Workflows#

Import and export of requirements.txt files#

If you already have a requirements.txt file in an existing project, pipenv can resolve dependencies. If the requirements.txt file is in the same directory, simply with $ pipenv install or, if it is in a different directory, with $ pipenv install -r /path/to/requirements.txt.

Conversely, you can also create a requirements.txt file from an existing Pipenv environment with:

$ pipenv run pip freeze > requirements.txt

Upgrade workflow#

  1. Find out what has changed upstream:

    $ pipenv update --outdated
Package 'requests' out-of-date: '==2.13.0' installed, '==2.19.1' available.

  2. To update the Python packages, you have the following two options:

    • update everything with $ pipenv update

    • update individual packages, for example requests with $ pipenv update requests

Pipfile vs. setup.py#

A distinction must be made whether you are developing an application or a library.

Libraries

They offer reusable functions for other libraries and applications/projects. You have to work with other libraries, each with their own dependencies. To avoid version conflicts in dependencies between different libraries within a project, libraries should never commit dependency versions. However, you can specify lower or upper limits if you are relying on a particular feature or bug fix. Library dependencies are noted in install_requires of the setup.py file.

Applications

They use libraries and are mostly not dependent on other projects. They should be implemented in a specific environment and only then should the exact versions of all their dependencies and sub-dependencies be specified. Facilitating this process is the main goal of Pipenv.