Re: [apps-discuss] draft-ietf-appsawg-json-pointer

I agree. JSON specifies only strings as member names, so there is no
ambiguity when resolving within an object. This issue boils-down to the
difference between JSON objects and Python dictionaries.

To help answer your question, perhaps I could know more about the Python
JSON library you're using to serialize Python data structures. How will
it handle the case where a key is not a string?

Without knowing this, I would be inclined to fail on non-string keys,
because the data structure is not conforming to the JSON specification.

Paul

On Wed, 2012-10-24 at 16:02 +1100, Mark Nottingham wrote: 

> [forwarding to the WG]
> 
> Hi Stefan,
> 
> My .02 - this is really a question about JSON implementation, I think. JSON-Patch is designed to operate upon JSON documents (where this question doesn't come up), not upon Python data structures.
> 
> Cheers,
> 
> 
> On 24/10/2012, at 3:04 AM, Stefan Kögl <stefan@skoegl.net> wrote:
> 
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> > 
> > Dear Working Group,
> > 
> > I'm the maintainer of python-json-pointer [1], a Python implementation
> > of the json-pointer drafts.
> > 
> > In Python the natural representation of a JSON object is a dictionary.
> > One of the differences is that a Python dictionary can have keys of
> > different types. I could, for instance, define the following dictionary
> > 
> > {"1": "a", 1: "b"}
> > 
> > and try to resolve the pointer /1. I see several possible ways of
> > resolving such cases:
> > 
> > * Ignore any non-string key
> > * Fail on non-string keys
> > * Interpret both keys as equal and fail according to the following
> >  sentence of the current draft:
> >  " If a referenced member name is not unique in an object, the member
> >    that is referenced is not unique in an object, the member that is
> >    referenced is undefined, and evaluation fails (see below). "
> > 
> > I understand that this problem can not occur in "pure" JSON where keys
> > can only be strings, but I assume that also implementations in other
> > languages will face the same issue. What is the opinion of the working
> > group on this? Maybe it could be addressed in future drafts.
> > 
> > 
> > Best regards,
> > 
> > Stefan Kögl
> > 
> > 
> > [1] https://github.com/stefankoegl/python-json-pointer
> > -----BEGIN PGP SIGNATURE-----
> > Version: GnuPG v1.4.11 (GNU/Linux)
> > Comment: Using GnuPG with Mozilla - http://www.enigmail.net/
> > 
> > iEYEARECAAYFAlCGwCUACgkQ638pvJMAJ9BbsQCfYeu9SVDSKE2Z6DBYUKaAl1/b
> > Z9gAoLRaMjzBgun8K8Wd3QEic0bYTxaC
> > =ttNc
> > -----END PGP SIGNATURE-----
> 
> --
> Mark Nottingham   http://www.mnot.net/
> 
> 
> 
> _______________________________________________
> apps-discuss mailing list
> apps-discuss@ietf.org
> https://www.ietf.org/mailman/listinfo/apps-discuss