DBGYM_CONFIG_PATH is now only set in one place

.github/workflows/tests.yaml

@@ -52,7 +52,7 @@ jobs:
       # Integration tests do require external systems to be running (most commonly a database instance).
       # Unlike end-to-end tests though, they test a specific module in a detailed manner, much like a unit test does.
       env:
-        # We set `INTENDED_DBDATA_HARDWARE` so that it's seen when `integtest_pg_conn.py` executes `set_up_gymlib_integtest_workspace.sh`.
+        # We set `INTENDED_DBDATA_HARDWARE` so that it's seen when `integtest_pg_conn.py` executes `_set_up_gymlib_integtest_workspace.sh`.
         INTENDED_DBDATA_HARDWARE: ssd
       run: |
         . "$HOME/.cargo/env"

env/tests/set_up_gymlib_integtest_workspace.sh → env/tests/_set_up_gymlib_integtest_workspace.sh

@@ -1,5 +1,10 @@
 #!/bin/bash
 
+# DO NOT RUN THIS SCRIPT DIRECTLY.
+# This script only runs correctly when run by GymlibIntegtestManager.set_up_workspace() as it sets the necessary envvars.
+# By allowing GymlibIntegtestManager.set_up_workspace() to set the envvars, we ensure that the envvars are only defined
+# in a single location (inside GymlibIntegtestManager).
+
 # Gymlib integration tests relies on Postgres being built and workloads/dbdata being generated.
 # Generating these things is not considered a part of the test which is why it's in its own shell script.
 # The reason there's a shell script generating them instead of them just being in the repo is because (a)
@@ -10,8 +15,6 @@ set -euo pipefail
 
 # INTENDED_DBDATA_HARDWARE can be set elsewhere (e.g. by tests_ci.yaml) but we use hdd by default.
 INTENDED_DBDATA_HARDWARE="${INTENDED_DBDATA_HARDWARE:-hdd}"
-export DBGYM_CONFIG_PATH=env/tests/gymlib_integtest_dbgym_config.yaml # Note that this envvar needs to be exported.
-WORKSPACE_PATH=$(grep 'dbgym_workspace_path:' $DBGYM_CONFIG_PATH | sed 's/dbgym_workspace_path: //')
 
 python3 task.py benchmark $BENCHMARK data $SCALE_FACTOR
 python3 task.py benchmark $BENCHMARK workload --scale-factor $SCALE_FACTOR

env/tests/integtest_util.py

@@ -27,28 +27,30 @@ class GymlibIntegtestManager:
     inside a class makes it clear that they are related to the gymlib integration tests.
     """
 
-    # These constants are also used by set_up_gymlib_integtest_workspace.sh.
+    # These constants are also used by _set_up_gymlib_integtest_workspace.sh.
     BENCHMARK = "tpch"
     SCALE_FACTOR = 0.01
-    # TODO: make DBGYM_CONFIG_FPATH an envvar.
-    DBGYM_CONFIG_FPATH = Path("env/tests/gymlib_integtest_dbgym_config.yaml")
+    DBGYM_CONFIG_PATH = Path("env/tests/gymlib_integtest_dbgym_config.yaml")
 
     # This is set at most once by set_up_workspace().
     DBGYM_WORKSPACE: Optional[DBGymWorkspace] = None
 
     @staticmethod
     def set_up_workspace() -> None:
         workspace_path = get_workspace_path_from_config(
-            GymlibIntegtestManager.DBGYM_CONFIG_FPATH
+            GymlibIntegtestManager.DBGYM_CONFIG_PATH
         )
 
         # This if statement prevents us from setting up the workspace twice, which saves time.
         if not workspace_path.exists():
             subprocess.run(
-                ["./env/tests/set_up_gymlib_integtest_workspace.sh"],
+                ["./env/tests/_set_up_gymlib_integtest_workspace.sh"],
                 env={
                     "BENCHMARK": GymlibIntegtestManager.BENCHMARK,
                     "SCALE_FACTOR": str(GymlibIntegtestManager.SCALE_FACTOR),
+                    # By setting this envvar, we ensure that when running _set_up_gymlib_integtest_workspace.sh,
+                    # make_standard_dbgym_workspace() will use the correct DBGYM_CONFIG_PATH.
+                    "DBGYM_CONFIG_PATH": str(GymlibIntegtestManager.DBGYM_CONFIG_PATH),
                     **os.environ,
                 },
                 check=True,

util/workspace.py

@@ -293,6 +293,7 @@ def make_standard_dbgym_workspace() -> DBGymWorkspace:
     default path of dbgym_config.yaml.
     """
     dbgym_config_path = Path(os.getenv("DBGYM_CONFIG_PATH", "dbgym_config.yaml"))
+    assert dbgym_config_path == Path("env/tests/gymlib_integtest_dbgym_config.yaml")
     dbgym_workspace_path = get_workspace_path_from_config(dbgym_config_path)
     dbgym_workspace = DBGymWorkspace(dbgym_workspace_path)
     return dbgym_workspace