Re: [apps-discuss] Review of draft-ietf-appsawg-file-scheme

[Matthew Kerwin <[e-mail]>]

On 13 April 2016 at 20:34, Graham Klyne <[e-mail]> wrote:

>
> As for other URI specs that don't define a protocol...
>
>     urn:
>     acct:
>     about:
>     cid:
>     mid:
>     ni:
>     (etc)
>
> (That's from a very brief look at
> http://www.iana.org/assignments/uri-schemes/uri-schemes.xhtml)
>
>
Urgh, of course you're right, and I knew about those. In my mind I was
still hung up on the difference between a purely naming scheme (like urn)
and a very dereferenceable one (like mid).

As far as precedent, mid: and cid: are both dereferenceable URLs (that is,
they support an operation like HTTP's GET), but they're really old so don't
have the formalisms we expect in a modern spec. (In fact, they seem to
assume that the fact that they can get 'got' at all is basically implied by
the fact that they're URLs. (I'm sorry for writing that sentence.))

acct: and ni: are nice and modern, but don't define any methods per se
(although they do point to webfinger and/or define a mapping to HTTP URIs).

Given the precedent, and Dave's protocol-vs-format comment, I'm wondering
if it's even appropriate to include methods in the file: spec at all (BCP
35 only says the definition SHOULD include operations.)

Cheers
-- 
  Matthew Kerwin
  http://matthew.kerwin.net.au/