13
13
)
14
14
15
15
JOB_TABLES_URL = "https://event.cwi.nl/da/job/imdb.tgz"
16
+ JOB_QUERIES_URL = "https://event.cwi.nl/da/job/job.tgz"
16
17
JOB_QUERY_NAMES = [
17
18
"1a" ,
18
19
"1b" ,
@@ -159,7 +160,7 @@ def job_workload(
159
160
dbgym_cfg : DBGymConfig , query_subset : str , scale_factor : float
160
161
) -> None :
161
162
assert scale_factor == DEFAULT_SCALE_FACTOR
162
- _clone_job_queries (dbgym_cfg )
163
+ _download_job_queries (dbgym_cfg )
163
164
_generate_job_workload (dbgym_cfg , query_subset )
164
165
165
166
@@ -172,24 +173,8 @@ def _download_job_data(dbgym_cfg: DBGymConfig) -> None:
172
173
)
173
174
174
175
175
- def _clone_job_queries (dbgym_cfg : DBGymConfig ) -> None :
176
- expected_symlink_dpath = (
177
- dbgym_cfg .cur_symlinks_build_path (mkdir = True ) / "job-queries.link"
178
- )
179
- if expected_symlink_dpath .exists ():
180
- logging .getLogger (DBGYM_LOGGER_NAME ).info (
181
- f"Skipping clone: { expected_symlink_dpath } "
182
- )
183
- return
184
-
185
- logging .getLogger (DBGYM_LOGGER_NAME ).info (f"Cloning: { expected_symlink_dpath } " )
186
- real_build_path = dbgym_cfg .cur_task_runs_build_path (mkdir = True )
187
- subprocess_run (
188
- f"./clone_job_queries.sh { real_build_path } " , cwd = dbgym_cfg .cur_source_path ()
189
- )
190
- symlink_dpath = link_result (dbgym_cfg , real_build_path / "job-queries" )
191
- assert expected_symlink_dpath .samefile (symlink_dpath )
192
- logging .getLogger (DBGYM_LOGGER_NAME ).info (f"Cloned: { expected_symlink_dpath } " )
176
+ def _download_job_queries (dbgym_cfg : DBGymConfig ) -> None :
177
+ _download_and_untar_dir (dbgym_cfg , JOB_QUERIES_URL , "job.tgz" , "job-queries" )
193
178
194
179
195
180
def _download_and_untar_dir (
0 commit comments