W3C home > Mailing lists > Public > public-openannotation@w3.org > June 2012

Re: Consistency for FragmentSelector value

From: Stian Soiland-Reyes <soiland-reyes@cs.manchester.ac.uk>
Date: Tue, 26 Jun 2012 12:29:35 +0100
Message-ID: <CAPRnXtm3gWacgG34qDbCgAjXhd-GpMCv=YkyenEjsZQT9ExmwA@mail.gmail.com>
To: James Smith <jgsmith@gmail.com>
Cc: Robert Sanderson <azaroth42@gmail.com>, public-openannotation@w3.org
Agreed  - so go for rdf:value - there is only one way to have
fragments (you can't do a fragment as a set of bytes), the fragment
string does fully represent the FragmentSelector, and it does not have
additional properties like content type.

Also going into the area of string encoding of fragment selectors
sounds like a mine field..

On Wed, Jun 20, 2012 at 5:51 PM, James Smith <jgsmith@gmail.com> wrote:
> On Jun 20, 2012, at 12:36 PM, Robert Sanderson <azaroth42@gmail.com> wrote:
>> Dear all,
>> Please could you weigh in on a minor issue of consistency.
>> In the current specification, oa:FragmentSelector uses rdf:value to
>> record the fragment, whereas other resources use the Content in RDF
>> specification to include their data into the graph.
>> Should oa:FragmentSelector also use cnt:chars, or is it more like
>> TextOffsetSelector/TextQuoteSelector in that the properties should be
>> just part of the graph and not able to be exported?  Is there a clear
>> rule that we can use to determine which is appropriate?
> My feeling is that FragmentSelector is like TextOffsetSelector. It's value is a valid fragment, which isn't open to mime type interpretation or character encoding. Fragments have a very specific format (RFC 3986, section 3.5 [pg. 24]) that assumes a UTF-8 superset of US-ASCII (or other suitable superset of US-ASCII) meaningful in the context of the resource to which the fragment is attached.
> We might interpret fragments, but that interpretation should be outside the scope of the FragmentSelector schema.
> I think the general rule could be that if a resource could stand on its own as a resource (i.e., if we have examples in the wild of similar content being placed at a URL for consumption), then it can use the cnt:chars and related properties. If the value is just a value that depends on some other related context for meaning, then it doesn't need the cnt:chars & co. properties.
> -- Jim

Stian Soiland-Reyes, myGrid team
School of Computer Science
The University of Manchester
Received on Tuesday, 26 June 2012 11:30:28 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:22:00 UTC