Environment Setup Repo

This repo captures and reproduces your Conda+pip environment cleanly.

Quick Start

Option A — Use environment.yml (recommended)

conda env create -f environment.yml
# If the env already exists:
conda env update -f environment.yml --prune
# Activate (uses the name defined inside environment.yml):
# conda activate <name_from_yaml>

Option B — Use split files (conda + pip)

# Create and activate
conda create -n myenv --file conda_requirements.txt -y
conda activate myenv

# Then install pip packages
pip install -r pip_requirements.txt

---

Exporting your current environment (kept in repo)

After you add or change packages, regenerate the files with:

./scripts/export_env.sh <your_env_name>

This updates:

• environment.yml (portable full spec)
• conda_requirements.txt (conda-only spec)
• pip_requirements.txt (pip-only spec excluding pip/setuptools/wheel)
• LAST_EXPORTED_UTC.txt (timestamp marker)

Commit and push these changes to keep the repo current.

---

Recreating the environment on another machine

From a fresh clone of this repo:

# Preferred
./scripts/recreate_env.sh <target_env_name>

# Or do it manually:
conda env create -f environment.yml  # then conda activate <name_from_yaml>
# OR:
conda create -n <target_env_name> --file conda_requirements.txt -y
conda activate <target_env_name>
pip install -r pip_requirements.txt

Tip: If environment.yml defines a different name than the one you want to use, either:

• Use ./scripts/recreate_env.sh <target_env_name> (it will try to activate the name in the YAML if present), or
• Edit the name: in environment.yml before running the command.

---

One-liners you may find handy

• List only the pip packages you installed (excludes bootstrap tools):

pip list --format=freeze | grep -vE '^(pip|setuptools|wheel)=='

• Save pip-only packages for migration:

pip list --format=freeze | grep -vE '^(pip|setuptools|wheel)==' > pip_requirements.txt

• Export everything into a single YAML:

conda env export --no-builds > environment.yml

---

Windows notes

• Use Anaconda Prompt or PowerShell where conda is initialized.
• On Windows, the Bash scripts will run under Git Bash or WSL. Alternatively, copy the commands inside the scripts and run them manually in Anaconda Prompt.

---

Best practices

• Prefer conda install when packages exist on conda channels for easier resolution.
• Use pip install for packages not available on conda.
• After changes, re-run ./scripts/export_env.sh <env> and commit updates.