Re: Social API: Scope

On 3 Aug 2014, at 20:47, Harry Halpin <hhalpin@w3.org> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> 
> 
> On 08/03/2014 07:57 PM, henry.story@bblfish.net wrote:
>> 
>> On 3 Aug 2014, at 19:14, Erik Wilde <dret@berkeley.edu> wrote:
>> 
>>> the assumption here being that tomorrow, RDF might be cool. maybe
>>> it will be, and we'll see tomorrow, i guess. to quote tim bray,
>>> what matters is the bits on the wire. it's kind of hard to get
>>> around this simple truth. cheets, dret.
>> 
>> Teenagers make their decisions based on coolness. As you get older
>> you try to make them based on experience, and where possible try to
>> use foundations that are secure. In this case maths, logic and web
>> architecture.
> 
> Note that the charter quite clearly says we are to focus on JSON as
> *syntax* - i.e. "A JSON-based syntax".  If somehow people have
> real-world software and actual deployments that require XML or some
> other non-JSON compatible RDF, go for it but this is very much
> secondary and should not tackled until we get the JSON serialization
> agreed upon, at least in this Working Group. If you want to start
> working on that now, a Community Group can be made at any time.

As you know I did not suggest an XML. I just suggest that one should 
agree on the vocabulary in terms of URIs and what they refer to. Eg 
foaf:knows refers to the relation of two people who have some interaction 
as defined by the http://xmlns.com/foaf/0.1/knows URI which if you
dereference it ( do an HTTP GET on it ) will give you a document that
explains what the relation is. With that one can produce over the wire
formats in RDF/XML, Turtle, JSON-LD, or any format that can be mapped 
to RDF via a transformation such as json-ld macros

  https://github.com/antoniogarrote/json-ld-macros

That is what I mean by RDF. It is just a question of separating
concerns between what a URI means and how to serialise graphs of
relations. 

> 
> Furthermore, the WG should also use URIs - "should include at least
> the ability to describe the data using URIs in an extensible manner".
> 
> The charter is clear: JSON and URIs. This isn't up for debate.
> 
> Now rather than rat-holing, please provide competing JSON-based syntax
> alternatives to ActivityStreams. That *would* be useful.

In that case the syntax should be JSON-LD, IMHO.
Can one map activity streams to JSON-LD? That should be the principal
question.

Whatever the syntax, we can already come to an agreement on the
semantics. That way we can be solve one issue ( semantics ) without 
the syntax issues getting in the way. And we can agree to a syntax
without having to make decisions about the semantics.

It's just good engineering.

> 
> 
> 
>> 
>> The idea is simple: You specify the meaning of the words, then you
>> allow the data to be expressed in whatever syntax is more
>> convenient.  Because it is useful to have a default, you take the
>> fashion of the moment - currently JSON-LD - as the serialisation on
>> the wire to agree on.
>> 
>> For those with legacy software one can the write a JSON to JSON-LD
>> mapper to make integration easier.
>> 
>> Then if between the time you start this process and the time it
>> ends you find another serialisation more fashionable ( say as
>> happend with Atom between Tim Bray - father of XML - pushing it,
>> and a few years later when JSON started becoming cool ), then you 
>> don't have to change all the existing software.
>> 
>> That also makes a lot of economic sense, which it is true is what
>> managers tend to think as being cool. So I can't completely escape
>> being cool. (sigh!)
>> 
>> 
>>> 
>>>> On Aug 3, 2014, at 10:01, "henry.story@bblfish.net"
>>>> <henry.story@bblfish.net> wrote:
>>>> 
>>>> 
>>>>> On 3 Aug 2014, at 18:53, Harry Halpin <hhalpin@w3.org>
>>>>> wrote:
>>>>> 
>>>>> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
>>>>> 
>>>>> 
>>>>> 
>>>>>> On 08/03/2014 06:02 PM, Erik Wilde wrote: hello james.
>>>>>> 
>>>>>>> On 2014-07-31, 10:32 , James M Snell wrote: FWIW, AS2
>>>>>>> does not *re-base* itself on JSON-LD, it aligns with 
>>>>>>> JSON-LD. It's a critical difference.
>>>>>> 
>>>>>> i think this will get interesting when it comes to defining
>>>>>> an extension model. what was great about AS1 was that it
>>>>>> had both an XML and a JSON syntax, so it was useful for
>>>>>> both communities. once you subscribe to some layer higher
>>>>>> than that, it gets a bit trickier to have a well-defined
>>>>>> domain-based extension model, without resulting in rather
>>>>>> horrible structures in one of the underlying syntaxes.
>>>>>> 
>>>>>> i tried to work on an AS2 XML encoding for a little while 
>>>>>> (analogous to http://activitystrea.ms/specs/atom/1.0/),
>>>>>> because it might be helpful to also serve the XML/Atom
>>>>>> community. but it gets rather tricky to translate AS2's
>>>>>> "alignment" with JSON-LD into reasonable XML constructs.
>>>>>> that's because as an XML user, you'd like to see
>>>>>> XML's/Atom's extension model to be used rather than some
>>>>>> more complicated way of folding what's required by JSON-LD 
>>>>>> into some generic XML mapping.
>>>>>> 
>>>>>> i think it wold be important to discuss whether an XML
>>>>>> syntax is a requirement. if it is, my guess is that this
>>>>>> will have some implications for how much layered models
>>>>>> such as JSON-LD can be used, and where the line has to be
>>>>>> drawn to avoid dependencies on their implicit models.
>>>>> 
>>>>> Actually, according to the charter only a JSON-based syntax
>>>>> is a requirement. The WG can of course have an XML syntax,
>>>>> but the focus on should be on JSON.
>>>>> 
>>>>> cheers, harry
>>>> 
>>>> If the group would manage to agree at the semantic level ( ie,
>>>> one an RDF vocabulary for whatever ) with a default syntax (
>>>> say JSON ), then these issues would just go away.
>>>> 
>>>> Otherwise you'll just spend two years debating syntax issues.
>>>> Yesterday XML was cool. Right now JSON is. Sometime in the
>>>> future something else will be....
>>>> 
>>>> Henry
>>>> 
>>>> 
>>>> 
>> 
>> Social Web Architect http://bblfish.net/
>> 
>> 
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.11 (GNU/Linux)
> Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
> 
> iQIcBAEBAgAGBQJT3oOzAAoJEPgwUoSfMzqczFkP/3o0baI+695BuE53nA/EICBH
> wf/ivA8vDfk+6trB1CMTpQlVJybG8W8oI6lNU5yplcBoDak2mwP7YWcIWyNvxGJ2
> ilwD/xJPOzqt/rBbulpp9wgIcmruzG0TMLuSy/tq0wI9iDDV7qOQ6ge6X1NHw13r
> pxtqPJSBxLyJpNqyBSgxXiTLdxVCwrQvqScWy/u4ZnC/SYO/T+fUn9VMxif+8Wwd
> G1cZ36gSHAPEQ7vNWheEMVg6Dv8lXs445rfyPkt09KNEK251YdsQC3NDInOZvngK
> VdLjEIS3Lk1uMKR7JfzaIr7ESmsqsN/gl2M4Zxupg8L831tyhQvtYxrHrQ9spq55
> XZ852colase8TRy+wJk0G6erKm1ysj1IH8Fm2uMlThydZq2mIxtIRlrhZlRkXKmm
> vqifuZXmmaPKxh2JlMHVIjNsudznPzy26vwc9jivlkZOIGTAHaKhSIOA+kdXvGVO
> H0jkipp7Pft8QM7BbVDktiOT+gdAirleuBZB8epIEVBtaYTVRk0T8pebUdo+t4fJ
> owEyaztfpl3jwohzg3zSw0Rqf89rjBpfgtpAUt2g3u2PZ0VGrw0amEgnXTZMMze+
> J/R+yZcg4p0X8XE7EvttEq4FMH8Ljo8whi6pd0E5pdJ7rg+RHlzsJ9jR1bIUEvDu
> PJpaESjcq2gLunrslwpc
> =NAga
> -----END PGP SIGNATURE-----

Social Web Architect
http://bblfish.net/

Received on Sunday, 3 August 2014 19:04:18 UTC