feat: improve error message on unequal schemas during set ops

ibis/expr/operations/relations.py

@@ -288,10 +288,24 @@ class Set(Relation):
     values = FrozenOrderedDict()
 
     def __init__(self, left, right, **kwargs):
-        # convert to dictionary first, to get key-unordered comparison semantics
-        if dict(left.schema) != dict(right.schema):
-            raise RelationError("Table schemas must be equal for set operations")
-        elif left.schema.names != right.schema.names:
+        err_msg = "Table schemas must be equal for set operations."
+        try:
+            missing_from_left = right.schema - left.schema
+            missing_from_right = left.schema - right.schema
+        except ValueError as e:
+            if "Conflicting values for key" in str(e):
+                raise RelationError(err_msg) from e
+            else:
+                raise
+        if missing_from_left or missing_from_right:
+            msgs = [err_msg]
+            if missing_from_left:
+                msgs.append(f"Columns missing from the left:\n{missing_from_left}.")
+            if missing_from_right:
+                msgs.append(f"Columns missing from the right:\n{missing_from_right}.")
+            raise RelationError("\n".join(msgs))
+
+        if left.schema.names != right.schema.names:
             # rewrite so that both sides have the columns in the same order making it
             # easier for the backends to implement set operations
             cols = {name: Field(right, name) for name in left.schema.names}

ibis/expr/schema.py

@@ -67,7 +67,7 @@ def _name_locs(self) -> dict[str, int]:
         return {v: i for i, v in enumerate(self.names)}
 
     def equals(self, other: Schema) -> bool:
-        """Return whether `other` is equal to `self`.
+        """Return whether `other` is equal to `self`. Order matters.
 
         Parameters
         ----------
@@ -77,12 +77,13 @@ def equals(self, other: Schema) -> bool:
         Examples
         --------
         >>> import ibis
-        >>> first = ibis.schema({"a": "int"})
-        >>> second = ibis.schema({"a": "int"})
-        >>> assert first.equals(second)
-        >>> third = ibis.schema({"a": "array<int>"})
-        >>> assert not first.equals(third)
-
+        >>> xy = ibis.schema({"x": int, "y": str})
+        >>> xy2 = ibis.schema({"x": int, "y": str})
+        >>> yx = ibis.schema({"y": str, "x": int})
+        >>> xy_float = ibis.schema({"x": float, "y": str})
+        >>> assert xy.equals(xy2)
+        >>> assert not xy.equals(yx)
+        >>> assert not xy.equals(xy_float)
         """
         if not isinstance(other, Schema):
             raise TypeError(