Add ci for pyodide marimo

[paddymul]

Checks

• I have checked that this enhancement has not already been requested

How would you categorize this request. You can select multiple if not sure

Developer Experience/CI (feature to make it easier to devlop on Buckaroo)

Enhancement Description

it fails because of fasparquet error. make sure we build pyodide marimo in ci, check to make sure renders in browser
https://marimo.io/p/@paddy-mullen/buckaroo-styling-gallery

Pseudo Code Implementation

No response

Prior Art

n/a

[paddymul]

Implementation Plan

Overview

Add a CI job that tests Buckaroo in a pyodide/marimo WASM environment.

Steps

1. Identify the blocker — the issue mentions fastparquet fails in pyodide. Determine:

   • Is fastparquet imported unconditionally? → Make it conditional
   • Can we use a WASM-compatible serialization fallback (JSON, or Arrow IPC via pyarrow-wasm)?

2. Make fastparquet optional — wrap import fastparquet in try/except everywhere it's used. Fall back to JSON serialization in pyodide environments (detect via sys.modules.get('pyodide')).

3. Add CI job in .github/workflows/:

   • Use pyodide-build or a Playwright-based approach that loads pyodide in a browser
   • Install buckaroo via micropip.install("buckaroo")
   • Run a basic smoke test: create a DataFrame, render with BuckarooWidget, verify no errors

4. Coordinate with Re-enable full WASM marimo Playwright test suite #513 — the WASM marimo Playwright tests are related. This CI job could be the same infrastructure.

Files to modify

• buckaroo/serialization_utils.py — conditional fastparquet import
• .github/workflows/ — new or extended CI job
• pyproject.toml — make fastparquet an optional dependency if not already

@codex

[chatgpt-codex-connector]

Summary

• Made fastparquet non-mandatory for base installs by removing it from core dependencies and introducing a new optional parquet extra (fastparquet>=2024.5.0). This helps Pyodide installs avoid pulling an incompatible package by default. pyproject.tomlL34-L41 pyproject.tomlL108-L111
• Added explicit Pyodide runtime detection in serialization utilities and introduced a fallback path in to_parquet:
  • If fastparquet is unavailable and runtime is Pyodide, Buckaroo now serializes with pyarrow (after JSON-encoding object dtype columns), instead of immediately erroring.
  • Non-Pyodide behavior remains unchanged (still raises informative ImportError when fastparquet is missing). buckaroo/serialization_utils.pyL21-L23 buckaroo/serialization_utils.pyL182-L200
• Added a new CI workflow job PyodideBuckarooSmoke that runs a browser-based Playwright smoke test:
  • loads Pyodide in Chromium,
  • runs micropip.install("buckaroo"),
  • creates a pandas DataFrame and BuckarooWidget,
  • fails if browser/page errors are detected. .github/workflows/checks.ymlL526-L587

Testing

• ✅ python -m py_compile buckaroo/serialization_utils.py
• ⚠️ pytest -q tests/unit/serialization_utils_test.py (failed in this environment because pandas is not installed)
• ⚠️ uv sync --dev --extra test (failed due network/tunnel fetch error to PyPI metadata in this environment)

Committed on branch work as:

• e396219 — Add pyodide smoke CI and make fastparquet optional.

Also created the follow-up PR via the make_pr tool with title:

• “Follow-up: make fastparquet optional and add Pyodide Buckaroo smoke CI”

View task →

[paddymul]

Related: PRs #532, #509, #549

• PR fix: use Pyodide-bundled fastparquet for WASM marimo test #532 (merged): Uses Pyodide-bundled fastparquet for WASM marimo tests
• PR Feat/fix marimo wasm #509 (merged): Fixes for marimo WASM environment
• PR refactor: switch transport from Parquet to Arrow IPC #549 (open): Arrow IPC transport would remove the fastparquet dependency entirely, which is the root cause of pyodide failures

The fastparquet blocker will be resolved when PR #549 merges. The WASM marimo CI is partially working via PR #532's approach. See also #513 (re-enable full WASM tests).