Re: [lyo-dev] FW: [Bug 533213] New: rdf:Seq deserialization in JMH conta

[Andrii Berezovskyi <andriib@xxxxxx>]

Sure, Ricardo,

 

If something is possible to implement with a rather limited effort, let’s do it. It is always great to be able to handle any RDF that comes from the outside!

 

/Andrew

 

Den 2018-05-08  18:42 skrev "lyo-dev-bounces@xxxxxxxxxxx på uppdrag av Ricardo Javier Herrera Ledesma" <lyo-dev-bounces@xxxxxxxxxxx på uppdrag av neormx@xxxxxxxxx> följande:

 

Andrii, I agree with you that Bags are useless, and I've seen also solutions where an extra property is added to the model in order to keep the order, however sometimes it may be useful to leave the order concerns to the Jena/List themselves. Currently I'm working in a model where an entity is related sometimes with 0..* entities of the same type and the order matters:

 

class Entity {

   List<Entity> children;

}

 

Adding an extra property to keep the order...

 

class Entity {

   int index;

   List<Entity> children;

}

 

has the inconvenient of you having to sort the elements by yourself once deserialized by JMH, not to mention that a model with a property that is used only if the instance is part of a collection is not so good... this also hits performance, if you have a complex hierarchy of parent/children, the sorting algorithm will have to run once per children list, it would be better if the children get sorted since the deserialization process.

 

Regarding rdf:Seq are unqueryable in SPARQL, does this have to do with the fact that they are stored as blank nodes? If so, I think we can modify JMH to correct that, actually in this moment I have a copy with such changes and it is working quite well.

 

I think that, in general, if OSLC lyo library has an @OslcRdfCollectionType annotation that accepts @OslcRdfCollectionType(collectionType = "Seq"), it is worth that these changes be applied, they do not impact other parts of the code. I'll work in a Gerrit patch to show it you all and you can considere it.