Advice on NID for media fingerprint

[Jonas Oberg <[e-mail]>]

Dear all,

over the last few years, I've worked on fingerprint algorithms for
media files (mostly images, but lately also video). As part of the
work my team and I have done on images, we put together a draft urn
scheme for a perceptual image blockhash algorithm intending to request
the "blockhash" Namespace ID.

You will find this here:

  https://raw.githubusercontent.com/commonsmachinery/blockhash-rfc/master/draft-commonsmachinery-urn-blockhash-00.txt

We believe this to be generally useful, and it would definitely help
ascertain without doubt what is meant with a fingerprint expressed
using this urn scheme.

However, we also realise the necessity of being able to facilitate the
inclusion of additional algorithms, and to publish new and revised
versions of our blockhash algorithm to take into account new
developments.

So our current line of thought is to request an informal URN
namespace, where we define the NSS in a way that it includes both an
identifier (possibly with version number) for the algorithm, and the
fingerprint itself.

A URN in this way could take the form of:

  urn:urn-n:blockhash:v1:XYZ

This seems more in line with our original intention and much more
flexible as we move forward with this. We also do not see there's
a need to establish IETF Consensus for this purpose (which would be
needed for a formal URN namespace, as we understand it).

Any thoughts on this matter would be well appreciated, otherwise I
would intend to move on with defining this informal URN namespace in
some more detail and request this for approval sometime soon.


Sincerely,

-- 
Jonas Öberg, Executive Director
Free Software Foundation Europe | [e-mail]
Your donation enables our work (fsfe.org/donate)