Re: [apps-discuss] draft-saintandre-json-namespaces-00 comments

My 2¢: Leave the field wide-open. Let implementers pick the namespace
that gives them the best comfort of avoiding collisions. As far as I can
tell, the best to pick from would be: reverse-domain-name-prefixed
dot-delimited value, URI w. scheme, and even OID in some cases. Allowing
all of these namespace conventions seems to avoid all the collisions
that I can currently imagine.

Paul

On Tue, 2011-11-08 at 20:23 +0100, Julian Reschke wrote:

> On 2011-11-08 20:20, Mark Nottingham wrote:
> >
> > On 08/11/2011, at 1:18 PM, Julian Reschke wrote:
> >
> >> On 2011-11-08 20:10, Mark Nottingham wrote:
> >>> I don't think URIs should be used for this.
> >>> ...
> >>
> >> Well, one use case that the proposal is addressing is the transport of data from frameworks that *already* use URIs; such a WebDAV properties or JCR identifiers. In these cases you really have only the choice of using the identifiers you have, or establishing a completely new identifier system.
> >>
> >> And yes, it would be helpful if the draft was just saying that and not make the impression that it was solving the generic problem for JSON.
> >
> > Ah. I see that as almost an application-specific issue, nothing that should be standardised. Doing that would be leaving sharp tools around small children.
> > ...
> 
> I think it's worthwhile to have a spec that says:
> 
> "If you're identifiers are XML names, then this how to use them in JSON. 
> But if you don't have identifiers like these, don't *start* using them 
> if you want to use JSON."
> 
> Which of course leaves the question open what to do instead :-) (and 
> yes, I saw your proposal)
> 
> Best regards, Julian
> _______________________________________________
> apps-discuss mailing list
> apps-discuss@ietf.org
> https://www.ietf.org/mailman/listinfo/apps-discuss