AX structured objects and querying them
Joseph A Holsten
joseph at josephholsten.com
Thu Dec 10 03:37:53 UTC 2009
What's the holdup on that official WG, anyway? Is what's in the wiki
the proposed WG scope? Seems good enough to work with. I'd bet JSON
the namespacing and query language are big enough problems that
this'll fork into at least one other spec, but that's for a WG to
investigate.
I notice that there's no explicit mention of Breno's versioning use
cases on the wiki page, though it does mention AX validate mode. That
spec will have to be updated to work with the new assertion coding at
least, yes?
--
j
On Dec 9, 2009, at 8:56 PM, Allen Tom wrote:
> +1
> I believe the current thinking is that JSON will be used, since it's
> widely
> used and is fairly compact. Obviously nothing has been specified yet
> since
> there's no official WG.
>
> Allen
>
>
> On 12/8/09 8:44 PM, "David Recordon" <recordond at gmail.com> wrote:
>
>> I'd prefer that we use the JSON structure defined in Portable
>> Contacts
>> which is also now being used by OpenSocial.
>> http://portablecontacts.net/
>>
>> --David
>>
>> On Tue, Dec 8, 2009 at 8:42 PM, Joseph A Holsten
>> <joseph at josephholsten.com> wrote:
>>> What is the thinking about that serialization format? I note the
>>> wiki page
>>> doesn't mention anything. XML, JSON with some namespacing,
>>> something custom
>>> and or backwards compatible?
>>>
>>> When you request attributes, would you normally expect the objects
>>> in
>>> standard types?
>>> as in "I want an email address" -> {"email": "foo at bar.com",
>>> "verification-timestamp": "2009-12-08"}
>>> Or should you be able to request subattributes?
>>> "I want an email address including a timestamp and method of
>>> verification"
>>> -> ...
>>>
>>> Seems like asking for subattributes would be nifty, but quite
>>> nasty to spec.
>>> I sure can't think of any existing way to do this short of SPARQL
>>> or some
>>> NOSQL query language.
>>> --
>>> j
>>>
>>>
>>> On Dec 8, 2009, at 7:59 PM, Allen Tom wrote:
>>>
>>>> I think the more general idea for AX 2.0 is to define a way for
>>>> structured
>>>> data (aka objects) to be shared. This might conflict with the
>>>> goal of
>>>> making
>>>> the messages compact.
>>>>
>>>> At least with the case of email address, we'd like a way to
>>>> return the
>>>> email
>>>> address, as well as attributes about the address. Ideally, we'd
>>>> make this
>>>> mechanism generic so that any attribute can have metadata
>>>> associated with
>>>> it.
>>>>
>>>> Allen
>>>>
>>>>
>>>>> The spec would need to require the colons, and probably permit
>>>>> empty
>>>>> strings in place of the numbers when an OP wanted to say nothing
>>>>> at all
>>>>> about validation, e.g.
>>>>>
>>>>> openid.ax.d.c=::+1-555-555-1212
>>>>>
>>>>
>>>
>>> _______________________________________________
>>> specs mailing list
>>> specs at lists.openid.net
>>> http://lists.openid.net/mailman/listinfo/openid-specs
>>>
>
More information about the specs
mailing list