Inspired by the holds method of https://github.com/linkeddata/rdflib.js I figured we should have something similar in Attean, so I attempted this default implementation, which would work under the assumption that the count estimate would be at least positive if there is one match. Perhaps that assumption is something we should spec.
However, the test fails for me, and I don't understand why. Figured I'd let you have a look anyway.
Inspired by the
holds
method of https://github.com/linkeddata/rdflib.js I figured we should have something similar in Attean, so I attempted this default implementation, which would work under the assumption that the count estimate would be at least positive if there is one match. Perhaps that assumption is something we should spec.However, the test fails for me, and I don't understand why. Figured I'd let you have a look anyway.