Open jonnycrunch opened 5 years ago
@mikeal @vmx what's your take on URIs for IPLD nodes? I think it's important to have one defined even if we don't try to register with ietf (there's clearly interest and I don't see any downside to recommending a format now). Maybe dweb:/ipld/<cid>
to mirror dweb:/ipfs/
?
Really, I think we need both.
ipld://Cid/path
. This is technically incorrect as the "Cid" isn't an "authority". However, we already have to use this format for ipfs://
and ipns://
as browsers use the authority to determine the origin.dweb:/ipld/...
. I actually prefer this one because it's (a) "more correct" and (b) preserves the path form.Note: In our applications, we still plan on using paths (for now, at least). We use paths for everything because they're composable (although, IMO, we need to write an internal standard/registry for path namespaces).
I can live with dweb:/ipld/ipld:
As far as authority, since there isn't one, it doesn't require the two slashes//
, just ipld:<cid>
( my mistake) . I have conceded that in the Decentralized Identifiers spec (https://w3c-ccg.github.io/did-spec/), the DID is the @id
, basically a valid URL , not a web URL but a valid Uniform Resource Location that is still a mutable document, just a decentralized one that doesn't have a single point of failure.
@Stebalien as discussed, this would be helpful with working in standards development that demand a valid URI schema.