search

multiformats / multihash

Self describing hashes - for future proofing
https://multiformats.io/multihash/
MIT License
895 stars 114 forks source link

Is there an agreed-upon URI and namespace for multihash? #132

Open IS4Code opened 3 years ago

IS4Code commented 3 years ago

In XML signatures and linked data, one would use a URI for a particular hash function, like http://www.w3.org/2001/04/xmldsig-more#md5 for the MD5 function, and the prefix urn:md5: to identify a byte stream with its hash (for use in a magnet: link for example). Sorry if this is not the right place to ask, but are these conventions used here? I have found one instance of urn:multihash:, so that seems usable enough, but I am clueless about the URI.

One could say that if you already have a way to specify the hash function together with its result, there is not much advantage to using multihash, but not all hashes have existing identifiers, so a way to use its identifier table would be useful.

vmx commented 3 years ago

I guess if a URI is required, it would then just point to Multihash, without pointing to a specific algorithm as the Multihash value itself already contains the information about the hash algorithm.

troyronda commented 3 years ago

An urn:multihash namespace would be useful.

It would also be good to include the multibase character.

urn:multihash:<mb><multihash>
melvincarvalho commented 2 years ago

@IllidanS4 I raised this same question in the W3C community group on multihash

See these two specs proposed to the IETF:

https://www.ietf.org/id/draft-multiformats-multibase-05.html https://www.ietf.org/id/draft-multiformats-multihash-04.html

FYI: here is the issue I raised:

https://github.com/w3c-ccg/multihash/issues/20

It seems that we could register a mh:// URI scheme perhaps, but this does not take into account multibase and CIDs. Still room for discussion but it would be good to get something stable, nailed down quite soon for those interested in the URI part.

Currently mh: is preferred slightly to multihash: as it is more compact. While less descriptive, I think most could live with mh and it mirrors things like ni

rvagg commented 2 years ago

could mh:// just dictate that the following string be a multibase representation of the multihash?

joakim commented 1 year ago

When there's no authority, there shouldn't be a //, shaving off a few more bytes.