PINGInstaller

Light-weight application for installing PINGMapper and associated packages. PINGInstaller is designed to install a conda environment from a yml specified as a URL or localy hosted yml.

Example yml file structure:

name: ping
channels:
  - conda-forge
dependencies:
  - python<3.13
  - gdal
  - numpy
  - git
  - pandas
  - geopandas
  - pyproj<3.7.1
  - scikit-image
  - joblib
  - matplotlib
  - rasterio
  - h5py
  - opencv
  - pip
  - pip:
      - pingverter
      - pingmapper
      - pingwizard
      - pinginstaller
      - doodleverse_utils
      - psutil
      - tensorflow
      - tf-keras
      - transformers
      - rsa

The special thing about PINGInstaller is that it will install the conda environment based on the conda prompt it is launched from. This enables end-users with multiple conda installations to choose the flavor of conda as needed.

Supported prompts include (but may not be limited to):

• Miniforge
• Miniconda
• Anaconda
• ArcGIS Python Command Prompt

PINGInstaller is also compatible with projects in the Doodlevers.

Installation & Usage

Step 1

Open (download, if not already available) the conda prompt you want to use (ex: On Windows 11 - Start --> All --> Anaconda (miniconda3) --> Anaconda Powershell Prompt).

Step 2

Install PINGInstaller in the base environment with:

pip install pinginstaller

Step 3

Then install the environment from a web or locally hosted yml with:

python -m pinginstaller https://github.com/CameronBodine/PINGMapper/blob/main/conda/PINGMapper.yml

That's it! Your environment is now ready to use.

If you want to update the environment, simply re-run the environment installation script with:

python -m pinginstaller https://github.com/CameronBodine/PINGMapper/blob/main/conda/PINGMapper.yml

Ta-ta for now!

Troubleshooting

"Access is denied" When Updating from Wizard

Problem: When updating via PINGWizard, you see:

remove_all: Access is denied.

Cause: In rare cases, file locking can prevent updates when the wizard is running from the environment being updated.

Solution: PINGWizard (v1.0.12+) automatically runs updates from the base environment to avoid this. If you still see this error:

1. Update PINGWizard: pip install pingwizard -U
2. Or manually run from base: conda activate base && python -m pingwizard

"Non-conda folder exists at prefix" Error

Problem: When creating an environment, you see:

error    libmamba Non-conda folder exists at prefix - aborting.
critical libmamba Non-conda folder exists at prefix - aborting.

Cause: A leftover directory exists at the environment location (e.g., Z:\miniforge3\envs\ping) that's not a valid conda environment. This can happen after an incomplete removal or failed installation.

Solution: Manually remove the directory and try again:

# Windows PowerShell
Remove-Item -Recurse -Force "Z:\miniforge3\envs\ping"

# Then retry installation
python -m pinginstaller

# Linux/Mac
rm -rf ~/miniforge3/envs/ping

# Then retry installation
python -m pinginstaller

Slow Environment Solving

Problem: Environment creation takes a very long time (>10 minutes).

Solutions:

1. Use mamba (much faster): Install mamba in your base environment:

   conda install -n base mamba -y
   

   PINGInstaller automatically detects and uses mamba when available.

2. Check your network connection: Slow downloads can cause delays.

3. Clear conda cache:

   conda clean --all -y
   

"Package Not Found" or Solver Errors

Problem: Conda/mamba cannot find required packages or conflicts prevent solving.

Solutions:

1. Update conda/mamba:

   conda update -n base conda -y
   # or if using mamba
   mamba update -n base mamba -y
   

2. Check channel configuration: Ensure conda-forge is available:

   conda config --show channels
   # Should include conda-forge
   

3. Update all packages in base:

   conda update --all -y
   

Installation Hangs During Housekeeping

Problem: Installation appears stuck during the "Updating conda/mamba" step.

Solution: Press Ctrl+C to cancel, then run with quiet mode to skip housekeeping updates:

python -m pinginstaller <yml_url> -q

Or skip housekeeping by commenting out the install_housekeeping() call temporarily.

Wrong Conda Installation Being Used

Problem: PINGInstaller is using a different conda installation than expected.

Solution:

1. Verify which conda is active:

   which conda  # Linux/Mac
   where conda  # Windows
   

2. Check CONDA_PREFIX:

   echo $CONDA_PREFIX  # Linux/Mac/PowerShell
   echo %CONDA_PREFIX%  # Windows CMD
   

3. Activate the correct conda first, then run pinginstaller from that environment.

Mamba Not Being Detected

Problem: PINGInstaller uses conda even though mamba is installed.

Solution:

1. Verify mamba is in base environment:

   conda activate base
   mamba --version
   

2. Reinstall mamba if needed:

   conda install -n base mamba -y
   

3. Check output: PINGInstaller will print "Using mamba for faster installation" if detected.

Permission Denied Errors

Problem: Cannot write to conda directories.

Solutions:

• Windows: Run PowerShell or Command Prompt as Administrator
• Linux/Mac: Check directory ownership, or reinstall conda in user directory (not system-wide)

Import Errors After Installation

Problem: Packages installed but cannot be imported.

Solutions:

1. Activate the environment first:

   conda activate ping
   python -c "import pingmapper"
   

2. Verify environment location:

   conda env list
   # Ensure 'ping' environment is listed
   

3. Check package installation:

   conda list | grep pingmapper
   

Getting Help

If you encounter issues not covered here:

1. Check the GitHub Issues
2. Enable debug mode for detailed output: python -m pinginstaller <yml> --debug
3. Open a new issue with the full error output and your system details