fix query order per spec

src/lakebench/benchmarks/_load_and_query/_load_and_query.py

@@ -230,7 +230,11 @@ def __init__(
         self.engine.extended_engine_metadata["analyze"] = analyze_mode
 
         self._power_test_uses_full_stream = query_list is None or query_list == ["*"]
-        if query_list is not None:
+        if self._power_test_uses_full_stream and self.QUERY_STREAMS:
+            query_plan = self._query_plan_for_stream(self.POWER_TEST_STREAM)
+            self.query_progress = [progress for progress, _ in query_plan]
+            self.query_list = [query_name for _, query_name in query_plan]
+        elif query_list is not None:
             expanded_query_list = []
             for query in query_list:
                 if query == "*":
@@ -244,9 +248,10 @@ def __init__(
                     f"Query list contains unsupported queries: {unsupported_queries}. Supported queries: {self.QUERY_REGISTRY}."
                 )
             self.query_list = expanded_query_list
+            self.query_progress = None
         else:
             self.query_list = self.QUERY_REGISTRY
-        self.query_progress = None
+            self.query_progress = None
 
         for base_engine, benchmark_impl in self.BENCHMARK_IMPL_REGISTRY.items():
             if isinstance(engine, base_engine):

src/lakebench/benchmarks/tpcds/tpcds.py

@@ -21,7 +21,7 @@ class TPCDS(_LoadAndQuery):
     scenario_name : str
         The name of the benchmark scenario.
     query_list : list of str, optional
-        List of queries to execute. Use '*' for all queries. If not specified, all queries will be run.
+        List of queries to execute. Use '*' or omit this parameter to run query stream 0.
     input_parquet_folder_uri : str, optional
         Path to the input parquet files. Must be the root directory containing a folder named after
         each table in TABLE_REGISTRY.
@@ -37,9 +37,9 @@ class TPCDS(_LoadAndQuery):
         Runs the benchmark in the specified mode.
         Supported modes are:
             - 'load': Sequentially executes loading the 24 tables.
-            - 'query': Sequentially executes the 99 queries.
+            - 'query': Executes query stream 0 by default, or the explicitly requested query list.
             - 'power_test': Executes query stream 0 without loading data.
-            - 'load_and_query': Executes the load test followed by the query test.
+            - 'load_and_query': Executes the load test followed by the default stream or requested query list.
     _run_load_test()
         Loads the data for the benchmark.
     _run_query_test()

src/lakebench/benchmarks/tpch/tpch.py

@@ -21,7 +21,7 @@ class TPCH(_LoadAndQuery):
     scenario_name : str
         The name of the benchmark scenario.
     query_list : list of str, optional
-        List of queries to execute. Use '*' for all queries. If not specified, all queries will be run.
+        List of queries to execute. Use '*' or omit this parameter to run query stream 0.
     input_parquet_folder_uri : str, optional
         Path to the input parquet files. Must be the root directory containing a folder named after
         each table in TABLE_REGISTRY.

tests/test_query_streams.py

@@ -7,6 +7,22 @@
 from lakebench.benchmarks._load_and_query import _LoadAndQuery
 from lakebench.benchmarks.tpcds import TPCDS
 from lakebench.benchmarks.tpch import TPCH
+from lakebench.engines.duckdb import DuckDB
+
+
+def _benchmark_with_queries(benchmark_class, query_list):
+    engine = MagicMock(spec=DuckDB)
+    engine.SUPPORTS_MOUNT_PATH = True
+    engine.version = "test"
+    engine.extended_engine_metadata = {}
+    engine.get_total_cores.return_value = 1
+    engine.get_compute_size.return_value = "test"
+    return benchmark_class(
+        engine=engine,
+        scenario_name="test",
+        query_list=query_list,
+        input_parquet_folder_uri="/tmp/data",
+    )
 
 
 def test_power_test_runs_queries_without_loading():
@@ -50,6 +66,16 @@ def test_query_streams_are_complete_permutations(benchmark_class, stream_count,
     assert all(set(stream) == set(range(1, query_count + 1)) for stream in benchmark_class.QUERY_STREAMS)
 
 
+@pytest.mark.parametrize("benchmark_class", [TPCDS, TPCH])
+@pytest.mark.parametrize("query_list", [None, ["*"]])
+def test_full_query_shortcuts_default_to_stream_zero(benchmark_class, query_list):
+    benchmark = _benchmark_with_queries(benchmark_class, query_list)
+    query_plan = benchmark_class._query_plan_for_stream(0)
+
+    assert benchmark.query_list == [query_name for _, query_name in query_plan]
+    assert benchmark.query_progress == [progress for progress, _ in query_plan]
+
+
 def test_tpcds_power_test_uses_stream_zero_order():
     benchmark = object.__new__(TPCDS)
     benchmark.query_list = ["q1"]