smith6jt-cop Jupyter Autoreload Awareness
Experiment Overview
| Item | Details |
|---|---|
| Date | 2025-01-14 |
| Goal | Stop telling users to restart kernels or reload notebooks unnecessarily |
| Environment | Jupyter notebooks with autoreload extension |
| Status | Critical workflow improvement |
Context
KINTSUGI notebooks use IPython's autoreload extension:
%load_ext autoreload
%autoreload 2
With %autoreload 2, all imported modules are automatically reloaded before executing any code. This means:
- Edits to
.pyfiles (Kio.py, modules in Kdecon/, Kstitch/, etc.) take effect immediately - No kernel restart needed
- No notebook reload needed
- Just re-run the cell that imports/uses the module
NEVER Say These Things
| Bad Advice | Why It's Wrong |
|---|---|
| "Restart your Jupyter kernel" | Autoreload handles module updates automatically |
| "Reload the notebook" | Wastes time, loses output/state, unnecessary |
| "Close and reopen the notebook" | Even worse - complete productivity killer |
| "You may need to restart..." | Creates doubt; autoreload makes this unnecessary |
Correct Guidance
After editing a Python module that a notebook imports:
CORRECT: "Re-run the cell" or just confirm the edit was made WRONG: "Restart kernel and reload notebook"
When Kernel Restart IS Needed
Only tell user to restart kernel when:
- Installing new packages (
pip install) - Changing environment variables that were read at import time
- C extension modules were recompiled
- Autoreload is NOT enabled in the notebook
- User explicitly asks to restart
Trigger Conditions
This skill applies when:
- About to tell user to restart Jupyter kernel
- About to tell user to reload/reopen notebook
- After editing .py files imported by notebook (Kio.py, kintsugi/*.py, etc.)
- User is working in a Jupyter notebook environment
Key Insight
Telling users to restart kernels or reload notebooks lowers productivity significantly. With autoreload enabled, the only action needed after editing imported modules is to re-run the relevant cell.