paratii-live.md

Objective

---

Get paratii to play livepeer streams in a seamless way.

What's the current structure of paratii player?

Paratii mainly compromises of 3 core parts:

• paratii-js : This has all the logic required to interact with paratii-contracts on the blockchain, and IPFS.

• paratii-portal: the main React app where the player lives

• paratii-mediaplayer : a clappr based player that has various custom loaders to allow Clappr to pull content from IPFS and/or XHR.

What needs to change?

Mainly, the paratii-mediaplayer loader needs to be able to integrate with livepeer and resolve livepeer channels. the mediaplayer has a loader that is based on hlsjs-ipfs-loader it's currently able to:

• get playlists via IPFS or normal XHR loading (through the IPFS gateway)
• get ts segments and feed them into hls.js.

so this loader need to be extended to be able to get both playlists & segments from the livepeer network smoothly, it'll use livepeerJs various libraries to achieve that and it'll follow the same logic used in @livepeer/player

differeniating a livefeed from a normal video on paratii would probably require a simple URL parser so paratii would know whether to use livepeer or IPFS to get the required video/stream

example normal video: https://lab.paratii.video/play/cugXqidPKCR9

livepeer stream: https://lab.paratii.video/lp/<eth_address>

UI changes to paratii

there needs to be a live mode (red live LED), and no start/end timestamps without breaking the interface.

I'm looking into how to make the changes in an efficient manner, I'll keep you updated regarding this.

Good start here @ya7ya.

Can you elaborate on your vision about the live streaming workflow? If someone creates a live stream in Livepeer as a broadcaster, how does she get the Paratti player to start playing that stream? (For example, does the Paratti player simply take a url param? Should this url be uploaded somewhere so the Paratti player can get it?)

Note, currently a new job is created when a live stream is created in Livepeer. But it has proven to be a bit of a bottleneck and we are thinking about taking it away (sorry I forgot to mention this on the call - had to think this through a little more). So I think for now, we can expect the Livepeer broadcaster (operator of the Livepeer node) is in charge of publishing the viewing URL.

Can you elaborate on your vision about the live streaming workflow? If someone creates a live stream in Livepeer as a broadcaster, how does she get the Paratti player to start playing that stream? (For example, does the Paratti player simply take a url param? Should this url be uploaded somewhere so the Paratti player can get it?)

hey @ericxtang

Yeah, sure, My current understanding is the broadcaster is in charge of sharing his eth address, and I can use the existing tools (@livepeer/sdk and graphql ) to pull the latest job and get that manifest. so it's going to mimic @livepeer/player

But this is my initial assumption (which i'm testing right now) so i'm open to ideas here 👍

The job onchain only provides the stream ID, but not the entire URL. So the URL has to be provided somehow. In the current player it's hard-coded. But that's bad.