On @iri/@subject, @type/@datatype (#15, #31) and coercion (was: Re: [json-ld.org] Are @subject and @iri redundant? (#15))

I have the opposite position on both accounts.. ;)

On issue #15: @iri and @subject are synonymous, and I see little value
in keeping both. I have only used @iri (actually I use "iri": "@iri"
in my context and thus use "iri"). You can always use aliases if you
want sugar for certain cases.

On issue #31: @type is fundamentally different from @datatype. It is
an alias for "http://www.w3.org/1999/02/22-rdf-syntax-ns#type",
whereas @datatype is part of the expanded syntax for literals. Where I
to alias "@type" to e.g. "a" (which I do), I do *not* want "a" to be
possible to use to give the datatype of an expanded literal (nor in
coercion, if we are to replace "@coerce" with "@datatype" in the
context). Granted, I could define "a" to be "just" the IRI for
rdf:type, but in any case having the key for datatype mean rdf:type
outside of literals seems confusing and non-pedagogical. I think this
weighs heavier than that since they aren't intended for use in the
same place their meaning *could* be inferred.

.. For completeness, I'm also wary of using "@datatype" instead of
"@coerce" in the context. This is because "@coerce": "@iri" means
something different from what datatypes are about. Of course, we can
do something similar to Gregg's proposal of "@list": true (which I
like), and use "@iri": true instead of coercion. But that would also
mean forbidding/overriding any "@datatype" coercion directive. I'm may
still be open to going down that path though, as I'm also interested
in investigating the relative value of e.g. using "@language" to
define language-specific terms (e.g. "titleSv: {"@iri": "dc:title",
"@language": "sv"}).

.. Also, as mentioned before, I need a mechanism for saying that the
values for a certain term will *always* be a JSON array (i.e. a
"set"). I have evaluated if I can leave that aspect to the (much more
advanced) use of framing, and I've concluded that that is much less
viable.

Best regards,
Niklas



On Tue, Nov 29, 2011 at 1:10 PM, Markus Lanthaler
<markus.lanthaler@gmx.net> wrote:
> This issue was originally raised by Gregg:
>
> http://lists.w3.org/Archives/Public/public-linked-json/2011Jul/0060.html
>
> I think in this case it makes sense to keep both @subject and @iri even though @subject is just syntactic sugar. So for me it would be fine to close this issue.. let's see what Gregg thinks about it - I've CCed him.
>
> P.S.: For me this issue is quite different from the @type/@datatype issue (#31). I wouldn't like to treat them the same way.
>
>
>
>> -----Original Message-----
>> From: Manu Sporny [mailto:reply+i-1477642-
>> 03cf6838214958976e2f68a8ac53ef0de003a547-456407@reply.github.com]
>> Sent: Tuesday, November 29, 2011 12:42 PM
>> To: Markus Lanthaler
>> Subject: Re: [json-ld.org] Are @subject and @iri redundant? (#15)
>>
>> Technically, they are redundant. However, I think that just like @type
>> and @datatype, we should keep the two concepts separate to help people
>> understand what is going on in the markup a bit better.
>>
>> ---
>> Reply to this email directly or view it on GitHub:
>> https://github.com/json-ld/json-ld.org/issues/15#issuecomment-2912823
>
>
>

Received on Tuesday, 29 November 2011 13:27:00 UTC