A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.
Python artificial-intelligencedagdata-sciencedata-visualizationdataflowdeveloper-toolsmachine-learningnotebookspipelinepythonreactivesqlweb-app
# Copyright 2026 Marimo. All rights reserved.
from __future__ import annotations
import subprocess
import sys
from typing import TYPE_CHECKING
if TYPE_CHECKING:
from pathlib import Path
NOTEBOOK_SOURCE = """import marimo
app = marimo.App()
with app.setup:
import marimo as mo
@app.cell
def _():
with mo.persistent_cache("extensionless"):
print("computed")
value = 42
print(f"value={value}")
return
if __name__ == "__main__":
app.run()
"""
def test_persistent_cache_in_extensionless_script(tmp_path: Path) -> None:
"""`mo.persistent_cache` works when the notebook runs from a spool copy.
Slurm executes sbatch scripts from an extensionless copy of the submitted
file. The cache block must still resolve its enclosing cell.
"""
spool_path = tmp_path / "slurm_script"
spool_path.write_text(NOTEBOOK_SOURCE, encoding="utf-8")
for run in range(2):
result = subprocess.run(
[sys.executable, str(spool_path)],
capture_output=True,
text=True,
cwd=tmp_path,
timeout=120,
)
assert result.returncode == 0, (run, result.stderr)
assert "value=42" in result.stdout
# The first run computes; the second restores from the cache, so
# the block body must not re-execute.
block_ran = "computed" in result.stdout
assert block_ran == (run == 0), (run, result.stdout)
cache_dir = tmp_path / "__marimo__" / "cache"
assert cache_dir.is_dir()
assert any(cache_dir.iterdir())