reqwire 0.1a2

reqwire: a wrapper for easily managing requirements with pip-tools.

reqwire wires up your pip requirements with pip-tools.

Install

pip install reqwire, how else?

Features

• Manage multiple requirement source files, or tags

• Add requirements from the command line (à la npm --save)

• Compile tagged source files with pip-compile (from pip-tools)

Rationale

Until PEP 518 and Pipfile become reality, maintaining one or more requirements.txt files for Python projects will continue to be subject to personal preference and differing opinions on best practices [1]. Typical workflows involve maintaining multiple requirements.txt files, and many projects have some form of tooling, be it Makefile targets or external tools, such as Vincent Driessen’s excellent pip-tools.

reqwire is a glorified wrapper around pip-tools, and imposes a slightly opinionated workflow:

• Python requirements are split into source files and built files, with the built files being the output from pip-compile, containing pinned versions of the entire dependency graph. (Use reqwire init to quickly scaffold the necessary directory structure.)

• Source files (with the .in extension) represent a project’s immediate dependencies, and are always pinned to specific versions or version ranges.

• Source filenames are synonymous with tags, which can be passed to reqwire add and reqwire build to maintain requirements for entirely separate environments.

Roadmap

• Unit tests

• Provide a command to update requirements (user-specified version)

• Provide a command to freshen requirements

• Provide a command to combine tags to a single output file (easily possible with pip-compile)

• Provide a utility for generating setup and runtime requirements in setup.py scripts using setuptools.

• Abandon in favor of Pipfile 👌

[1]

• http://nvie.com/posts/pin-your-packages/

• https://www.kennethreitz.org/essays/a-better-pip-workflow

• https://mail.python.org/pipermail/distutils-sig/2015-December/027954.html

• https://devcenter.heroku.com/articles/python-pip#best-practices