[JENA-1477] Updating from Jena 2.12 to Jena 3.0.1 produces different RDF/XML output IF ARQ is loaded

• Type: Question
• Status: Closed
• Priority: Minor
• Resolution: Done
• Affects Version/s: Jena 3.0.1
• Component/s: RDF/XML, RIOT

Environment

Jena 3.0.1

• Assignee: Unassigned
• Reporter: Andrew Berezovskyi
• Created: 03/Feb/18 16:53
• Updated: 03/Feb/18 22:15
• Resolved: 03/Feb/18 21:57

Description

We are in the process of a Jena 2>3 transition in Eclipse Lyo and discovered the following: if the ARQ is used in the application, then the typed node elements will be used by default, which did not occur in Jena 2. It causes problems for some legacy apps that rely on Lyo.

We were able to eliminate this behaviour by calling the static method

RDFWriterFImpl.alternative(null);

Technically, this ensures that the 'org.apache.jena.rdfxml.xmloutput.impl.Basic' implementation is not getting substituted for the 'org.apache.jena.riot.adapters.RDFWriterRIOT'.

Questions:

1. Is it the best way to ensure backwards compatibility of the RDF/XML output when moving from Jena 2>3?
2. Is there any way to control the RIOT writer to produce XML w/o typed node elements so that we can avoid resetting the alternative writer?
3. (if you have some extra time) What purpose do alternative writers serve? What is the difference between the writers in the respective packages as the two writers above?