Skip to content

pipeio_nb_exec: python_bin as string breaks subprocess.run when using conda env

Problem

In src/projio/mcp/pipeio.py:1120, when env is set: 

python_bin = f"{conda} run -n {env} python"

This creates a single string like "conda run -n cogpy python". But downstream in packages/pipeio/src/pipeio/notebook/lifecycle.py:1004-1006, _require_jupytext passes it as: 

_sp.run([python_bin, "-m", "jupytext", "--version"], ...)

The single string "conda run -n cogpy python" is treated as a single executable path (element 0 of the list), not split into args. This causes FileNotFoundError caught by the except block, resulting in "jupytext not found".

Same issue in _jupytext() and the papermill command construction in mcp_nb_exec.

Fix

python_bin should be a list when it's a multi-part command, or lifecycle helpers should handle splitting. Simplest approach: change python_bin type from str | None to str | list[str] | None and use shlex.split() or pass as list from the wrapper: 

python_bin = [conda, "run", "-n", env, "python"]

Then in lifecycle.py and mcp.py, normalize: cmd_prefix = [python_bin] if isinstance(python_bin, str) else python_bin.

Files to change

  • src/projio/mcp/pipeio.py:1120 — pass list instead of string
  • packages/pipeio/src/pipeio/notebook/lifecycle.py — _require_jupytext() and _jupytext() need to handle list python_bin
  • packages/pipeio/src/pipeio/mcp.py:4275 — papermill cmd construction needs same fix

Source context: pixecog

PixEcog (pixecog): Neuropixels and ECoG dataset and analysis

Recent commits: 

9b2f6fa Scaffold ecephys TTL removal mod, flow overview + mod docs, demo notebook
80194af Add TTL characterization & removal demo notebook (preprocess_ieeg)
dc93496 Update mkdocs pipeline nav

README:

type: readme

Quick Start for Collaborators

Follow this checklist to get started with Pixecog documentation and workflows.

🐀 Pixecog Project — Compact Overview

Core principles

  • One immutable BIDS raw dataset (raw/) as the canonical baseline
  • Each analysis pipeline ha