W3C home > Mailing lists > Public > public-rdf-wg@w3.org > May 2011

Re: Proposal for ISSUE-40 Skolemization

From: Steve Harris <steve.harris@garlik.com>
Date: Fri, 13 May 2011 12:42:59 +0100
Message-Id: <0B4317B0-558A-4BE7-A101-2AF5C12CBF91@garlik.com>
To: "RDF-WG public-rdf-wg@w3.org" <public-rdf-wg@w3.org>
On 2011-05-13, at 08:42, Ivan Herman wrote:
> 
> On May 12, 2011, at 17:58 , Pat Hayes wrote:
> 
>> 
>> On May 12, 2011, at 9:15 AM, Steve Harris wrote:
>> 
>>> On 2011-05-12, at 14:46, Thomas Steiner wrote:
>>> 
>>>>> I'd be tempted to put the proposal to the vote with only the HTTP option and see if it draws any objections.
>>>> 
>>>> +1 for making the URIs dereferencable. Leaves all doors open for
>>>> actually putting useful content there.
>>> 
>>> I'm not keen on enforcing dereferencability. bNodes aren't dereferencable, so it's quite large step.
>> 
>> +1  Being agnostic on this is not closing doors, but it allows skolem-URI-generators to operate at high speed without having to get involved with setting up a web service as backup.
> 
> I think the main value of this whole discussion and proposals is to provide general agreement on how to do skolemization if needed, wanted, (outside of the system, etc), so that parts of the community would not reinvent the wheel. On the other hand, it seems that the issue of dereferencability is important for ones and not important or even to-be-avoided for others. As a consequence I would think that the original proposal of Sandro is the right one, ie, to provide two different approaches (and making the necessary admin steps by the IETF). One is the .well-known for http{s}, and the other is a short form. 
> 
> In terms of compactness the
> 
> urn:steveH:XXXX
> 
> seems to be the best, but that is not fully o.k., because two different organizations might generate the same URI (which is avoided with the HTTP version due to the domain being part of the URI). I guess we may fall back then to the original Sandro approach (having made a really large circle:-)
> 
> tag:yourdomain,2011:steveH:xxxxxxx
> 
> or
> 
> urn:steveH:yourdomain:XXXX
> 
> which slightly less verbose. The advantage of the latter over the tag scheme that it is possible to register steveH for urn, which makes it clearer...

I think we can reasonably put the onus on the minter of the URIs to make them globally unique.

In Nstore we'll just use:
"urn:steveH:".$store_uuid."/".$local_id    ['scuse the perl syntax]

Or something equivalent. If we have to include a domain in there as well it will just make it longer for no good reason.

- Steve

-- 
Steve Harris, CTO, Garlik Limited
1-3 Halford Road, Richmond, TW10 6AW, UK
+44 20 8439 8203  http://www.garlik.com/
Registered in England and Wales 535 7233 VAT # 849 0517 11
Registered office: Thames House, Portsmouth Road, Esher, Surrey, KT10 9AD
Received on Friday, 13 May 2011 11:43:30 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:42 GMT