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

And another +1 to the -10.

To make the reason clear, the problem this causes is that names are no longer able to be treated opaquely.  Having "{$foo}x" with more than one potential semantic in the same context is a crazy sort of problem that only an unnecessary optimization can produce.

I'm generally with Mark, and Paul.  The lessons learned from "X-" and XML indicate that a private playpen is nice.  Proscribing the shape of that playpen seems unnecessary, at least for now.  

On the one hand, you could attempt to establish a convention for the shape of the playpen.  This draft does that with {} and URIs.  There's certainly merit in convention, but the proposal is heavyweight.

On the other, all you need is a mechanism to avoid collisions in a given host format.  Given the identifier's inherent opacity, extensions can take any form (c.f. Paul's comment) and the goal is still achieved.  You can use a registry and bear the management overhead; use dotted names, uris or something similar; you can just use real names and risk collisions; or you can combine the approaches.

For instance, you could say: "official" extensions wont use '.', come see us if you want one; otherwise, try to play nice.  For instance, I haven't seen too many problems with namespace collisions in Java class names.

--Martin

On 2011-11-09 at 04:10:44, Mark Nottingham wrote:
> +1 to Paul's -10.
> 
> My take on it:
> 
> http://www.mnot.net/blog/2011/10/12/thinking_about_namespaces_in_json
> 
> Let's keep this simple and avoid repeating the mistakes of XML, OK?
> 
> 
> On 08/11/2011, at 11:05 AM, Paul Hoffman wrote:
> 
> > On Nov 8, 2011, at 4:42 AM, Dominik Tomaszuk wrote:
> >
> >> Clark Notation can sometimes be too heavy, especially when
> namespaces are placed in many keys.
> >
> > +1
> >
> >> I propose to give possibility to declare namespace.
> >
> > -1. Actually, -10. While somewhat useful for XML, namespaces have
> also proven to cause many headaches for corner cases.
> >
> > Please strongly consider using a much simpler naming scheme, such as
> "names with periods in them".
> >
> > --Paul Hoffman
> >
> > _______________________________________________
> > apps-discuss mailing list
> > apps-discuss@ietf.org
> > https://www.ietf.org/mailman/listinfo/apps-discuss
> 
> --
> Mark Nottingham
> http://www.mnot.net/
> 
> 
> 
> 
> _______________________________________________
> apps-discuss mailing list
> apps-discuss@ietf.org
> https://www.ietf.org/mailman/listinfo/apps-discuss