What a weird reaction by that company. The time it took them to discuss it, investigate, and respond could've easily been used to comply with the license. I'm already happy with my current hosting but they've been added to my shitlist either way.
Most of these sites simply vanish at some point, with kind of defeats the purpose of using them to simplify their operation as you need to be ready to look for an alternative on short notice and it's a security concern.
OTOH what are the chances that the code changes would be of any use to the author?
It could be that the time it took to discuss and investigate the issue led them to realize that they don't want to release the material, or possibly cannot.
If a program is only GPLed, then it can be put into service with GPL-incompatible changes. In that form, it cannot be redistributed, but since that is not required, there is no problem.
If a program is AGPLed and is put into service with GPL-incompatible changes, then there is a double problem. It must be redistributed, but in its present form it cannot.
It doesn't cure the past violation, but a whole lot of legal importance rest on knowing violation, and continuing after being informed of the violation can, in addition to being evidence that you violation after the notice was intentional, weaken your argument that the violation before was not, so stopping a violation once you are informed of it reduces, though does not entirely eliminate, the legal risk. So, a hard stop on use when you are informed your currente use may be illegal is a perfectly sensible response in most cases even if you are evaluating whether your use was illegal and whether, if it was, you should, in the long term, either stop the use permanently or conform to whatever legal requirements attach to use.
The copyright holder can choose not to pursue the matter if the web site hosters send the source code, but that has nothing to do with the requirements in the actual AGPL.
Where I work, we use a lot of free software to build our product (various licenses) and I am happy to hand over the sources of these libraries to anyone asking. So far, no one has asked, though.
Not quite. From [1]:
> If you commercially distribute binaries not accompanied with source code, the GPL says you must provide a written offer to distribute the source code later.
So if you don’t send the source along with the binaries then you need to make a written offer to your users. However, [2] states that:
> If you make object code available on a network server, you have to provide the Corresponding Source on a network server as well.
So you can’t publish the binaries online but only distribute the source via physical media, for example.
[1] https://www.gnu.org/licenses/gpl-faq.html#WhatDoesWrittenOff...
[2] https://www.gnu.org/licenses/gpl-faq.html#AnonFTPAndSendSour...
It definitely does not suffice to simply send the source code on demand without additional up-front work.
I agree it's a weird reaction to take the site down when pushed, and try to get away with it to begin with, but I also think that them removing the site entirely was a very likely outcome from the beginning, and probably the best move for that company.
If they missed the AGPL requirement from day 1 (likely) and they're developing at the typical level of a small business who misses things like licencing requirements, then their source code is likely not something they can release. It's likely to have hard-coded credentials, and it's even possible that the licence for their translations don't allow them to distribute them elsewhere! They might have conflicting licences.
As unsatisfying as it is, taking the site down is a reasonable step – it mostly addresses the complainant's request by ceasing to violate the AGPL.
The AGPL is pretty much impossibly to comply with. Read the part about "Corresponding Source" and think about just how much Corresponding Source is out there.
E.g.: If this company ran the service on AWS, then the source code for their provisioning scripts for AWS to run the service would be in scope. And those scripts probably pull in internal libraries.
Or not. You may find out after a multi-year court battle that this part of the AGPL is unenforcable. Who knows? It's lose-lose anyway.
It's absolutely not worth it to run AGPL. It's not a weird reaction at all. It's probably the only possible action.
A script that calls a piece of software is not a derived work. Otherwise, all build systems that have ever been used with GCC would have to be GPL-licensed.
Why would your example be different?
If that script relies on that software, it is a derived work. If that script is GPLed, then that software must be GPL-compatible, or else it must be interpreted as falling under the "system library exception".
> all build systems that have ever been used with GCC would have to be GPL-licensed.
The tools required to build GCC on a proprietary OS can be regarded as landing under the system exception.
It would almost certainly be a GPL violation to add a patch to GCC which causes it to require a third-party proprietary program or script in order to build (something not coming from a target system itself).
This is comparable to a provisioning script.
Probably not, but it is "Corresponding source" in the AGPL license.
> Otherwise, all build systems that have ever been used with GCC would have to be GPL-licensed.
No, because:
1. AGPL is not GPL. GPL doesn't have the concept of "corresponding source". AGPL does.
2. Build systems were not built for the purpose of building software X, they were built for building any software. In other words, a completely freestanding source code that can ONLY link with the Linux kernel IS in scope for GPL, as derivative work. E.g. see legal status of out-of-tree binary kernel modules.
Or another comparison: Your LD_PRELOAD library that overloads write() will work with any software, and therefore is not derivative work. Your LD_PRELOAD that is made only for a GPL tool does in fact link with it, and probably (don't know if there's a court case on this yet) does mean it's GPL.
Your script that can only be used to launch an AGPL software thingy IS "corresponding source" and/or (could be both, a court has not tested this) "derivative work". It's certainly not neither, and therefore it falls under AGPL and must be published.
That's the purpose of the AGPL.
It's a political tool for the GNU ideology whose intent it is to prevent the free software from being used for building services that are are also built on proprietary software.
Those situations cannot comply with the license, by design.
The AGPL is not about the user freedom at all. The user has no freedom because the user has no administrative access to alter the service software.
The software can do bad things that are harmful to the users, yet fully comply with the AGPL. The users have to accept that, or not use the service. The FSF wants it that if the users are going to be abused by an online service, at least this should be done with entirely free software, and that's where the AGPL comes in.
The AGPL (and copyright licensing in general) is completely the wrong tool for actually doing something to combating the social problem that people are becoming dependent on SaaS that is out of their control.
Or: even if Facebook were thoroughly AGPLed, running your own copy would be next to useless because only you and three other people would be there.
Not having a copy of the exact software that services are running is not what is gnawing away at your freedom. It's completely the wrong problem to be focusing on.