enh: robuster NFS lock for download of DIPY data on GHA

Author: oesteban

test/test_motion_viz.py

@@ -22,6 +22,7 @@
 #
 
 from pathlib import Path
+from time import sleep
 
 import numpy as np
 import pandas as pd
@@ -39,9 +40,26 @@
     plot_volumewise_motion,
 )
 
-if not (Path.home() / ".dipy" / "stanford_hardi").exists():
+# Folders are atomic operations in NFS, therefore, creating folders is a good
+# way to implement rudimentary semaphores/locks.
+# OE: we are running into race conditions for the way dipy downloads data.
+# The delayed creation of the "stanford_hardi" directory makes this test
+# for data existence very flaky.
+# Replacing with a pure lock based on directory creation.
+dipy_datapath = Path.home() / ".dipy"
+if not (lock_folder := dipy_datapath / ".lock-stanford_hardi").exists():
+    lock_folder.mkdir(parents=True)
     fetch_stanford_hardi()
 
+MAX_CHECKS = 20
+try_num = 1
+while not (dipy_datapath / "stanford_hardi").exists():
+    sleep(5 * try_num)
+    try_num += 1
+
+    if try_num > MAX_CHECKS:
+        raise RuntimeError("Error downloading DIPY's stanford_hardi dataset")
+
 img, _ = read_stanford_hardi()
 img_data = img.get_fdata()