W3C home > Mailing lists > Public > public-rdf-comments@w3.org > July 2012

Re: Skolemization .well-known prefix: genid --> bnode or blanknode

From: Steve Harris <steve.harris@garlik.com>
Date: Thu, 19 Jul 2012 10:39:33 +0100
Cc: <nathan@webr3.org>, "'Gavin Carothers'" <gavin@carothers.name>, "'Gregg Kellogg'" <gregg@greggkellogg.net>, "'David Booth'" <david@dbooth.org>, "'Richard Cyganiak'" <richard@cyganiak.de>, "'public-rdf-comments'" <public-rdf-comments@w3.org>
Message-Id: <E4E0D9B8-CAB5-432B-97EC-274253C99C7D@garlik.com>
To: Markus Lanthaler <markus.lanthaler@gmx.net>
On 2012-07-19, at 09:48, Markus Lanthaler wrote:

> Nathan wrote: 
>> Gavin Carothers wrote:
>>> On Wed, Jul 18, 2012 at 7:05 PM, Gregg Kellogg
>> <gregg@greggkellogg.net> wrote:
>>>> On Jul 18, 2012, at 11:01 AM, David Booth wrote:
>>>> Well, I wasn't a WG member when this was decided, but I did see it
>> go by. IMO, the term "gemid" means nothing. To me, it says just
>> "generated identifier", and nothing that would indicate that it denotes
>> an unnamed resource (or existential identifier). My preference would be
>> for something more like http://example.com/.well-
>> known/anonid/d26a2d0e98334696f4ad70a677abc1f6, which provides a better
>> linguistic clue as to what the IRI is intended to represent.
>>> 
>>> http://example.com/.well-known/existential/blahblah or perhaps
>>> http://example.com/.well-known/∃/blahblah (lets break some bad IRI
>>> implementations ;) )
>>> 
>>> Only half kidding. The name for the name for something that is
>>> explicitly not a name or denotation is always going to be funky.
>> 
>> "funky" is one way of putting it - weird thing is, outside of all the
>> conversations about skolemization in the RDF WG, this seems like a
>> ridiculously stupid idea that breaks about every principal of logic and
>> web architecture and common sense.
>> 
>> But then somehow all these clever people talking about it makes it seem
>> rationale. I think.
> 
> Just as Gregg I'm quite new to this WG and wasn't involved in the discussions. I'm wondering though why the spec (somewhat) encourages to use http(s) IRIs for this even though they are clearly not intended to identify a globally unique and de-referenceable resource.
> 
> Perhaps, from an Web architecture point of view, establishing a new URI scheme or at least using something like the uuid scheme would be a better option!? Just my two cents.

Well, the key point is that the server holding the data sometimes /can/ mint a globally unique and dereferencable URI for them. HTTP(S) is described because that's where the .well-known/ scheme is defined. Systems are free to use URNs or whatever for them when they can't be dereferenced, or ignore it altogether if they can't support the behaviour.

- Steve

-- 
Steve Harris, CTO
Garlik, a part of Experian
+44 7854 417 874  http://www.garlik.com/
Registered in England and Wales 653331 VAT # 887 1335 93
Registered office: Landmark House, Experian Way, Nottingham, Notts, NG80 1ZZ
Received on Thursday, 19 July 2012 09:40:10 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:59:30 UTC