@be@shauna@luis_in_brief@cwebber@root This is really interesting. Will need to check if it would stop us dual-licensing (which is the only reason for the CLA), but there is ZERO intention to switch license to non-OSI... which obviously Apache allowed too.
@shauna@luis_in_brief@cwebber@root I'd encourage them to add a clause to the CLA like Signal and Slint (https://github.com/slint-ui/slint/discussions/244) that obliges them to continue publishing the software with an OSI approved license, so that they can't just decide to make it proprietary one day (which they could have before with the Apache license).
@luis_in_brief@cwebber@root (not a lawyer or license expert) Doesn't switching to AGPL make it *harder* to then switch to a closed license in the future?
@be@element@shauna@luis_in_brief@cwebber@root KDE Free Qt Foundation has done something very similar for over 20 years now, if Element needs someone to talk to about the particulars of it, I would recommend reaching out to them.
Granted I'm the one who is grumpy with the change and repeatedly saying, "we'd prefer no CLA," and not the one who decided to fork ... but nonetheless, I'll send this over to my colleagues at Element.
This would at least soften the blow. Thanks for sharing!
@josh@shauna@luis_in_brief@cwebber@root Note that this type of licensing arrangement isn't new; it's similar to the agreement between The Qt Company and KDE Free Qt Foundation.
@josh@shauna@luis_in_brief@cwebber@root Considering that there already is a Matrix Foundation set up, involving the foundation in some sort of legally binding agreement like the KDE Free Qt Foundation may be worth considering. I'd suggest getting in touch with the KDE Free Qt Foundation people if that's considered seriously. https://kde.org/community/whatiskde/kdefreeqtfoundation/
@josh@shauna@luis_in_brief@cwebber@root If this is communicated well, it could be better than "softening the blow". With a noncopyleft license (as was the case before with the Apache license), contributors and users have no guarantee that the maintainers won't turn the project proprietary in the future. A copyleft license with a CLA *can* provide that, if it's carefully considered.
@be@shauna@luis_in_brief@cwebber@root Yeah, excellent point. More than softening the blow, it would be an actual, binding commitment to keeping it open source.
@katco@be@josh@shauna@cwebber@root Much depends on the quality and maintenance story of the other protocol implementations. If there’s a lot of money flowing to one implementation, or contracts that specify the tool rather than the spec (which their post suggests may be the case?) it becomes very easy to change/ignore/manipulate a protocol spec.
Is this situation made less risky in that we're talking about reference servers that implement a protocol, and that there are other servers that do the same, e.g. https://conduit.rs/?
It seems like this is a good example of why protocols are good.
@katco Absolutely. If we didn't have an open source protocol with open governance, and other open source server implementations, this would be catastrophic.
@luis_in_brief Very true, and there's tragic prior art for that.
This heightens the need to cultivate a vibrant, diverse ecosystem. I am cautiously optimistic, but ultimately only time will tell whether today represents an inflection point for the better or for the worse.
@josh is matrix.org, the instance of Matrix, owned and governed by the Matrix foundation or Element? Because that seems like significant leverage. If Element attempted to fork/ignore the spec, I think it'd be difficult to fork away from that instance at this point.
@katco Right. This is one of the reasons I think there's good reason to remain optimistic.
A lot of my time in leadership at OSI was punctuated by for-profits extracting from, and enclosing, the commons. I do think there are meaningful differences in how this is playing out in the Matrix ecosystem 🤞🏻
@josh right. This helps me take this announcement in good faith (unless there's some deep entanglement to Element in operating that instance I don't know about, e.g. there's a CVE disclosed, no one at Matrix to fix it, and Element has forked the spec).