fix(pyspark): fail when aggregation contains a having filter

ibis/backends/pyspark/compiler.py

@@ -388,6 +388,13 @@ def compile_literal(t, op, *, raw=False, **kwargs):
 def compile_aggregation(t, op, **kwargs):
     src_table = t.translate(op.table, **kwargs)
 
+    if op.having:
+        raise com.UnsupportedOperationError(
+            "The PySpark backend does not support `having` because the underlying "
+            "PySpark API does not support it. Use a filter on the aggregation "
+            "expression instead."
+        )
+
     if op.predicates:
         predicate = functools.reduce(ops.And, op.predicates)
         src_table = src_table.filter(t.translate(predicate, **kwargs))

ibis/backends/tests/test_generic.py

@@ -1125,6 +1125,11 @@ def test_distinct_on_keep(backend, on, keep):
     raises=com.OperationNotDefinedError,
     reason="backend doesn't implement ops.WindowFunction",
 )
+@pytest.mark.notimpl(
+    ["pyspark"],
+    raises=com.UnsupportedOperationError,
+    reason="backend doesn't support `having` filters",
+)
 def test_distinct_on_keep_is_none(backend, on):
     from ibis import _