I’m looking to try self-hosting an XMPP server for my family to use as a secure communication platform. I realize that end2end encryption with XMPP doesn’t seem as strong as something like Matrix but my self-hosted Matrix server has been very unreliable.
I’m looking for recommendations and resources. I’v considered running Prosody and Openfire but both of them look like a pretty involved installation process with plenty of room to fuck up. Does anyone know of something similar to matrix-ansible-install for an XMPP server? Should I be looking at something besides Prosody or Openfire?
Please, no YouTube tutorials. I prefer written instructions.
moving to AGPL is “less permissive” than the apache license they currently use? no. this is just plain wrong. the supposedly less-permissive part is the CLA they are asking contributors to sign so they can dual-license the software in some situations, but the CLA isn’t even written yet and they are actively listening to feedback from the community to determine how best to shape this license agreement so that all parties are happy.
this is FUD.
Oh yes, but it is.
Apache > You can do what you like with the software, as long as you include the required notices. AGPL > You may do what you like BUT you need to track changes and provide the source code - this essentially kills any company trying to build a product around it.
We dont want a bunch of proprietary extensions to an open communications standard, do we? This is something positive.
That said, I dont have much hope for matrix. Implemented in python with the initial goal of “bridging every chat platform in existence” is just bound to be a disaster.
Maintaining anything beyond a couple of hundred lines in python becomes tedious imo.
The rewrite in go has been spoken about since like 2018, and matrix.org still runs synapse iirc. Synapse should have been trashed immediately after MVP demonstration.
Theres also conduit, but to be honest, i feel like the lesson here is to avoid feature creep. Safe, fast and distributed dm text chat should have been the target functionality, with a lean, mean codebase.
Thanks for coming to my ted talk
Just like Chia, who on their right minds would do those things in Python.