Conversation
🚀 Build Preview on IPFS ready
|
| ## Extension To Delegated Routing | ||
|
|
||
| Making content discoverable can be achieved by offering a delegated routing endpoint | ||
| ([[http-routing-v1]]). An Edge IPFS server responds to routing requests for CIDs |
There was a problem hiding this comment.
I would be more specific here about the endpoint only implementing section 4 of the spec, i.e., the Content Routing API.
| that it knows about with a peer routing response the `Protocols` array of which | ||
| must have a `rasl` entry. | ||
|
|
||
| Correspondingly, there must then be a `rasl` field in the peer response object, |
There was a problem hiding this comment.
| Correspondingly, there must then be a `rasl` field in the peer response object, | |
| Correspondingly, there must then be a `rasl` field in the Peer Schema object, |
One thing that's slightly awkward is that the peer-schema requires an ID:
ID: the Peer ID as Multihash in Base58btc or CIDv1 with libp2p-key codec.
So the question is whether we want to require every RASL server to also have a key, or to work on adjusting the spec. I'm sure the latter will get push back, given that this assumption is baked in to a bunch of implementations — but still worth having the discussion, given that with HTTP, you are most likely relying on TLS and certificate authorities for transport encryption, rather than key based authentication, transport encryption.
Just playing around with this idea.