To get a sense of where the project is going, have a look at our Roadmap

There are a number of ways to contribute:

  • Create new issues for feature requests or to report bugs

  • Adding / correcting documentation

  • Adding a new unit test

  • Contributing a new node that accesses a specific datasource

  • Contributing a new node that implements a domain-specific algorithm

  • Commenting on issues to help out other users

To contribute:

  • Fork the PODPAC repository on github

  • Create a new feature branch from the develop branch

git checkout develop  # Assuming you've already checked out and tracked the develop branch
git branch feature/my_new_feature
  • Make your changes / additions

  • Add / modify the docstrings and other documentation

  • Write any additional unit tests

  • Create a new pull request

At this point we will review your changes, request modifications, and ultimately accept or reject your modifications.

Coding style

  • Generally try to follow PEP8, but we’re not strict about it.

  • Code should be compatible with both Python 2 and 3


Podpac uses Python Black autoformatting. Configuratin can be found in pyproject.toml. During installation for development, pre-commit hooks are automatically installed, including a hook that will run Black. Code that is not formatted with Black will fail in CI.


All classes and methods should be properly documented with docstrings. Docstrings will be used to create the package documentation.

Many IDE’s have auto docstring generators to make this process easier. See the AutoDocstring sublime text plugin for one example.


Podpac adheres to the numpy format for docstrings:




Note that class attributes can be documented multiple ways.

  • Key public attributes and properties should be documented in the main class docstring under Parameters.

  • All public attributes (traits) should be documented in the subsequent line by setting its __doc__ property.

  • All public properties created with the @property decorator should be documented in the getter method.

class ExampleClass(tl.HasTraits):
    """The summary line for a class docstring should fit on one line.

    Additional details should be documented here.

    attr1 : str
        Description of attr1.
    attr2 : dict, optional
        Description of attr2.

    attr1 = tl.Str()
    attr.__doc__ = ":str: Description of attr1"

    attr2 = tl.Dict(allow_none=True)
    attr2.__doc__ = ":dict: Description of attr2"

    attr3 : tl.Int()
    attr2.__doc__ = ":int: Description of secondary attr3"

    def attr4(self):
        """:bool: Description of attr4."""

        return True


All references to podpac classes (:class:), methods (:meth:), and attributes (:attr:) should use the public path to the reference. If the class does not have a public reference, fall back on the full path reference to the class. For example:

def method(coordinates, output=None):
    """Class Method.  
    See method :meth:``.
    See attribute :attr:``.

    coordinates : :class:`podpac.Coordinates`
      Coordinate input
    output : :class:`podpac.core.units.UnitsDataArray`, optional
      Container for output

      Returns a UnitsDataArray


To help adhere to PEP8, we use the pylint module. This provides the most benefit if you configure your text editor or IDE to run pylint as you develop. To use pylint from the command line:

$ pylint podpac                 # lint the whole module
$ pylint podpac/     # lint single file

Configuration options are specified in .pylintrc.


We use the python logging library to support library logging. To include logging in your module, use:

import logging
log = logging.getLogger(__name__)  # creates a logger with the current module name

# log a message
log.debug('Debug message')'Info message')
log.warning('Warning message')
log.error('Error message')

Do not set levels, handlers, or formatters in your modules. See the python logging documentation for details on how to construct log messages.

To handling logging in your application, you can use a simple config:

import logging
logging.basicConfig(level=logging.INFO)  # log to console
logging.basicConfig(level=logging.INFO, filename='podpac.log')  # log to a file podpac.log

or a more complicated configuration which handles only podpac logs:

# log only podpac logs
log = logging.getLogger('podpac')

# log only podpac logs in a file
import logging
from logging import FileHandler
log = logging.getLogger('podpac')
log.addHandler(FileHandler('podpac.log', 'w'))

We have created a convience method create_logfile() in the podpac.utils module to automatically create a log file for only podpac logs.

Import Conventions / API Conventions

Public API

The client facing public API should be available on the root podpac module. These imports are defined in the root level podpac/ file.

The public API will contain a top level of primary imports (i.e. Node, Coordinate) and a second level of imports that wrap more advanced public functionality. For example, podpac.algorithm will contain “advanced user” public imports from podpac.core.algorithm. The goal here is to keep the public namespace of podpac lean while providing organized access to higher level functionality. The most advanced users can always access the full functionality of the package via the podpac.core module (Developer API). All of this configuration and organization should be contained in podpac/, if possible.

For example:

import podpac

 # Public Classes, Functions exposed here for users

 # organized submodules

 # the settings module

 # developer API goes here. i.e. any non-public functions, or rarely used utility functions etc.

Developer API

The Developer API follows the hierarchical structure of the core directory. All source code written into the core podpac module should reference other modules using the full path to the module to maintain consistency.

For example:

import podpac


In source code /podpac/core/

Podpac Module


from podpac import settings
from podpac.core.units import Units, UnitsDataArray
from podpac.core.coordinates.coordinates import Coordinates
from podpac.core.utils import common_doc

Note: The modules podpac.settings and podpac.units.ureg MUST be imported without using the from syntax. For example:

import podpac.settings                 # yes
from podpac.settings import CACHE_CIR  # no


We use pytest to run unit tests. To run tests, run from the root of the repository:

$ pytest
$ pytest -k "TestClass"    # run only the TestClass

Configuration options are specified in setup.cfg.

Integration testing

We use pytest to write integration tests. Generally these tests should be written in seperate files from unit tests. To specify that a test is an integration test, use the custom pytest marker @pytest.mark.integration:

import pytest

def test_function():

class TestClass(object):

    def test_method(self):

Integration tests do not run by default. To run integration tests from the command line:

$ pytest -m integration

See working with custom markers for more details on how to use markers in pytest.

Skip Tests on CI

In some circumstances, we don’t want tests to run during the CI process, e.g. when the test will incur charges from a cloud provider).

To invoke tests with --ci command line option:

$ pytest --ci podpac

AWS Marker

To specify that a test uses AWS and should not be run during CI tests, mark it as such:

import pytest
def test_function():
class TestClass(object):

    def test_method(self):

Alternatively, you can specifically run only AWS tests by using the -m command line option:

$ pytest -m aws podpac

Adding Markers

See to add additional custom markers and behavior.

Code Coverage

We use pytest-cov (which uses coverage underneath) to monitor code coverage of unit tests. To record coverage while running tests, run:

$ pytest --cov=podpac --cov-report html:./artifacts/coverage podpac   # outputs html coverage to directory artifacts/coverage

We use coveralls to provide coverage status and visualization. Commits will be marked as failing if coverage drops below 90% or drops by more than 0.5%.


  • We encourage and welcome contributions from the wider community

  • Presently, a small group of core developers decide which contributions will be incorporated

    • This is a complex software library

    • Until the library is mature, the interfaces and features need tight control

    • Missing functionality for your project can be implemented as a 3rd party plugin

    • For now, we are trying to be disciplined to avoid feature creep.