[split]P2P chat, tor etc

[Lyberta]

Moderator notice: split from here: https://forum.freegamedev.net/viewtopic ... 81&p=86932

Well, I'm also using Tox and it has voice and video but it is p2p.

[fluffrabbit]

I tried TRIfA, the only Tox client currently on F-Droid. It made my phone heat up, the echo service can't do more than 7 FPS, it seems to be stuck in vertical mode, and when I exit the app it crashes rather than closing gracefully. I also don't think the long hexadecimal IDs are so great. They could make them shorter and more memorable by using base32 and less entropy. It's pretty janky overall, but I guess it works.

Jami has a much nicer interface, though it's not pure SIP; some other shit seems to be added onto the protocol. It's also changed names a couple times now. Probably less secure than Tox, but it doesn't turn your phone into a toaster and the developers respect material design. Not a pure protocol though, so technically it's disqualified. I'm merely using it as an example of how to make a communication app more user-friendly.

[fluffrabbit]

Here's an idea: Store usernames in a distributed hash table.

Much like generating an onion URL, the first few characters of your chosen username go at the beginning of the hash, and the protocol would specify that those characters must match the associated username in the DHT or it's invalid. The computational cost of generating vanity hashes would deter spammers from eating up all the desirable usernames. If you don't care what your username is, one can be generated based on the first few characters of a random hash so that it matches and there is very little computational cost.

With this system, you could address people by username. It checks the DHT for the username, there can be only one, and the entropy of usernames could be very low to start with and grow with the size of the network, like with the transition from IPv4 to IPv6. Then you would only have to type something like 5 characters rather than 20.

[fluffrabbit]

More off-topic: A serverless SIP/XMPP type infrastructure would be ideal. Read about DHTs, I2P, and Freenet. The tech is in place, it's just a matter of somebody making an app.

[Julius]

More off-topic: A serverless SIP/XMPP type infrastructure would be ideal. Read about DHTs, I2P, and Freenet. The tech is in place, it's just a matter of somebody making an app.

Isn't that what Jami (formerly called Ring) does: https://jami.net/ ?

[fluffrabbit]

I thought so, but

a) Jami, formerly Ring, formerly SFLPhone, is not a protocol and stands alone as an implementation. May not be the best for security and interop.
b) There must be some kind of Tor-like server layer, else my carrier would block it. Not pure P2P.

[Lyberta]

I just installed Jami and I don't see a SOCKS proxy option to run it via Tor, wth?

[fluffrabbit]

Tor isn't the only anonymity solution in the world. Personally, I think it's over-hyped and constructed in a more dangerous fashion than alternatives. It may be worthwhile to read Jami's source code, but in general, I don't think we need Tor.

[Lyberta]

Tor isn't the only anonymity solution in the world. Personally, I think it's over-hyped and constructed in a more dangerous fashion than alternatives. It may be worthwhile to read Jami's source code, but in general, I don't think we need Tor.

Well except Tor I know VPN and I2P. But getting anonymous VPN is extremely hard because you have to switch exit nodes manually and I2P is slow. With Tor I can watch 1080p@60fps videos on YouTube, maybe not always but it is still fast enough. I2P seems to be much slower.

EDIT: Wait, I installed i2pd, set Firefox proxy to 127.0.0.1:4444 and I can't connect to clearnet websites, wth????

EDIT2: Oh, I2P nodes don't allow clearnet access by default which means I2P is 99% useless.

[fluffrabbit]

99% of the time when people say "clearnet" they are referring to the web, which we're trying to move beyond. You have to approach this from a forward-looking perspective. A completely decentralized Internet means moving away from the client-server model entirely. That's how the best of communication software already works, with document retrieval systems close behind and e-commerce soon to follow.

I don't appreciate Julius splitting every thread. In this particular case, since he wants to DevOp it up and explore cutting-edge tech, this type of communication software might be of interest for FGD's chat system. Not that this site is lacking in interested parties or anything. Maybe I should look to these systems to drive traffic bring people to my own forum.

[Julius]

I don't appreciate Julius splitting every thread. In this particular case, since he wants to DevOp it up and explore cutting-edge tech, this type of communication software might be of interest for FGD's chat system. Not that this site is lacking in interested parties or anything. Maybe I should look to these systems to drive traffic bring people to my own forum.

this was clearly getting totally off topic in the original topic...

If you want to discuss decentralized systems like this, I don't mind at all, but I really don't see the relevance to the question of having a chat system for FGD. Don't get me wrong, it is cool tech, but FGD is more or less a public space... and these primarily non-discoverable and e2e encrypted services (while nice from a privacy perspective) are the opposite of a public space. If it wasn't for a lack of features and convenient moderation options I would definitely prefer IRC as the public chat as anyone can easily join and participate.

[fluffrabbit]

Why do you see secure chat and public spaces as mutually exclusive? The best public "chats" are done anonymously, whichever medium they assume. With many of these systems, granted, you have an ID which separates your messages from those of others, but that's no different from IRC, Mattermost, or any other chat system in that regard. Anyways, the only thing that Mattermost does that IRC don't seems to be the newsfeed, but I'm not fond of that anyhow.

[Julius]

Maybe I haven't used these systems sufficiently, but those that I tried only had encrypted 1 on 1 chats and closed group chats that would not scale well beyond maybe 10 users or so. There is also a discoverability problem in general. I think these systems have their use-cases but public spaces isn't it.

The news feed isn't even a core feature of Mattermost, but just an addon with an RSS feed bot. That could be easily done in IRC as well. I don't even know where to start explaining how much more really useful features for group collaboration Mattermost has compared to IRC... did you even look at it? It's probably closer in functionality to a modern forum system than a pure IRC chat. But we are getting off-topic in an off-topic tread

Edit: RTFM https://docs.mattermost.com/guides/user.html

[fluffrabbit]

While discoverability is an issue, I would call that a design flaw that has yet to be addressed as opposed to a diametric antithesis of secure chat. I can't speak for that scalability of 10 users you mention; what would the limiting factor be? Also, how is any of this different from Mumble, which is hosted here?

As for Mattermost, I have not bothered with it beyond the shitty Android app, and from what I've seen, it sucks.

EDIT: Okay, I'll RTFM.

EDIT 2: Okay, I still don't get it. It seems like the kind of thing that a controlling boss would set up to organize a team of engineers. That's great if you want to get stuff done, but what the hell is anybody getting done on FGD? This is a place for lazy open source gamers and floundering game developers. There is no need for (((enterprise collaboration solutions))) but there is need for lulz.

[Julius]

AFAIK they are not scalable to public group chats due to two issues, 1. a general problem with too many p2p connections versus a central node to relay messages 2. an encryption issue with having to have 1 on 1 keys for every user and device. (2. also seems to effect some other e2ee but centralized messenger, but in my experience it is less noticeable).

The Android App of Mattermost, especially the "classic" one found on Fdroid is sadly a weak point. It is really all in the web-client for now. The new react native mobile app is slowly catching up though.

edit: I think the future lies in federated messengers with nomadic identities, not fully decentralized p2p ones.

[fluffrabbit]

You could do massively scalable text chat using a system similar to ZeroNet. For A/V chat of course you would be limited to a number of users, and you wouldn't want to hear that many voices at once anyways.

[Julius]

Sure, A/V chat is even more problematic. But text chat needs to be more or less real-time (I am already getting annoyed by the laggy [Matrix] chats all the time)... I doubt it can be done without federated relay nodes or centralized servers. But to be honest, I have very little expertise in this topic, so take my opinion with the needed skepticism

[fluffrabbit]

With a 1-on-1 connection through several hops, expect 500 ms latency at worst. That's pretty good IMHO. ZeroNet is more like a blockchain so I suppose the latency would be longer, but nobody has problems with high latency in email or forums.

[Julius]

EDIT 2: Okay, I still don't get it. It seems like the kind of thing that a controlling boss would set up to organize a team of engineers. That's great if you want to get stuff done, but what the hell is anybody getting done on FGD? This is a place for lazy open source gamers and floundering game developers. There is no need for (((enterprise collaboration solutions))) but there is need for lulz.

heh , you might be right about that impression... so what can we do to get more lulz into it?

[fluffrabbit]

EDIT 2: Okay, I still don't get it. It seems like the kind of thing that a controlling boss would set up to organize a team of engineers. That's great if you want to get stuff done, but what the hell is anybody getting done on FGD? This is a place for lazy open source gamers and floundering game developers. There is no need for (((enterprise collaboration solutions))) but there is need for lulz.

heh , you might be right about that impression... so what can we do to get more lulz into it?

Okay, I guess since you're fixed on Mattermost, I might as well throw some ideas out to that end, though I don't know how feasible these things are.

* Allow posting under the name "Anonymous" without registration
* More memes in the standard emoticon set, especially GIFs with lots of motion
* Reverse the direction of scrolling such that newer posts appear at the top rather than the bottom
* Any post to a thread bumps it

EDIT: Since this is a FLOSS site, meme licensing might be an issue. Here is a public domain meme for starters. There are probably more on OGA. Why do I feel weird about this? I remember when the word "meme" used to imply something more organic, but now it's something one has to work for.