Re: [apps-discuss] proposing draft-farrell-ni as an appsarea wg item

[Peter Saint-Andre <[e-mail]>]

On 7/28/11 9:41 AM, Stephen Farrell wrote:
> 
> 
> On 28/07/11 02:25, Linyi Tian 01 wrote:
>>
>>
>> On 11-7-27 上午11:49, "Stephen Farrell" <[e-mail]> wrote:
>>
>>>
>>> Hi,
>>>
>>> Yesterday, I presented a draft [1] on a URI scheme for naming
>>> stuff to the decade WG. There's interest in that there but
>>> the idea behind this is really broader than just decade so
>>> if there's interest here, doing this work in the appsarea WG
>>> would I think be much better.
>>>
>>> The scheme is basically like HTTP URIs with one additional
>>> thing, which is that it specifies a way to include a hash
>>> function output in the name (and the related input, in case
>>> you want to verify something later) which seems to me to be
>>> a generally useful thing that various protocols could use
>>> in various ways. Hence the idea of bringing it here.
>>
>> I am wondering how this would be used. Will it be used together with HTTP
>> URIs? Is there any real life use case how I can use this URI.
> 
> There are two ways in which this could be used I think, one is
> where a protocol carries a URI and then the new scheme could be
> used. For HTTP URLs one could use the hash-string construct
> within the pathname I guess and that might be useful to document
> as an option.
> 
> As I said the decade wg are thinking about this for their protocol
> work. There are also many HTTP URLs that contain hashes, the extent
> to which it'd be useful for someone to be able to know what was input
> to the hash is hard to tell. I suspect it'd be useful sometimes
> though.

I've heard about other folks who are interested in a very simple method
for either a URN or a URI for identifying the output of a hash function.

Possible examples of URNs and URIs:

urn:hash:sha256:NDc0NzgyMGVmOGQ3OGU0...

sha256:NDc0NzgyMGVmOGQ3OGU0...

I don't think this should be a URN, because a URN namespace is managed.
(In this sense RFC 4122 is misguided and deserves to be deprecated.)

I don't see the need for an authority component, but maybe I'm missing
something in the use cases.

Peter

-- 
Peter Saint-Andre
https://stpeter.im/