feat: allow input_types, output_type, and dataset to be used positionally in remote_function (#1560)

Author: tswast

packages/bigframes/bigframes/pandas/__init__.py

@@ -65,10 +65,14 @@
 
 
 def remote_function(
-    *,
+    # Make sure that the input/output types, and dataset can be used
+    # positionally. This avoids the worst of the breaking change from 1.x to
+    # 2.x while still preventing possible mixups between consecutive str
+    # parameters.
     input_types: Union[None, type, Sequence[type]] = None,
     output_type: Optional[type] = None,
     dataset: Optional[str] = None,
+    *,
     bigquery_connection: Optional[str] = None,
     reuse: bool = True,
     name: Optional[str] = None,

packages/bigframes/bigframes/session/__init__.py

@@ -1202,10 +1202,14 @@ def _check_file_size(self, filepath: str):
 
     def remote_function(
         self,
-        *,
+        # Make sure that the input/output types, and dataset can be used
+        # positionally. This avoids the worst of the breaking change from 1.x to
+        # 2.x while still preventing possible mixups between consecutive str
+        # parameters.
         input_types: Union[None, type, Sequence[type]] = None,
         output_type: Optional[type] = None,
         dataset: Optional[str] = None,
+        *,
         bigquery_connection: Optional[str] = None,
         reuse: bool = True,
         name: Optional[str] = None,

packages/bigframes/tests/system/large/functions/test_remote_function.py

@@ -109,9 +109,11 @@ def test_remote_function_multiply_with_ibis(
     try:
 
         @session.remote_function(
-            input_types=[int, int],
-            output_type=int,
-            dataset=dataset_id,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            [int, int],
+            int,
+            dataset_id,
             bigquery_connection=bq_cf_connection,
             reuse=False,
             cloud_function_service_account="default",
@@ -164,9 +166,11 @@ def test_remote_function_stringify_with_ibis(
     try:
 
         @session.remote_function(
-            input_types=[int],
-            output_type=str,
-            dataset=dataset_id,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            [int],
+            str,
+            dataset_id,
             bigquery_connection=bq_cf_connection,
             reuse=False,
             cloud_function_service_account="default",
@@ -213,9 +217,11 @@ def func(x, y):
             return x * abs(y % 4)
 
         remote_func = session.remote_function(
-            input_types=[str, int],
-            output_type=str,
-            dataset=dataset_id,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            [str, int],
+            str,
+            dataset_id,
             bigquery_connection=bq_cf_connection,
             reuse=False,
             cloud_function_service_account="default",
@@ -251,9 +257,11 @@ def func(x, y):
             return [len(x), abs(y % 4)]
 
         remote_func = session.remote_function(
-            input_types=[str, int],
-            output_type=list[int],
-            dataset=dataset_id,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            [str, int],
+            list[int],
+            dataset_id,
             bigquery_connection=bq_cf_connection,
             reuse=False,
             cloud_function_service_account="default",
@@ -286,9 +294,11 @@ def test_remote_function_decorator_with_bigframes_series(
     try:
 
         @session.remote_function(
-            input_types=[int],
-            output_type=int,
-            dataset=dataset_id,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            [int],
+            int,
+            dataset_id,
             bigquery_connection=bq_cf_connection,
             reuse=False,
             cloud_function_service_account="default",
@@ -333,9 +343,11 @@ def add_one(x):
             return x + 1
 
         remote_add_one = session.remote_function(
-            input_types=[int],
-            output_type=int,
-            dataset=dataset_id,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            [int],
+            int,
+            dataset_id,
             bigquery_connection=bq_cf_connection,
             reuse=False,
             cloud_function_service_account="default",
@@ -385,8 +397,10 @@ def add_one(x):
             return x + 1
 
         remote_add_one = session.remote_function(
-            input_types=input_types,
-            output_type=int,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            input_types,
+            int,
             reuse=False,
             cloud_function_service_account="default",
         )(add_one)
@@ -415,8 +429,10 @@ def test_remote_function_explicit_dataset_not_created(
     try:
 
         @session.remote_function(
-            input_types=[int],
-            output_type=int,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            [int],
+            int,
             dataset=dataset_id_not_created,
             bigquery_connection=bq_cf_connection,
             reuse=False,
@@ -469,9 +485,11 @@ def sign(num):
             return NO_SIGN
 
         remote_sign = session.remote_function(
-            input_types=[int],
-            output_type=int,
-            dataset=dataset_id,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            [int],
+            int,
+            dataset_id,
             bigquery_connection=bq_cf_connection,
             reuse=False,
             cloud_function_service_account="default",
@@ -517,9 +535,11 @@ def circumference(radius):
             return 2 * mymath.pi * radius
 
         remote_circumference = session.remote_function(
-            input_types=[float],
-            output_type=float,
-            dataset=dataset_id,
+            # Make sure that the input/output types can be used positionally.
+            # This avoids the worst of the breaking change from 1.x to 2.x.
+            [float],
+            float,
+            dataset_id,
             bigquery_connection=bq_cf_connection,
             reuse=False,
             cloud_function_service_account="default",