made all defaults below workload path fns

Author: wangpatrick57

dbms/postgres/cli.py

@@ -35,10 +35,10 @@
 from util.workspace import (
     WORKSPACE_PATH_PLACEHOLDER,
     DBGymConfig,
-    default_dbdata_parent_dpath,
-    default_pgbin_path,
     fully_resolve_path,
     get_dbdata_tgz_name,
+    get_default_dbdata_parent_dpath,
+    get_default_pgbin_path,
     is_fully_resolved,
     is_ssd,
     link_result,
@@ -78,7 +78,7 @@ def postgres_build(dbgym_cfg: DBGymConfig, rebuild: bool) -> None:
     "--pgbin-path",
     type=Path,
     default=None,
-    help=f"The path to the bin containing Postgres executables. The default is {default_pgbin_path(WORKSPACE_PATH_PLACEHOLDER)}.",
+    help=f"The path to the bin containing Postgres executables. The default is {get_default_pgbin_path(WORKSPACE_PATH_PLACEHOLDER)}.",
 )
 @click.option(
     "--intended-dbdata-hardware",
@@ -90,7 +90,7 @@ def postgres_build(dbgym_cfg: DBGymConfig, rebuild: bool) -> None:
     "--dbdata-parent-dpath",
     default=None,
     type=Path,
-    help=f"The path to the parent directory of the dbdata which will be actively tuned. The default is {default_dbdata_parent_dpath(WORKSPACE_PATH_PLACEHOLDER)}.",
+    help=f"The path to the parent directory of the dbdata which will be actively tuned. The default is {get_default_dbdata_parent_dpath(WORKSPACE_PATH_PLACEHOLDER)}.",
 )
 def postgres_dbdata(
     dbgym_cfg: DBGymConfig,
@@ -102,9 +102,9 @@ def postgres_dbdata(
 ) -> None:
     # Set args to defaults programmatically (do this before doing anything else in the function)
     if pgbin_path is None:
-        pgbin_path = default_pgbin_path(dbgym_cfg.dbgym_workspace_path)
+        pgbin_path = get_default_pgbin_path(dbgym_cfg.dbgym_workspace_path)
     if dbdata_parent_dpath is None:
-        dbdata_parent_dpath = default_dbdata_parent_dpath(
+        dbdata_parent_dpath = get_default_dbdata_parent_dpath(
             dbgym_cfg.dbgym_workspace_path
         )
 

env/integtest_pg_conn.py

@@ -16,9 +16,9 @@
 )
 from util.workspace import (
     DEFAULT_BOOT_CONFIG_FPATH,
-    default_dbdata_parent_dpath,
-    default_pgbin_path,
-    default_pristine_dbdata_snapshot_path,
+    get_default_dbdata_parent_dpath,
+    get_default_pgbin_path,
+    get_default_pristine_dbdata_snapshot_path,
 )
 
 
@@ -33,15 +33,17 @@ def setUp(self) -> None:
             "Make sure Postgres isn't running before starting the integration test. `pkill postgres` is one way "
             + "to ensure this. Be careful about accidentally taking down other people's Postgres instances though.",
         )
-        self.pristine_dbdata_snapshot_path = default_pristine_dbdata_snapshot_path(
+        self.pristine_dbdata_snapshot_path = get_default_pristine_dbdata_snapshot_path(
             IntegtestWorkspace.get_workspace_path(),
             INTEGTEST_BENCHMARK,
             INTEGTEST_SCALE_FACTOR,
         )
-        self.dbdata_parent_dpath = default_dbdata_parent_dpath(
+        self.dbdata_parent_dpath = get_default_dbdata_parent_dpath(
+            IntegtestWorkspace.get_workspace_path()
+        )
+        self.pgbin_dpath = get_default_pgbin_path(
             IntegtestWorkspace.get_workspace_path()
         )
-        self.pgbin_dpath = default_pgbin_path(IntegtestWorkspace.get_workspace_path())
 
         # The reason we restart Postgres every time is to ensure a "clean" starting point
         # so that all tests are independent of each other.

scripts/run_protox_e2e_test.py

@@ -11,14 +11,14 @@
 from benchmark.tpch.constants import DEFAULT_TPCH_SEED
 from util.pg import get_is_postgres_running
 from util.workspace import (
-    default_pristine_dbdata_snapshot_path,
-    default_replay_data_fpath,
-    default_repo_path,
-    default_tuning_steps_dpath,
     get_default_embedder_path,
     get_default_hpoed_agent_params_path,
+    get_default_pristine_dbdata_snapshot_path,
+    get_default_replay_data_fpath,
+    get_default_repo_path,
     get_default_tables_path,
     get_default_traindata_path,
+    get_default_tuning_steps_dpath,
     get_default_workload_path,
     get_workload_name,
 )
@@ -119,13 +119,13 @@ def run_e2e_for_benchmark(benchmark_name: str, intended_dbdata_hardware: str) ->
         )
         assert workload_dpath.exists()
 
-    repo_dpath = default_repo_path(workspace_dpath)
+    repo_dpath = get_default_repo_path(workspace_dpath)
     if Stage.DBRepo in STAGES_TO_RUN:
         assert not repo_dpath.exists()
         subprocess.run(f"python task.py dbms {DBMS} build".split(), check=True)
         assert repo_dpath.exists()
 
-    pristine_dbdata_snapshot_fpath = default_pristine_dbdata_snapshot_path(
+    pristine_dbdata_snapshot_fpath = get_default_pristine_dbdata_snapshot_path(
         workspace_dpath, benchmark_name, scale_factor
     )
     if Stage.DBData in STAGES_TO_RUN:
@@ -170,7 +170,7 @@ def run_e2e_for_benchmark(benchmark_name: str, intended_dbdata_hardware: str) ->
         )
         assert hpoed_agent_params_fpath.exists()
 
-    tuning_steps_dpath = default_tuning_steps_dpath(
+    tuning_steps_dpath = get_default_tuning_steps_dpath(
         workspace_dpath, benchmark_name, workload_name, False
     )
     if Stage.TuneTune in STAGES_TO_RUN:
@@ -182,7 +182,7 @@ def run_e2e_for_benchmark(benchmark_name: str, intended_dbdata_hardware: str) ->
         assert tuning_steps_dpath.exists()
 
     # Post-training (replay)
-    replay_data_fpath = default_replay_data_fpath(
+    replay_data_fpath = get_default_replay_data_fpath(
         workspace_dpath, benchmark_name, workload_name, False
     )
     if Stage.Replay in STAGES_TO_RUN:

tune/demo/main.py

@@ -5,9 +5,9 @@
 from util.workspace import (
     DEFAULT_BOOT_CONFIG_FPATH,
     DBGymConfig,
-    default_dbdata_parent_dpath,
-    default_pgbin_path,
-    default_pristine_dbdata_snapshot_path,
+    get_default_dbdata_parent_dpath,
+    get_default_pgbin_path,
+    get_default_pristine_dbdata_snapshot_path,
     make_standard_dbgym_cfg,
 )
 
@@ -43,13 +43,13 @@ class Demo:
 
     def __init__(self) -> None:
         self.dbgym_cfg = make_dbgym_cfg_cached()
-        self.pristine_dbdata_snapshot_path = default_pristine_dbdata_snapshot_path(
+        self.pristine_dbdata_snapshot_path = get_default_pristine_dbdata_snapshot_path(
             self.dbgym_cfg.dbgym_workspace_path, Demo.BENCHMARK, Demo.SCALE_FACTOR
         )
-        self.dbdata_parent_dpath = default_dbdata_parent_dpath(
+        self.dbdata_parent_dpath = get_default_dbdata_parent_dpath(
             self.dbgym_cfg.dbgym_workspace_path
         )
-        self.pgbin_dpath = default_pgbin_path(self.dbgym_cfg.dbgym_workspace_path)
+        self.pgbin_dpath = get_default_pgbin_path(self.dbgym_cfg.dbgym_workspace_path)
         self.pg_conn = PostgresConn(
             self.dbgym_cfg,
             DEFAULT_POSTGRES_PORT,

tune/protox/agent/hpo.py

@@ -35,14 +35,14 @@
     WORKSPACE_PATH_PLACEHOLDER,
     DBGymConfig,
     TuningMode,
-    default_dbdata_parent_dpath,
-    default_pgbin_path,
-    default_pristine_dbdata_snapshot_path,
     fully_resolve_path,
     get_default_benchbase_config_path,
     get_default_benchmark_config_path,
+    get_default_dbdata_parent_dpath,
     get_default_embedder_path,
     get_default_hpoed_agent_params_fname,
+    get_default_pgbin_path,
+    get_default_pristine_dbdata_snapshot_path,
     get_default_workload_name_suffix,
     get_default_workload_path,
     get_workload_name,
@@ -144,7 +144,7 @@ def __init__(
     "--pristine-dbdata-snapshot-path",
     type=Path,
     default=None,
-    help=f"The path to the .tgz snapshot of the dbdata directory to use as a starting point for tuning. The default is {default_pristine_dbdata_snapshot_path(WORKSPACE_PATH_PLACEHOLDER, BENCHMARK_NAME_PLACEHOLDER, SCALE_FACTOR_PLACEHOLDER)}.",
+    help=f"The path to the .tgz snapshot of the dbdata directory to use as a starting point for tuning. The default is {get_default_pristine_dbdata_snapshot_path(WORKSPACE_PATH_PLACEHOLDER, BENCHMARK_NAME_PLACEHOLDER, SCALE_FACTOR_PLACEHOLDER)}.",
 )
 @click.option(
     "--intended-dbdata-hardware",
@@ -156,13 +156,13 @@ def __init__(
     "--dbdata-parent-dpath",
     type=Path,
     default=None,
-    help=f"The path to the parent directory of the dbdata which will be actively tuned. The default is {default_dbdata_parent_dpath(WORKSPACE_PATH_PLACEHOLDER)}.",
+    help=f"The path to the parent directory of the dbdata which will be actively tuned. The default is {get_default_dbdata_parent_dpath(WORKSPACE_PATH_PLACEHOLDER)}.",
 )
 @click.option(
     "--pgbin-path",
     type=Path,
     default=None,
-    help=f"The path to the bin containing Postgres executables. The default is {default_pgbin_path(WORKSPACE_PATH_PLACEHOLDER)}.",
+    help=f"The path to the bin containing Postgres executables. The default is {get_default_pgbin_path(WORKSPACE_PATH_PLACEHOLDER)}.",
 )
 @click.option(
     "--workload-path",
@@ -278,15 +278,15 @@ def hpo(
     if benchbase_config_path is None:
         benchbase_config_path = get_default_benchbase_config_path(benchmark_name)
     if pristine_dbdata_snapshot_path is None:
-        pristine_dbdata_snapshot_path = default_pristine_dbdata_snapshot_path(
+        pristine_dbdata_snapshot_path = get_default_pristine_dbdata_snapshot_path(
             dbgym_cfg.dbgym_workspace_path, benchmark_name, scale_factor
         )
     if dbdata_parent_dpath is None:
-        dbdata_parent_dpath = default_dbdata_parent_dpath(
+        dbdata_parent_dpath = get_default_dbdata_parent_dpath(
             dbgym_cfg.dbgym_workspace_path
         )
     if pgbin_path is None:
-        pgbin_path = default_pgbin_path(dbgym_cfg.dbgym_workspace_path)
+        pgbin_path = get_default_pgbin_path(dbgym_cfg.dbgym_workspace_path)
     if workload_path is None:
         workload_path = get_default_workload_path(
             dbgym_cfg.dbgym_workspace_path, benchmark_name, workload_name

tune/protox/agent/replay.py

@@ -31,9 +31,9 @@
 from util.workspace import (
     DBGymConfig,
     TuningMode,
-    default_tuning_steps_dpath,
     fully_resolve_path,
     get_default_replay_data_fname,
+    get_default_tuning_steps_dpath,
     get_default_workload_name_suffix,
     get_workload_name,
     link_result,
@@ -143,7 +143,7 @@ def replay(
     workload_name = get_workload_name(scale_factor, workload_name_suffix)
 
     if tuning_steps_dpath is None:
-        tuning_steps_dpath = default_tuning_steps_dpath(
+        tuning_steps_dpath = get_default_tuning_steps_dpath(
             dbgym_cfg.dbgym_workspace_path,
             benchmark_name,
             workload_name,

tune/protox/embedding/datagen.py

@@ -38,11 +38,11 @@
     WORKLOAD_NAME_PLACEHOLDER,
     WORKSPACE_PATH_PLACEHOLDER,
     DBGymConfig,
-    default_dbdata_parent_dpath,
-    default_pgbin_path,
-    default_pristine_dbdata_snapshot_path,
     fully_resolve_path,
     get_default_benchmark_config_path,
+    get_default_dbdata_parent_dpath,
+    get_default_pgbin_path,
+    get_default_pristine_dbdata_snapshot_path,
     get_default_traindata_fname,
     get_default_workload_name_suffix,
     get_default_workload_path,
@@ -89,14 +89,14 @@
     "--pgbin-path",
     type=Path,
     default=None,
-    help=f"The path to the bin containing Postgres executables. The default is {default_pgbin_path(WORKSPACE_PATH_PLACEHOLDER)}.",
+    help=f"The path to the bin containing Postgres executables. The default is {get_default_pgbin_path(WORKSPACE_PATH_PLACEHOLDER)}.",
 )
 # TODO(phw2): need to run pgtune before gathering data
 @click.option(
     "--pristine-dbdata-snapshot-path",
     type=Path,
     default=None,
-    help=f"The path to the .tgz snapshot of the dbdata directory to build an embedding space over. The default is {default_pristine_dbdata_snapshot_path(WORKSPACE_PATH_PLACEHOLDER, BENCHMARK_NAME_PLACEHOLDER, SCALE_FACTOR_PLACEHOLDER)}.",
+    help=f"The path to the .tgz snapshot of the dbdata directory to build an embedding space over. The default is {get_default_pristine_dbdata_snapshot_path(WORKSPACE_PATH_PLACEHOLDER, BENCHMARK_NAME_PLACEHOLDER, SCALE_FACTOR_PLACEHOLDER)}.",
 )
 @click.option(
     "--intended-dbdata-hardware",
@@ -108,7 +108,7 @@
     "--dbdata-parent-dpath",
     type=Path,
     default=None,
-    help=f"The path to the parent directory of the dbdata which will be actively tuned. The default is {default_pristine_dbdata_snapshot_path(WORKSPACE_PATH_PLACEHOLDER, BENCHMARK_NAME_PLACEHOLDER, SCALE_FACTOR_PLACEHOLDER)}.",
+    help=f"The path to the parent directory of the dbdata which will be actively tuned. The default is {get_default_pristine_dbdata_snapshot_path(WORKSPACE_PATH_PLACEHOLDER, BENCHMARK_NAME_PLACEHOLDER, SCALE_FACTOR_PLACEHOLDER)}.",
 )
 @click.option(
     "--benchmark-config-path",
@@ -211,13 +211,13 @@ def datagen(
     if benchmark_config_path is None:
         benchmark_config_path = get_default_benchmark_config_path(benchmark_name)
     if pgbin_path is None:
-        pgbin_path = default_pgbin_path(dbgym_cfg.dbgym_workspace_path)
+        pgbin_path = get_default_pgbin_path(dbgym_cfg.dbgym_workspace_path)
     if pristine_dbdata_snapshot_path is None:
-        pristine_dbdata_snapshot_path = default_pristine_dbdata_snapshot_path(
+        pristine_dbdata_snapshot_path = get_default_pristine_dbdata_snapshot_path(
             dbgym_cfg.dbgym_workspace_path, benchmark_name, scale_factor
         )
     if dbdata_parent_dpath is None:
-        dbdata_parent_dpath = default_dbdata_parent_dpath(
+        dbdata_parent_dpath = get_default_dbdata_parent_dpath(
             dbgym_cfg.dbgym_workspace_path
         )
     if max_concurrent is None:
@@ -294,7 +294,7 @@ def datagen(
         generic_args.dbdata_parent_dpath,
     )
     pgbin_path = fully_resolve_path(
-        dbgym_cfg, default_pgbin_path(dbgym_cfg.dbgym_workspace_path)
+        dbgym_cfg, get_default_pgbin_path(dbgym_cfg.dbgym_workspace_path)
     )
     start_postgres(dbgym_cfg, pgbin_path, dbdata_dpath)
     _gen_traindata_dpath(dbgym_cfg, generic_args, dir_gen_args)