Tighten workflow-failure handling comments

Author: Ghesselink

backend/apps/ifc_validation/tasks/task_runner.py

@@ -158,15 +158,13 @@ def on_workflow_failed(self, *args, **kwargs):
     id = args[1]
     request = ValidationRequest.objects.get(pk=id)
 
-    # Idempotency (IVS-750): both error_handler (link_error) and chord_error_handler
-    # (chord on_error) can fire for the same failure. If already finalized, do nothing
-    # so we don't re-send the failure emails.
+    # Both error callbacks can fire for one failure; skip if already finalized so the
+    # failure emails aren't sent twice.
     if request.status == ValidationRequest.Status.FAILED:
         logger.debug(f"Request {id} already marked FAILED; skipping duplicate failure handling.")
         return
 
-    # IVS-750: record a clear, human-readable reason instead of dumping raw celery
-    # args/kwargs into status_reason. Full detail stays in the logs (logger.debug above).
+    # Record a readable reason rather than the raw celery args/kwargs (full detail is logged above).
     exc = next((a for a in args if isinstance(a, BaseException)), None)
     reason = "Validation failed before any check could run (workflow-level error)."
     if exc:

backend/apps/ifc_validation/tests/tests_orphan_failed_request.py

@@ -1,16 +1,13 @@
 """
-IVS-750 "failing request without relating task" -- option B (the minimal/elegant fix).
-
-This variant does NOT fabricate a synthetic task. Instead it fixes the two real
-defects in on_workflow_failed:
-  1. it could fire twice (error_handler via link_error AND chord_error_handler via
-     chord on_error), double-sending the failure emails -> add an idempotency guard;
-  2. it dumped raw celery args/kwargs into status_reason -> replace with a clean,
-     human-readable reason (full detail stays in the logs).
-
-Note: under option B the orphaned request still has NO ValidationTask
-(request.tasks.count() == 0). That is intentional for B -- it explains the failure
-on the request itself rather than inventing a task that never ran.
+Workflow-level failure handling in on_workflow_failed.
+
+Fixes two defects without fabricating a synthetic task:
+  1. it could fire twice (link_error + chord on_error), double-sending failure
+     emails -> idempotency guard;
+  2. it dumped raw celery args/kwargs into status_reason -> readable reason instead.
+
+The failed request keeps zero tasks by design: the failure is explained on the
+request itself rather than by inventing a task that never ran.
 """
 
 from unittest.mock import patch