[JENA-828] Handling strings for RDF 1.1 in SPARQL

• Type: Task
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Jena 2.12.1
• Fix Version/s: Jena 2.13.0
• Component/s: ARQ
• Assignee: Andy Seaborne
• Reporter: Andy Seaborne
• Created: 14/Dec/14 16:51
• Updated: 23/Dec/14 18:10
• Resolved: 23/Dec/14 18:10

Description

RDF 1.1 defines simple literals and simple literals to be the same (literals with datatype xsd:string). This affects the sorting, where the SPARQL 1.1 spec says simple literals sort before datatype literals.

Best compatibility is to make sorting: xsd:string < all other datatypes with same lexical form.

ARQ also defines ordering between any two RDF terms in a stable, predictable way based on lexical form first, then simple literals < lang literals < datatypes by URI. This is not covered by the SPARQL spec

For RDF 1.1 this becomes

1. Order by lexical form
2. Order by xsd:string < language literals.
3. Order by language literals, base don language tag (case-insensitive, then case senstive).
4. Datatypes by URI string

A language literal has datatype rdf:langStirng but also has a non-empty language tag.

Issue Links

• is depended upon by: JENA-193 Changes needed, or desirable, for RDF 1.1
• is related to: JENA-816 Handling simple literals,language literals and xsd:string in RDF 1.1