Schema without type

[ddeboer]

Hey @karelklima, thanks for this cool library! I just started using it.

In the docs you write:

Setting the RDF type is recommended for optimal performance.

Why, exactly?

What if one wants to query a polymorphic property? E.g. schema:potentialAction takes AchieveAction, AssessAction, ConsumeAction etc.

      potentialAction: {
        "@array": true,
        "@id": schema.potentialAction,
        "@schema": {
          "@type": ???,             // Leaving out @type altogether makes the query really slow.
          ...
        }

[karelklima]

Hi @ddeboer, thanks for the comment!

LDkit works in a way that it creates a single SPARQL query for each action that you do with it, e.g. if you want to query for a single entity, LDkit creates a SPARQL query based on the schema. The more properties you add to the schema, and the more schemas you nest, the less performant the query.

Now, when you specify a type in the schema, LDkit uses that type to limit the set of solutions in the SPARQL query, hence the improved performance.

Truth be told, SPARQL engines leave a lot to be desired when it comes to performance in general, but hopefully things will get better in the future. In the meantime, if you run into performance issues with LDkit, I recommend trying to split the schema into multiple smaller ones.

I also recommend setting the logQuery option so that you can get a callback with each SPARQL query that LDkit issues (so that you can e.g. log it in the console): https://ldkit.io/docs/components/options