revert to old fix because fictitious tasks spawn with array[0:]

ikrommyd · Oct 24, 2024 · 2705b0b · 2705b0b
1 parent 1e1c4a1
commit 2705b0b
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/scripts/run_analysis.py b/scripts/run_analysis.py
@@ -223,7 +223,7 @@ def main():
     if args.print_necessary_columns:
         import dask_awkward as dak
 
-        necessary_columns = dak.neccessary_columns(to_compute)
+        necessary_columns = dak.necessary_columns(to_compute)
         logger.info(f"The necessary columns are:\n{necessary_columns}")
     logger.info("Computing the task graph")
     if client:

diff --git a/src/egamma_tnp/utils/runner_utils.py b/src/egamma_tnp/utils/runner_utils.py
@@ -188,7 +188,7 @@ def save_array_to_parquet(array, output_dir, dataset, subdir, prefix=None, repar
 
     # Trick to reduce node multiplicity before to_parquet until it's fixed.
     # TODO: remove this when the issue is fixed
-    array = array[0:]
+    array = array[array.run > -999.0]
 
     # Repartition the array if needed
     if repartition_n: