Aggregation is awkward. It is not that composable. Sub-queries are not a thing. Prolog conventions make things awkward. The ammount of datatypes seems to be limited.

If you encounter datalog in the wild it will be something pretty-custom, like datomic, pydatalog, or even differential dataflow (that feels more like map-reduce with fix-points?)

Was going to write a comment about datalog but will piggy back on yours. Every query language has felt like garbage to me until discovering datalog. So simple I explained it to my non technical business partner and he grasped how to query things within five minutes.