Hacker News Reader

Comment by Elv13 | Hacker News Reader

As of now we have no plan to support Tox. We initially evaluated this and decided against. Ring is based on open standards such as SIP (rfc3261), TLSv1.2 (rfc5246), SRTP (rfc3711), vCards (rfc6350) and many more.

We are also inter-operable with existing SIP infrastructure such as corporate phone system and classical phone provider with optional SIP accounts. We will continue to be as standard compliant as we can rather than create/support yet another custom communication protocol. What Ring bring to this existing mesh of technologies is the ability to connect people using a decentralized peer-to-peer network.

View on HN

Support for non-obsolete and secure standardized protocols seems to me to be a good thing.

    As of now we have no plan to support Tox. We initially evaluated
    this and decided against.

I'm rather interested in the reasons of this decision.

    Ring is based on open standards (...).

Tox is open as well.

However, I see a difference between the protocols you've given and Tox: Tox is not standardized.

    What Ring bring to this existing mesh of technologies is the
    ability to connect people using a decentralized peer-to-peer network.

This technology for peer-to-peer discovery is also not standardized, I assume.

So, why creating your own protocol instead of using Tox for this?

> This technology for peer-to-peer discovery is also not standardized, I assume.

DHT has been used in the wild for more than a decade now, it is very well understood. SIP over DHT idea has also been studied[1] academically for a long period of time, even if it was never really implemented in commercial products. Finally, our P2P "protocol" is still using the standards mentioned above for cryptographic identity and other negotiation details, so there is very little that isn't fully standard.

[1] Example: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.183... (not used for the current Ring implementation)

    DHT has been used in the wild for more than a decade now,
    it is very well understood.

Tox also uses DHT for contact discovery [1].

Does Ring support communication without having to register any type of account on any third-party server?

[1]: https://jenkins.libtoxcore.so/job/Technical_Report/lastSucce...

No server account are required, it is what DHT is for. However, if you do have a server SIP account, we also support them. This is all about interoperability with existing infrastructure.

I see value in the support of server accounts.

However, the DHT-based account-less p2p part seems to redoing work Tox has done.

Why not use their efforts and experience? They offer their core separately and Ring could be a client for it.

That would duplicate most of our code and we chose not to do so. We currently have an unified architecture that handle both existing sip infrastructure and the distributed one. The distributed part is less or more done, it isn't a big part of the code (about 4% according to a quick rexgex) compared to SIP. We currently use an open network model, but will switch to a closed one based on the cryptographic chain of trust in the coming weeks. That about all there is to it, everything else is SIP/ICE/UPnP/TLS/RTP.

    That would duplicate most of our code

You wouldn't have had to write that code if using Tox were chosen in the first place.

    will switch to a closed one based on the cryptographic chain of trust

This could differentiate the decentralized part of Ring from Tox. Can you give more details on this?

Anyway, I still think you should try to join efforts.

> You wouldn't have had to write that code if using Tox were chosen in the first place.

First of all, it is important to understand that Ring is based on SFLphone. Our team have over 10 years of experience (and code) working with the existing SIP infrastructure. Have an inter-operable SIP based decentralized network is really the whole point of doing what we do.

> Can you give more details on this? (please note that this section talk about features in development, it can still change)

The classical phone network is an open system. Everybody call call everybody, some for emails. This can cause some issues, such as tele-marketing and scam (+ spam, called "SPIT" in the VoIP community). Some other IM networks (MSN messenger, Skype, in fact, most of them) use the opposite. You can only open communication with people who are whitelisted. Traditional SPAM methods are mostly useless on most VoIP media such as voice and images as there is no "ahead of time" processing (everything is RTC). A distributed server-less network also cannot ban accounts, there is none. With IPv6, an IP based blocklist is also impossible.

There is a bunch of ways to fix those issues. One is to build a validation network be asking the peers you currently trust to validate new persons. The downside of this is some serious anonymity issues. You basically tell the world who is calling you. The second option is to use a chain of trust. There is multiple ways this can be implemented. First, it can be done on the DHT itself. You only talk to nodes that are signed by a certificate authority you whitelisted and use TLS revocation and other protocol features to manage a pseudo-clean distributed closed-network from which intruders are banned. The second way is to use the full "global" DHT, but only allow calls signed by a trusted authority. For example, this allow a small startup to sign a bunch of certificates, place them on the DHT and let employees call each other. This network will be about as insulated from unwanted external calls as a private SIP server.

Obviously, those 3 strategies are complementary and can be combined to eliminate SPIT and to a certain extent third party meta-data collection.

I respect your effort to implement an open inter-operable SIP based network.

I only referred to the code that implements the p2p DHT part. That part was duplication of effort.

Now, you are talking about a serious and interesting problem in open systems: spam/spit/...

However, Tox faces the same problem. Their network was attacked at least once from the 4chan community. And they came up with a solution. Their solution doesn't force you to tell anybody your contacts and doesn't need certificate authorities.

How about talking with them about the problem to join efforts?

The Tox people are very open to other people using Tox just as a transportation core.

I'd just be unhappy to have another DHT-based communication platform that didn't talk to the other players first.

In all honesty, the decision was taken by the project sponsor and is final. We built this https://github.com/savoirfairelinux/opendht and it is already used by some other projects distributed outside of Savoir-faire Linux. So far things look well for OpenDHT so I don't think Savoir-faire Linux will change its mind of the decision at this point. I know you think this is a duplicated effort, but it might not be as true as you think. I respect and understand your frustration, but the decision is final.

However, do you have a link about the Tox 4chan DOS? A quick Google search failed to turn interesting results.

    the decision is final

I think here is a misunderstanding. First, if you feel offended, then please note that I didn't mean to. Second, what decision are you talking about?

You stated that the technology used by Ring to combat unwanted communication is still not finalized. I asked you

    How about talking with them about the problem to join efforts?

I'm not asking you to throw away all what you've done in the area of DHT-based p2p communication and to start to use Tox. However, how about talking to them, since they fought the same battle? Maybe they have valuable ideas or experiences. Or you might help them with your ideas.

    However, do you have a link about the Tox 4chan DOS?

I have no interesting link available. I'd suggest to take a look at their nospam ID's or to ask them.

On a more general note, I think defining some decision as final to silence criticism is deeply harmful. You might want to consider that.

Thank you for your information.

I am sorry if my message caused misunderstanding. In no way am I trying to silence criticism. You must understand, we had this discussion last year and choose to go our way willingly. We are aware Tox exist and everything. While you may see this as NiH syndrome (and you may be partially right), we chose to write a base DHT library, of which few exist open source one exist on the market right now, and go from there. We have been working for quite some time on this are are happy with the current state of affair. It is just that it is getting late and this discussion start to feel like "Microsoft should use WebKit". In no way I wish to sound offending or anything. It is simply that in the end, we are way too far down the path we willingly took to start re-architecturing everything. I agree that we share some objective with Tox. I also think that competition is a good thing. While communication network are sometime seen as natural monopolies, but it has been proved over and over again that they are not.

I really appreciate your honesty.

    this discussion start to feel like "Microsoft should use WebKit".

I'm not asking you to use Tox, but to look at what they are doing and to talk to them. Microsoft is for sure looking at what WebKit is doing.

Yes, competition is a good thing. But you should also learn from your competition, right?

Thank you.

Reply on news.ycombinator.com