Re: Absolute IRIs (Was: Re: IRI guidance)

On Mon, May 2, 2011 at 2:32 PM, Nathan <nathan@webr3.org> wrote:

> Eric Prud'hommeaux wrote:
>
>> * Ivan Herman <ivan@w3.org> [2011-05-02 19:39+0200]
>>
>>> On May 2, 2011, at 19:27 , Eric Prud'hommeaux wrote:
>>>
>>>  * Alex Hall <alexhall@revelytix.com> [2011-05-02 11:44-0400]
>>>>
>>>>> On Fri, Apr 29, 2011 at 2:28 PM, Ivan Herman <ivan@w3.org> wrote:
>>>>>
>>>>>
>>>>>> On 29 Apr 2011, at 19:50, Eric Prud'hommeaux <eric@w3.org> wrote:
>>>>>>
>>>>>>> I'm not personally keen on this absolute IRI restriction. I included
>>>>>>> it in this proposal in order to minimize the permutations being
>>>>>>> examined at once ("minimal change"). For usability, I find
>>>>>>> Data:
>>>>>>>  <s> <p> <o> .
>>>>>>> Query:
>>>>>>>  ASK { ?s <p> ?o }
>>>>>>>
>>>>>>> very intuitive when you don't have to specifically call out a base
>>>>>>> URI. Using IRI references instead of IRIs would permit the above
>>>>>>> query
>>>>>>> to work in e.g. Jena (which currently presumes absolute IRIs).
>>>>>>>
>>>>>>>  Is there a need for this outside the context of illustrating some
>>>>> simple
>>>>> test data and queries?
>>>>>
>>>> It's really just a minor usability/simplicity point. The Direct
>>>> Mapping of Relational Data to RDF maps a relational database to an RDF
>>>> graph with all relative IRIs. Custodians of the data can treat it as
>>>> they would a tarball of HTML docs in a filesystem, where the access,
>>>> be it e.g. HTTP backed by some Apache configuration, or directly via
>>>> file://localhost IRIs, determines the base. Like the browser's ability
>>>> to navigate relative links, SPARQL queries can elide the base,
>>>> matching RDF graphs regardless of access. When it doesn't work, I'd
>>>> say it's a usability obstacle a little worse than issue 18 .
>>>>
>>> But, at this moment, we are discussing RDF concepts and not a particular
>>> serialization. One can use relative URI-s with @base in turtle, or the
>>> equivalents in other serializations. But I do not see how the introduction
>>> of relative URI-s into the RDF Concepts, Semantics, etc, could be a minor
>>> point...
>>>
>>
>> Some implementors interpret the last URI references note in RDF Concepts
>> as saying that they are not allowed to put relative IRIs into an RDF graph.
>
>
I wasn't aware that this was a point of confusion -- if that's the case then
certainly we should clarify that note.  I can't suggest anything because I
still don't understand the source of the confusion.  Which notion of graph
is it that implementors want to put a relative IRI into -- g-snap, g-box, or
g-text?


> My question is, does the world get better or worse if we leave that note
>> out. I've found RDF pipeline processing (constructs, queries, service
>> federation, etc.) to be handy with relative IRIs, but we don't want
>> persistent aggregators (e.g. an RDF Google) to be polluted with lots of
>> ambiguous relative IRIs.
>>
>
I see the value of using relative IRIs internally, but I think that's an
issue of tool support and not standards.  As soon as you expose those IRIs
outside your own system, they get resolved as absolute IRIs relative to a
base IRI as determined by some context.  If you don't specify the base IRI
then it's generally determined to be the location where you published the
data.  I think the only way to prevent persistent aggregators from being
polluted with ambiguous relative IRIs is to disallow relative IRIs in the
abstract syntax.


>
> I'm missing something here, the value space is IRIs yes, but regardless of
> where you're working you always deal with some form of lexical space
> (serialization, memory structure, rdbms) - is there any limitation anywhere
> that says you can't store / represent an IRI as a relative reference rather
> than an absolute value? Surely that's behind the public interface and in the
> realm where anybody can do whatever they want.
>
>
I agree.  Another way of saying that is that the restriction to absolute
IRIs applies to g-snaps; g-boxes and g-texts are free to use relative IRI
references.  The limitations on storing an IRI as a relative reference are
generally imposed by tools which enforce the restriction to absolute IRIs.

-Alex



> Best,
>
> Nathan
>