blogged about some of those recently, too: https://writing.kemitchell.com/2019/06/15/Industry-Environment.html https://writing.kemitchell.com/2019/06/15/Industry-Environme... You're right that nothing in any GPL license requires handing patches back to developers as such. But in practice, they often function that way. By giving users source code and a GPL license, you're giving users permission to hand that source back to the developers. Often enough, that's exactly what happens, especially now that the Web makes it so easy to do. There's still some risk in confusing the practical effect of a license with what it requires. But very well known developers and projects, like Linux, get away with it, at least to some extent: https://youtu.be/1Mg5_gxNXTo?t=47m20s https://youtu.be/1Mg5_gxNXTo?t=47m20s Linus is wrong in that video, when he says that GPL means "I give you source code, you give me your changes back, we’re even." But being wrong isn't costing him too much. Others using the same license are't so fortunate. For more on patches back and the GPLs: https://writing.kemitchell.com/2018/08/28/Unhappy-Coincidences.html#software-freedom-doesnt-mean-patches-back https://writing.kemitchell.com/2018/08/28/Unhappy-Coincidenc...