How to Create and Manage Conda Environments: A Developer's Workflow

Setting up a robust development workspace often feels like a battle against dependency conflicts. You install a library for one project, and suddenly, another project breaks. This is where Conda shines. Unlike standard Python virtual environments, Conda manages not just Python packages but also binaries and external dependencies, making it the go-to tool for data science and complex backend projects.

In this guide, I will walk you through the most efficient workflow to create, manage, and share your Conda environments without the headache.

Quick Cheat Sheet

If you are already familiar with the concepts and just need the commands, here is the fast track.

# Create a new environment with a specific Python version
conda create --name myenv python=3.10
# Activate the environment
conda activate myenv
# Install packages (always specify the channel if possible)
conda install -c conda-forge numpy pandas
# List all environments
conda env list
# Remove an environment
conda env remove --name myenv

Step-by-Step: Creating a Conda Environment

Creating an environment is more than just running a command; it is about setting the right foundation for your project.

1. The Basic Setup

Open your terminal (or Anaconda Prompt on Windows). To create a clean workspace, run the following command. Replace myenv with a descriptive name for your project.

conda create --name myenv

2. Specifying Python Versions (Best Practice)

Never create an empty environment if you can avoid it. Specifying the Python version right at the start prevents conflicts later. If your production server runs on Python 3.9, your local environment must match it.

To understand why matching environments matters, you might want to read about the differences between development, staging, and production environments.

conda create --name myenv python=3.9

3. Using the conda-forge Channel

The default Conda channel is stable but often lags behind in updates. Most developers prefer conda-forge, a community-led collection that offers the latest versions and faster dependency solving.

conda create --name myenv -c conda-forge python=3.10 numpy pandas

Using the -c conda-forge flag ensures you get the most compatible and up-to-date packages.

Managing Your Environments

Once you create an environment, you need to know how to navigate between them.

Activating and Deactivating

To start working in your isolated space, you must activate it.

conda activate myenv

You will notice the terminal prompt changes from (base) to (myenv). This visual cue confirms that any package you install now will stay inside this specific environment.

To go back to the global system context:

conda deactivate

Listing All Environments

If you work on multiple projects, you might forget the exact names of your environments. The list command helps you track them.

conda env list

The output shows all available environments. The one currently active is marked with an asterisk (\*). If you are on a Linux system and need to locate the physical files of these environments, checking a guide on how to find a directory on linux can be helpful for manual inspection.

Removing an Environment

Over time, unused environments consume significant disk space. You should delete the ones you no longer need.

conda env remove --name myenv

Warning: This action is irreversible. Ensure you have backed up any important configurations before running this command.

Advanced Operations: Sharing and Cloning

A major advantage of Conda is reproducibility. You can share your exact setup with a colleague, ensuring code runs on their machine exactly as it does on yours.

Exporting to YAML

To share your environment, you export the configuration to a .yml file. This file lists every package and version currently installed.

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

I recommend using the --no-builds flag. This makes the file more cross-platform compatible, as it excludes OS-specific build strings. This concept is similar to managing environment variables where portability is key.

Creating from YAML

If a colleague sends you an environment.yml file, you can recreate their setup with a single command:

conda env create --file environment.yml

Cloning an Environment

Sometimes you want to test a risky upgrade without breaking your working setup. Instead of starting from scratch, create an exact copy (clone) of your current environment.

conda create --name myenv_backup --clone myenv

Now you can experiment on myenv_backup safely.

Troubleshooting Common Issues

Even with a solid workflow, you might encounter hurdles. Here are the most common ones I see developers face.

Command Not Found If your terminal says conda: command not found, it usually means Conda is not added to your system PATH.

• Solution: Run conda init in your terminal and restart the shell.

Slow Dependency Solving Conda can sometimes take a long time to solve the environment.

• Solution: Switch to the libmamba solver, which is significantly faster.

conda update -n base conda
conda install -n base conda-libmamba-solver
conda config --set solver libmamba

Pip vs. Conda Conflicts While you can use pip inside a Conda environment, mixing them carelessly causes breakage.

• Best Practice: Always try to install packages with conda install first. Only use pip as a last resort for libraries not available on Conda channels.