… oh, yeah, that makes sense. It’s the same intent as sending a patch upstream, with that framing this is an irrelevant issue, sorry I brought it up!
1 Like
I think this a very relevant topic to consider, even if it turns out it has no practical consequenes. Thanks for bringing it up!
Not necessarily. But let’s say you are a small coding shop doing consulting which includes some service development & configuration for customers:
- With Nix:
- Development and deployment of these internal systems comes with no strings attached beyond a copyright notice.
- With Aux:
- If you declare the system in nix code using Aux nix code then that whole config is also copyleft and thus must be made publicly available.
- If you declare a shim which manages all the systemd stuff by launching non-nix authored systemd config then only the shim is copyleft and thus has to be made publicly available.
- Objectively, for customer privacy and flexibility you are delivering a subpar service if you do not suggest a more flexibly licensed product. Nix is the vastly superior choice. But more likely the customer does not have that insight and will in stead go "No that does not work for us - give us something Ubuntu or Windows based’'.
Right now this is the reality the community is voting for. Dismissing the business case out of hand as “corpo can buzz off” is very easy and highly popular - no surprises there 
Only point of slight critique being that this stance ought to have been very clearly communicated in the key project principles rather than snuck in after the fact in a poll.
8 Likes
@isabel / @jakehamilton Incidentally, given that the poll closed yesterday, could we please have that clear communication explicited on the site now? As things are, people are still going in blind with the fair assumption of MIT - having been not told anything to the contrary.
2 Likes
Do you have an idea around the ways you would expect this to be communicated?
I would suggest adding it in as a point in the Values section of the site. It is prominently displayed at the top of the page, a key differentiator, and copyleft licensing is very much a choice solidly anchored in values & beliefs
Ah and ofc. to cover people who already signed up, a pinned post here in Announcements would probably suffice.
I don’t get why such a significant decision, which likely affects project values is being made without a ranked choice poll. Obviously, there are people who strongly prefer the copy-left license, and there are also those who strongly oppose it, with others falling somewhere in between.
2 Likes
When I voted in this poll, it barely even occurred to me that it would be binding.
Ideally, I’d like more time to think about it.
I did vote, and I didn’t complain at the time, so I have only myself to blame (seriously - not being pasag). But my current opinion, FWIW, is that we should aim to choose our licence later. (In the meantime, we’d need to use a licence that made it easy to relicence, and maybe we’d need to keep track of copyright holders just in case.)
4 Likes
Another reason why this is a difficult decision even for people (including me) who are strongly in favour of copyleft is that it’s not obvious, even for people with a lot of experience in software, what’s compatible with GPLv3 and what isn’t.
Clear evidence of that is the disagreement even among experts about the extent of incompatibility between ZFS and the Linux kernel. I’m pretty sure even expert lawyers are still disagreeing about that. Probably they shouldn’t be, but they are.
2 Likes
Agreed - there’s still parts of our architecture that we need to flesh out and consider. Far too early to be talking final choice of license, IMO. Things are still moving fast, and it’s very difficult to determine what a suitable license looks like without things settling down.
9 Likes
So just to be clear before I go to bed, because I’m in a stupid time zone: my proposal is that we start with whichever licence (within reason) we determine to be easiest to change later.
1 Like
To be honest the longer this poll went on the more I wanted to close it early since it held way too many points of contention. We should probably just licence on a per repo basis anyways and if we do vote again it’s worth looking at loomio and the voting integrity post
9 Likes
That approach is quite understandable, but without being explicited front & center I also don’t think it is quite fair to anyone considering investing time & effort in the project.
On top of that, “key parts of the project may at a later date turn copyleft” is essentially worse than “the intent is for this category of repositories to be copyleft and for this other category to retain the pre-fork MIT licensing”.
3 Likes
Okay, so can the current code contributors agree to hash this out (general licensing approach) at a later point and we end this thread for now? To be clear a side effect of decide-later, because no-license and copyleft AFAIK do not allow “decide later”, would mean temporarily using a permissive license.
As some kind of promise that “temporary” actually means temporary and not “long enough until people forget about it” it would be good to set a condition/trigger for when to re-open the license discussion.
(And not to exclude your point @AngryAnt, I think that would be part of the decided-later discussion)
2 Likes
That’s completely understandable. I would obviously excuse myself and still highly recommend that some indication of the current semi-radioactive state of project licensing is alluded to where other potentially interested contributors can clearly see it coming in and avoid feeling bait & switched on later.
1 Like
[…] then that whole config is also copyleft and thus must be made publicly available.
I don’t think that’s how copyleft works. The way I understand it is that if you distribute a work that is a derivative of or a part of some other copyleft work, then you must keep the same license. There is no requirement to make it publicly available. At least that’s the case with GPL (as I understand it). For example, a company that uses a modified GPL licensed program internally is not required to make these modifications available, either to the general public or to its employees: Frequently Asked Questions about the GNU Licenses - GNU Project - Free Software Foundation
Also: Frequently Asked Questions about the GNU Licenses - GNU Project - Free Software Foundation and Frequently Asked Questions about the GNU Licenses - GNU Project - Free Software Foundation
Personally, I am for copyleft generally, but I am unsure if it makes sense for packages. For programs, CLI utilities or whatnot, definitely copyleft. But for packages I think it will only cause confusion and be harder to comply with for no benefit to the Auxolotl project. Even for non-commercial, non-proprietary uses such as someone sharing their config publicly, they would have to license it under a license which is more complex to understand and harder to comply with than a permissive license. And well… the understanding of licenses among software developers is not great anyway, many don’t even care to comply with the attribution requirement of the MIT license 
2 Likes
My understanding differs, but more critically the likelihood of the understanding of business partners differing sufficiently to determine that they want nothing to do with the risk potential is off the charts. Which is how you suddenly find yourself an Ubuntu/Windows shop after having gone in with the best of intentions 
3 Likes
I agree with what you wrote, but wanted to point out that I also don’t believe @AngryAnt is quite wrong (which just goes to show how on point "more critically the likelihood of the understanding of business partners differing sufficiently to determine that they want nothing to do with the risk potential is off the charts. " is). But let me explain:
I believe the crucial thing about GPL is that it’s a ‘viral license’ (also called ‘strong copyleft’). That is, anything that links to it either never is published or has to be published under GPL. The problem: linking is a very under-specified term, with a very blurry definition. And depending on who you would ask, pulling in aux packages is considered linking; thus forcing GPL on anything that uses it.
(The following is my personal opinion on this thing, but I believe something like this has already been voiced on this thread: While I’m certainly in support of copyleft, I can totally see how that would both discourage other projects to build on top of auxolotl and people investing their time in auxolotl - since now usage of the work they put into aux forces them to use the GPL for anything they want to use aux with.)
8 Likes
that is kind of what I meant with:
3 Likes
And the courtroom, potentially a room of tech-illerate people, is the only “who you could ask” with real consequences. Thus feeding the perception of risky/unpredictable outcomes.
2 Likes