Question regarding `NodeRef`s about codepropertygraph

[mpollmeier]

The public api is designed so you'll do the right thing by default. In other words, you'll get NodeRefs, because a Call is a NodeRef:

joern> val call = cpg.call.head
call: Call = Call(
  id -> 350254634311902973L,
  argumentIndex -> 7,
  argumentName -> None,
  code -> "pat == null",
  columnNumber -> None,
  dispatchType -> "STATIC_DISPATCH",
  dynamicTypeHintFullName -> ArraySeq(),
  lineNumber -> None,
  methodFullName -> "<operator>.equals",
  name -> "<operator>.equals",
  order -> 7,
  signature -> "",
  typeFullName -> "<empty>"
)

joern> call.isInstanceOf[NodeRef[_]]
res6: Boolean = true

joern> call.isInstanceOf[NodeDb]
res7: Boolean = false

You can pass them around and store them in your maps/lists/whereever, they are just a thin reference to the actual data. What you should not do is hold onto the NodeDb instance, which you can get via call.get. On low memory, overflowdb drops the on-heap instance to clear some space, and will use the id from the NodeRef instance to load it back from disk if required.

The domain specific nodes and edges are generated based on the schema. To see the sources, simply run sbt compile and have a look in schema/target/scala-3.1.0/src_managed/main/overflowdb-codegen/io/shiftleft/codepropertygraph/generated

[mal-tee]

That makes sense, thank you very much!