On Dec 15, 2009, at 12:20 PM, Dick Hardt wrote:
On 2009-12-15, at 7:08 AM, Chris Obdam wrote:
Do we already have an inventorisation of what attributes and object
should be included into the model? There have been many
discussions, but I am losing grip. Maybe a Wiki? Where people can
suggest needed attr next to what already is available on axschema.org
And what do we do wit axschema? There is the schema (formal) and
the implementation (JSON, something else). Shouldn't we split these
discussions?
Yes they are two different discussions. There is the message schema
and the attributes that are being moved around.
+1
At IWW there was a discussion about what AX could look like. The
queries and responses need more syntax then is provided by simple
name/value pairs. It is a little embarrassing looking at AX 1.0 now
with the overloaded lvalue. XML and JSON are well understood
syntaxes for rich objects. XML looks to have more power/complexity
then needed, JSON looks just right.
I wish I could agree, but the current draft charter[1] mentions "Each
attribute property schema is bound to a unique attribute-type
namespace, can be described by a standard key string (does not need to
be defined through a URL value)." JSON doesn't have a formal or de
facto way to do that.
Am I just trolling on this issue? Has someone proposed a way to do
this namespacing in JSON? Dropped the requirement?
In the spirit of constructive criticism, here's a proposal: the top
level keys of the JSON object must be AXSchema attribute names. This
allows the existing attribute types to be used beside new structured
types. You request things by top level attribute name. That's a
tradeoff in granularity and means you either request the entire PoCo
record, or none of it.
1: http://wiki.openid.net/OpenID_Attribute_Exchange_Extension_2_0
--
Joseph Holsten
http://josephholsten.com
mailto:[email protected]
tel:+1-918-948-6747
_______________________________________________
specs mailing list
[email protected]
http://lists.openid.net/mailman/listinfo/openid-specs
