Federated social app: Discord-style servers and chat with ActivityPub.
https://github.com/y4my4my4m/harmony
I don’t remember hearing of it till now, yet it claims a lot of desired features. Github page shows only one contributor, one fork, 44 stars; lot of commits in the last week, oldest 3 years ago; GNU Affero General Public Licence v3.0.
Found via https://chat.btfree.org/@ozoned/3850a821-fc6d-431f-b235-feed68605f6e
Sweet. I’ve been looking for a federated alternative for my upcoming radio station project: [email protected]
the “instant messaging” line at the top is really funny in this context
If the author’s profile is any indication, it’s likely Harmony was written using AI:
Oh, they’re even working on KDE plugins! 🙄
It was. And they’ve also erased its generated disclosures.
@[email protected] who cares?
Most people? Certainly the 43 people who upvoted me.
Claude and cursor are mentioned in the
.gitignore.Looking to understand: what is the significance of this?
It means the project may be vibe coded (LLM generated with little to no verification of what the output is); meaning it’s a lot more likely to contain security vulnerabilities and privacy concerns.
And that’s ignoring the ethical concerns of using AI in addition to the lifespan of the project (vibe coded software often isn’t maintained very long).
Thanks for the explanation!
It’s wild that just the thumbnail alone can tell you it’s vibe-coded.
Not just that but the dev couldn’t even be bothered to write their own bio, what are the chances they took the time to look over a single line of code or test anything at all?
Ok but, Matrix? Or even XMPP?
deleted by creator
NOOOOOOOOOO WE MUST ALWAYS REPLACE $Product WITH “$Product BUT IT’S FEDERATED WITH MASTODON”!!!111
In my opinion, xmpp kinda fits perfectly? It was used in multiple platforms before (google, facebook, Gaia online, and gosh the list goes on, every platform 2005-2013???)
Why wouldn’t it be a decent solution for the fediverse software to pair with it, just like other networks did before?? Fediverse accounts already have the schema (user)@(domain.tld)
I understand Matrix needs more storage than xmpp, but I think either one could work.
fluxer.app 👍
Unfortunately, the interesting parts like group chats are not standardized, so not compatible to other Fediverse platforms.
And I’m generally a bit wary of AI written projects. It’s just hard to contribute and get a community going. And a lot of those kinds of projects show some crazy initial growth. And then they might get abandoned once the developer decides to start a new hobby or stop paying for the necessary tokens. Not sure if this is the case here, though.
It’d be nice to have something like this. Ideally an interoperable version.
It’s not vibe-coded and the developer published several FEPs to help others implement same features (e.g. FEP-bebd: Follow Invites)
@[email protected] i follow both and hope Mony will federate with Shoot. Don’t like AI? You can use shoot
@doesnm @hendrik @fediverse I am not that much against AI but 99.9% of slopware is garbage, and a developer needs to do something extraordinary to prove that their thing is worth my attention.
Nice, thanks. Though I’m afraid what I/we need, is a FEP for group chats and end-to-end-encryption. (And some kind of reference implementation.)
Shoot uses a variant of FEP-1b12 for group chats, it is the same protocol on which Lemmy communities are built.
However, chats typically require a more fine-grained permission system than Lemmy currently offers. This is what I am working on: FEP-5219: Groups and permissions
>end-to-end-encryption
Not a thing yet.
I had some time to think about this. And from what I gather, the missing encryption is probably one of the major showstoppers here.
For one people expect chat to come with some minimum standards. I mean proper e2ee and several other features have been part of practically every other chat solution out there, for quite some time now. And we need it to supply that amount of confidence to enable people to talk to their family, friends, colleagues… We’d need it to eat our own dogfood and discuss internal stuff within mod teams, etc…
And semi-public unencrypted chat messages are kind of a liability to admins as well. We expect people to do nefarious stuff once they think they’re in some sort of closed group. And that stuff can’t federate uncontrollably and we also can’t have it sit on our servers. And that means either encryption as well (so the admins don’t have to deal with it). Or a massive amount of effort and potential legal issues.
I’ve answered about AI involvement in the development of Harmony here https://codeberg.org/fediverse/fediparty/pulls/327#issuecomment-27235791
As for FEPs, I do plan on publishing AP extension proposals, but I’m making sure it federates nicely with other Harmony instances before having everyone implement their own versions.
That way I can say it’s not theorized but functional and demonstrably so.Definitely want to publish the drafts I have, soon, though.
Yeah, that’d be awesome. And a lot of work to get it right, communicate with people etc. If I remember correctly, there’s been efforts for years to get group chats standardized with proper end-to-end-encryption etc… And I’m not aware of any final result.
Edit: I probably didn’t remember correctly… Wasn’t group chats in itself pretty straightforward. But issue is people want some encryption to make it properly useful… And we never finalized any draft on how to handle all the keys and the complexity that comes with e2ee?
Yeah the long term upkeep tends to be the biggest problem.
But then again, normal projects have that too. Seen plenty of small projects get worked on like crazy for a few months and the author burns themselves out.
Ai like with most things just speed up the process. Lol
Fluxer has federation on their roadmap;, I’ll be waiting for their implementation.
Matrix already has federation
do they aim for voice and videochat?
They already have that too, screen sharing included. Most clients work with it just fine.
Please correct me but, isn’t matrix.org, the worse option, because of all the traffic? Calls didn’t work reliably in said server, but I got to do a few streams with friends and it was working fine with not much latency, but we were both in another homeserver.
I’d like to try gaming tho.
Is it low latency, suitable for gaming?
no, unless you have very low standards for “suitable for gaming”
Yeah no I want it to feel like having a conversation in my couch, kinda latency. 😅 Or virtually indistinguishable from that. Fluxer so far fits the shoe. 😁
i hope it isnt 100% ai generated
I don’t know about 100%, but I get the feeling a lot of it is vibe coded.
You can check the code frequency and contributors (there is only one, the maintainer).
its too frequent sounds like ai
Oh that’s me! :D Yes. @[email protected] made me aware of it. It’s brilliant. I’ve been loving it for the last day and I’m blown away. There’s no AI declaration, but this is a mountain of work for one person has done. So I honestly have my questions.
However, it honestly doesn’t matter to me. If it’s free code created from an open weight model that will exist without this project, then we need to be questioning how we stand on this.
Does LLM generated code matter when you get to self host it, it’s free software. And I’m honestly at the point where, I say use their weapon against Big Tech. This isn’t replacing anyone. This wouldn’t exist without it. It’s had commits over the last few years. I think it’ll depend on maintenance, but this can always be forked and should be if maintenance doesn’t work. Because I think this is incredible. The least we can do is learn from it.
I don’t think you quite understand why AI is so bad and why local models are no different. https://whynoai.info/
Like 70% of the points on that site are… Disingenuous at best. It’s not wrong over all at all. But I do wish people would stop throwing in a bunch of bullshit along with the good points.
I hate ai on many moral standards and see it as highly problematic, but also see it as a fun toy.
It makes it hard to have any honest conversations about the problems around ai when everyone is basically a religious zealot spouting as much bullshit as the ai tech bros are.
Is ai a problem? Yes. There’s no honest way to disagree with that. But the real conversation is far more nuanced than that.
I stopped reading at “fun toy”.
There’s always some shill willing to go “yeah this steals people’s work, mangles it and spews bullshit back at them, all while destroying the environment, to make a few dudes richer. It’s nuanced”. 🙄
You just described capitalism. I think it’s strange that so many people are pinning this flaw on AI. I guess AI just made it all start happening fast enough, and to the right people, that y’all noticed?
Anyway sure, let’s tear it down. I can’t be down with you if you just want to get back to last decade and before, when the “stealing mangling bullshit polluting machine” was only stealing people’s ideas and destroying environments outside your own though.
Or how communism operates, perhaps.
There’s using AI assistance for clean coding and following all the other guidelines of good software development otherwise.
And there’s giving Claude Code a simple prompt to generate a ready-to-run application and uploading the result to GitHub as-is because you want to be a coder, but you don’t know jack shit about coding or software development or licenses.
https://mony.world/ exists, hosted by us, if you want to play with it. (I wasn’t involved in the setup, so can’t answer much about it)
As someone not fond of Discord, it seems pretty decent and pretty Discord-like.
