LESSONS.md

Lessons

Table of Contents

1. 📦 Git LFS (Large File Storage)
2. 🌊 Marimo: Code Display Behavior

1. 📦 Git LFS (Large File Storage)

Git LFS (Large File Storage) is an extension for Git that allows you to track and version large files (e.g. .db, .csv, .model) without bloating your repository. Instead of storing the actual file content in Git, it stores a lightweight pointer and pushes the real content to a separate LFS server.

---

How to Install

sudo apt install git-lfs

After installation, initialize LFS:

git lfs install

---

How to Use (Basic Example)

1. Track the file type or name:

   git lfs track "olist.db"
   git lfs track "*.png"
   git lfs track "*.jpg"
   git lfs track "*.csv"
   
   # This adds patterns to a .gitattributes file. For exampple :
   # *.png filter=lfs diff=lfs merge=lfs -text
   

2. **Edit .gitignore **

   # Ignore all database files
   *.db
   # But include this one (make sure it's tracked with git LFS)
   !olist.db
   

3. Stage the tracking file (.gitattributes) and your large file:

   git add .gitattributes
   git add olist.db
   git commit -m "Track olist.db with Git LFS"
   

4. Push to your remote repo:

   git push origin main
   

5. (Optional) Fix Mistakes If You Added a Binary File the Wrong Way

   If you accidentally committed a binary file without LFS, you’ll need to:

   # Remove the file from Git history
   git lfs track "*.png"                     # if not already
   git rm --cached public/erd-schema.png     # remove from Git index
   git add public/erd-schema.png             # re-add to index, now tracked via LFS
   git commit -m "Re-add image using Git LFS"
   
   # Rewrite history to remove the original large file
   git filter-repo --path public/erd-schema.png --invert-paths
   git push --force origin main
   

2. 🌊 Marimo: Code Display Behavior

By default, Marimo hides the code cells in the deployed (read-only) version of your app, especially when running as a Hugging Face Space or a .py script.

This behavior is intentional and aligns with Marimo’s goal of turning Python notebooks into polished, interactive data apps — focusing on the results and UI rather than code.

Environment	Notes
Local Dev Mode (marimo run app.py)	Full interactivity + visible code
Deployed App (e.g., HF Spaces)	Only outputs / UI components shown

If you want to show the code:

You can explicitly render code using mo.ui.code() or mo.md() with fenced code blocks:

code = '''
revenue_by_month_year = query_results[QueryEnum.REVENUE_BY_MONTH_YEAR.value]
revenue_by_month_year
'''

mo.ui.code(code, language="python")

Or in Markdown:

mo.md("""```python
revenue_by_month_year = ...
revenue_by_month_year
```""")

Or set a global config int hte app header

__marimo__ = {
    "hide_code": False  # or True to hide all
}

💡 Tip:

If your app is intended as a tutorial or learning resource, you can:

• Create a separate notebook.md or .ipynb with all code visible

• Link to that from the main app using mo.md("[View full notebook](./notebook.md)")

• Use Marimo's UI components to conditionally show/hide code with toggles

  show_code = mo.ui.checkbox(value=False, label="Show Code")
  
  if show_code.value:
      mo.ui.code(code)
  

Set Dark Theme

Review the documentation. In short, add the following snippet at the top of your main app file:

# /// script
# [tool.marimo.display]
# theme = "dark"
# ///