Re: [Multiformats] Fwd: [media-types] [IANA #1272257] application/vnd.ipfs.ipns-record registration request

[Bumblefudge <[e-mail]>]

> Does anyone have additional details on this request? > Or perhaps a shorter explainer for how this might fit together with multihash?

Sorry, I was reviewing old emails and I realized I answered this out-of-band instead of on-list. For anyone else who might have been curious, IPNS records are a structured data representation defined here: https://specs.ipfs.tech/ipns/ipns-record/#record-serialization-format The short-ish explainer would be that IPNS records are a protobuf data structure similar in function to DID Documents, in that they contain: - a mutable pointer to immutable (and CID-addressed) data, - some metadata to make that pointer verifiable[^1], and - their "Names"[^2], a stable identifier derived from the specific keypair proving authority over the name over time. The resolution of IPNS names to IPNS records is implemented as part of the public IPFS DHT, but could just as easily use any other discovery and resolution mechanism. and resolvable to that document on the DHT, is a binary multihash of the public key serialized. They rely on multihash in the sense that longer keys are represented transformed and code-prefixed according to multihash logic; the legacy form and most common form, however, use shorter Ed25519 keys which are simply prefixed with the "identity" multihash code and represented in their entirety. It is also worth noting that since IPNS names and records alike are canonically binary (like CIDs more generally), but often multibased as CIDs.

But I feel like maybe the implicit question was why IPNS records, of all the higher-order constructs built on a multiformats basis, were registered as a content type. I'm not intimately familiar with the libp2p/networking section of the multicodecs registry or the multiaddr spec (a translation layer for addressing systems and identifiers), but I would assume it was for interoperability at the networking layer. I can chase down a more concrete rationale if this list is curious, although, as I said above, IPNS is at least as far out of scope as CIDs in general, if not more so.

Thanks, __bumblefudge [^1]: https://specs.ipfs.tech/ipns/ipns-record/#ipns-record [^2]: https://specs.ipfs.tech/ipns/ipns-record/#ipns-name [^3]: https://specs.ipfs.tech/ipns/ipns-record/#string-representation

---
bumblefudge
janitor @Chain Agnostic Standards Alliance
contractable via learningProof UG
mostly berlin-based