Re: Portability Requirement

On 03/15/2016 03:00 PM, Melvin Carvalho wrote:
>
>
> On 15 March 2016 at 19:54, Shane McCarron <shane@halindrome.com
> <mailto:shane@halindrome.com>> wrote:
>
>     I am not sure what to say.  I heard a strong requirement from the
>     group on the call today that this be included.  As far as I know,
>     all this means is that credentials (aka claims) need to have a well
>     defined data format that is transportable.  The repository for a
>     claim could be a folder fill of .json files on a disk, right?  Or
>     some fancy service provider.  The claims and their integrity are not
>     dependent upon the repository.
>
>
> Let me clarify slightly.
>
> If this simply means that the serialization reuses web standards such as
> JSON LD and Linked data then im +1, because that's portable by nature.
>
> If it means that the subject of those data structures requires the use a
> URI scheme other then HTTP (e.g. the did: scheme) URIs, that's the point
> I'd like to push back on.

There is no requirement for that to happen.

To give an example of how this might work with a WebID (but note that no 
technologies have been chosen yet), suppose that if you have some third 
party claims about the subject "https://example.com/people#melvin". You 
currently store those at "https://credbank.com", but now you'd like to 
move them to "https://best.credrepo.com". You can do that without having 
the claims reissued. The identifier you use for yourself has nothing to 
do with it. Your identifier can be *independent* of the services that 
issue claims to you and so can the repository where you store those claims.


-- 
Dave Longley
CTO
Digital Bazaar, Inc.
http://digitalbazaar.com

Received on Tuesday, 15 March 2016 19:32:02 UTC