09

9 - Discovering the world 🌐

🐥 Wow no you have done really well! We are really close to entering the final round of the workshop, where we are going to jump to put all that we've learned in practice in our chat app.

But first, we are going to study the last feature of Dat that we have skipped until now. It is about the old question; who do we find other peers?

We have looked mostly at using hyperdb, but the discovery in Dat is resolved by hypercore.

We can devide the discovery into two parts:

• At the network level, let's say the "physical" form, is how we create the swarm (this is the group of peers).
• Then we can also think of the discovery at the level of data, or to say, to search for information between the peers. To resolve this, Dat uses data structures like tries and merkle trees which let us get data in a highly efficient way. remember that Dat is designed to share big volumes of data, minimizing the quantity of data that is moved between peers.

But we are going to focus on how we create the swarm 🐝. To achieve this, hypercore uses the module discovery-swarm which in turn uses discovery-channel.

Dat has three strategies to connect the swarm of peers:

1. Find the peers in the local network. discovery-channel uses a module called 'dns-discovery' that has the ability to make queries in the local network using multicast dns (MDNS).
2. If we don't find eachother locally, we go to the internet. The next phase is to search using dns, also with dns-discovery.
3. If these methods fail, we are going to look in a DHT or distributed hashtable. For this, discovery-channel uses bittorrent-dht.

The first two methods are fast. In the 2nd or 3rd method a point appears to have in mind, they are centralized in some way. What does that mean? Maybe we aren't in a P2P workshop?

bootstrap nodes

Most of the decentralized solutions use bootstrap nodes. The bootstrap nodes function like known peers beforehand, that we use to start to know others and like this form our own swarm.

In other words, we need to know someone to enter the network.

In fact, Dat maintains it's bootstrap nodes in a separate module: dat-swarm-defaults.

What happens on the web?

We have seen specific concepts of Dat, that function wonders in the commandline. But in this workshop we want to create a web application.

When we work with P2P applications, we break the concept of client <-- server that we are accustomed to. P2P applications are useful in other architectures, if we can think of each participant as a client and server where the communication flows between pairs C/S <--> C/S <--> C/S.

To achieve this effect in the web we use webrtc. webrtc is a technology that allows us, among other things, communicate directly between peers.

It's fit to mention that webrtc also is used by bootstrap nodes. An interesting module to create a bootstrap node is signalhub.

Exercises

Introduction: Let's start by connecting the parts 🔌. In this exercise, we will work with two files. One is chat.js and the other is index.js, where we were working with Saga. chat.js will use Saga and is where we will create the swarm.

1. In the chat.js file we are going to create a swarm. We will use a module by GEUT: discovery-swarm-webrtc to do it 🆒. There are two important parameters that we have account for in the constructor:

• id: Saga exposes the hyperdb instance, and from there we can obtain the expone local feed and from this feed the key (in hexadecimal). This is what we have to pass to it, a unique identifer of our feed.
• stream: a stream to replicate with other peers. Luckly Saga exposes a replicate method. 😉

2. Then we are going to create a signalhub instance. The constructor receives two parameters, one of them we already have, signalUrls; the other, which we have to complete, is used like a namespace, here we will use the discoveryKey (in hex too) in hyperdb.
3. We now only lack to connect to Saga. In point (3), we have the swarm ready. swarm emits events, one of the most important parts is connection. This indicates when we have a new connected peer 💥 When this occurs we have to tell Saga. We will add here a new method (now we are going to implement it 😉): connect(peer) which receives a peer as a parameter.
4. On the Saga side, we haven't implemented the new connect method. Let's go to the index.js file. connect receives a peer as a parameter. peer is an object with a property that interests us, remoteUserData. From there we will get the data and the key (PK) of our new peer. Remember the earlier exercise? We will authorize it using this info.

Tips

• GEUT will rescue you. We will use a module from GEUT: discovery-swarm-webrtc to create tehs warm. This module maintains an API very similar to discovery-swarm.

Test

$ npm test ./09

test.js

Solution

solution.js

Extra 📚

• We recommend this article by Ranger Mauve

---

Woow this was a lot, but now you're already close to putting everything in action 🚀