Re: Survey for media-type evolution

Hi Pierre-Antoine,    From a theoretical standpoint, if RDF processing tools correctly implement RFC 6838 (which defines media types and their parameters), the version parameter should be gracefully ignored unless explicitly specified in the tool’s processing logic. RFC 6838 states that parameters not defined in a given media type specification must be ignored by default unless they alter the processing of the content. Therefore, legacy implementations that adhere to this standard should not be affected by the introduction of a version parameter.   Best,   Dominik 
     
      
       
        Dnia 14 lutego 2025 11:37 Pierre-Antoine Champin <pierre-antoine@w3.org> napisał(a):
       
    
       
         

    
  
   
     Dear Semantic Web community, 
     TL/DR: do your RDF tools encounter problems when loading the
      following resources?  
     
        perso.liris.cnrs.fr https://perso.liris.cnrs.fr/pierre-antoine.champin/index.nt  
         perso.liris.cnrs.fr https://perso.liris.cnrs.fr/pierre-antoine.champin/index.rdf  
         perso.liris.cnrs.fr https://perso.liris.cnrs.fr/pierre-antoine.champin/index.ttl 
     -----  
     
     As many of you know, the W3C RDF-Star Working Group [1] is
      currently extending the RDF abstract syntax and the corresponding
      serialization formats. A long standing question that we have to
      respond is whether we should change the media types of the
      formats, or keep them for the new versions of the format [2,3].  
     
     A middle-way was discussed yesterday by the Working Group [4],
      consisting in adding a 'version' parameter to the media-types,
      e.g. 'text/turtle;version=1.2'. The goal of this email is  not 
      to discus the merits and drawbacks of this approach, but to check
      an assumption that some of us have, namely: 
         Despite the fact that the current media-types do not support
      the version parameter currently, its presence will be gracefully
      ignored by current implementation. 
     To test this hypothesis, I have added such a version parameter on
      different versions of my own FOAF profile, and I have checked that
      the following tools load from these URLs without any trouble:  
     
     - sophia-cli ( github.com https://github.com/pchampin/sophia-cli )  
      - rapper 2.0.16 ( librdf.org https://librdf.org/ )  
      - Jena RIOT 5.2.0 ( jena.apache.org https://jena.apache.org/ )  
      - Ruby RDF command line tool 3.3.2 ( ruby-rdf.github.io https://ruby-rdf.github.io/ )  
      - Python RDFlib 7.1.1 ( rdflib.readthedocs.io https://rdflib.readthedocs.io/en/stable/  --
      via the method Graph.parse) 
     It would be great if we could collectively extend this list, to
      assess how disruptive (if at all) the use of the 'version'
      parameter would be on legacy implementations. Please respond to
      this email with your own experiment. 
       thanks in advance,  
        pa 
      
     
     PS: let me emphasize again that, if you want to discuss the pros
      and cons of this solution, you should contribute to [3], but do  not 
      do it in response to this email, to keep the thread focus on the
      survey itself :)  
     
     [1]  www.w3.org https://www.w3.org/groups/wg/rdf-star  
      [2]
       lists.w3.org https://lists.w3.org/Archives/Public/semantic-web/2025Jan/0015.html  
      [3]  github.com https://github.com/w3c/rdf-star-wg/issues/141  
      [4]  www.w3.org https://www.w3.org/2025/02/13-rdf-star-minutes.html#fc3f