Improved errors

[andreasjansson]

Important

Add prediction_index to improve error logging and tracking in prediction tasks across multiple modules.

• Error Handling:
  • Add prediction_index to cog_safe_push() in main.py to track prediction tasks.
  • Modify run_tasks() in main.py to log errors with prediction_index.
  • Update predict() in predict.py to include prediction_index in logs.
• Task Classes:
  • Add prediction_index attribute to CheckOutputsMatch, RunTestCase, and FuzzModel in tasks.py.
  • Log prediction_index in run() methods of these classes for better error tracking.
• Misc:
  • Update logging statements across main.py, predict.py, and tasks.py to include prediction_index for improved error context.

^{This description was created by for fa8a261. You can customize this summary. It will automatically update as commits are pushed.}

[Copilot]

Pull Request Overview

Adds a prediction_index to task runs and predictions so that logs and error messages are prefixed with an index for easier tracing.

• Introduce prediction_index on Task subclasses and propagate it into predict calls.
• Prefix verbose logs and raised errors with the index when supplied.
• Update main.py to assign and increment prediction_index for each task and adjust error handling to report indices.

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
cog_safe_push/tasks.py	Added prediction_index field; prefixed log.v and raise calls with index
cog_safe_push/predict.py	Extended predict signature to accept prediction_index and prefix logs
cog_safe_push/main.py	Initialized/incremented prediction_index for tasks and updated error aggregation to include indices

Comments suppressed due to low confidence (2)

cog_safe_push/tasks.py:177

• This if error is not None block is redundant and unreachable because any non-None error would have triggered an earlier check; consider removing this duplicate error check.

if error is not None:

cog_safe_push/predict.py:236

• [nitpick] There are no tests verifying that prediction_index is correctly propagated and prefixed in logs; consider adding unit tests to assert the behavior when prediction_index is set and unset.

prefix = f"[{prediction_index}] " if prediction_index is not None else ""