Hatch is a productivity tool designed to make your workflow easier and more efficient, while also reducing the number of other tools you need to know. It aims to make the 90% use cases as pleasant as possible.
For me personally, Hatch has entirely replaced the manual (or entire!) use of these:
- Cookiecutter PyPackage
- pip
- virtualenv
- pyenv
- Pew
- inve
- pytest
- Coverage.py
- twine
- bumpversion
- zest.releaser
- Incremental
python setup.py ...
/^\/^\
_|__| O|
\/ /~ \_/ \
\____|__________/ \
\_______ \
`\ \ \
| | \
/ / \
/ / \
/ / \ \
/ / \ \
/ / _----_ \ \
/ / _-~ ~-_ | |
( ( _-~ _--_ ~-_ _/ |
\ ~-____-~ _-~ ~-_ ~-_-~ /
~-_ _-~ ~-_ _-~
~--______-~ ~-___-~
Table of Contents
- Completely cross-platform (terminal colors too!) \(*_^)/
- Configurable project creation! CI service files, readme format/badges, licenses, etc.
- Easiest virtual environment management available
- Activation of virtual envs without disruption of current environments
- Ability to send commands to virtual envs without the need for activation
- Changed prompts when in a virtual env
- Installing/updating packages defaults to a per-user basis, allowing global usage with elevated privileges
- Configurable semantic version bumping
- Robust build/package cleanup
- Easy testing with code coverage
- Simple building and releasing for PyPI
- All commands are environment-aware w.r.t. python/pip
- Virtual envs can be the target for relevant commands
- Editable packages can be the target for relevant commands
Starting a new project is as easy as:
$ hatch new my-app
Created project `my-app`Now you have a fully function package that can be built and distributed.
$ tree --dirsfirst my-app
my-app
├── my_app
│ └── __init__.py
├── tests
│ └── __init__.py
├── LICENSE-APACHE
├── LICENSE-MIT
├── MANIFEST.in
├── README.rst
├── requirements.txt
├── setup.py
└── tox.ini
2 directories, 8 filesYou can also bump the version of most projects without any setup:
$ git clone -q https://github.com/requests/requests && cd requests
$ hatch grow build
Updated /home/ofek/requests/requests/__version__.py
2.18.4 -> 2.18.4+build.1
$ hatch grow fix
Updated /home/ofek/requests/requests/__version__.py
2.18.4+build.1 -> 2.18.5
$ hatch grow pre
Updated /home/ofek/requests/requests/__version__.py
2.18.5 -> 2.18.5-rc.1
$ hatch grow minor
Updated /home/ofek/requests/requests/__version__.py
2.18.5-rc.1 -> 2.19.0
$ hatch grow major
Updated /home/ofek/requests/requests/__version__.py
2.19.0 -> 3.0.0Checking code coverage is a breeze:
$ git clone https://github.com/ofek/privy && cd privy
$ hatch test -c
========================= test session starts ==========================
platform linux -- Python 3.5.2, pytest-3.2.1, py-1.4.34, pluggy-0.4.0
rootdir: /home/ofek/privy, inifile:
plugins: xdist-1.20.0, mock-1.6.2, httpbin-0.0.7, forked-0.2, cov-2.5.1
collected 10 items
tests/test_privy.py ..........
====================== 10 passed in 4.34 seconds =======================
Tests completed, checking coverage...
Name Stmts Miss Branch BrPart Cover Missing
-----------------------------------------------------------------
privy/__init__.py 1 0 0 0 100%
privy/core.py 30 0 0 0 100%
privy/utils.py 13 0 4 0 100%
tests/__init__.py 0 0 0 0 100%
tests/test_privy.py 57 0 0 0 100%
-----------------------------------------------------------------
TOTAL 101 0 4 0 100%Creating virtual envs is incredibly simple:
$ hatch env my-app
Already using interpreter /usr/bin/python3
Successfully saved virtual env `my-app` to `/home/ofek/.local/share/hatch/venvs/my-app`.
$ hatch env -ll
Virtual environments found in /home/ofek/.local/share/hatch/venvs:
my-app ->
Version: 3.5.2
Implementation: CPythonYou can nest activated virtual envs:
$ hatch use my-app
(my-app) $ hatch use -n fast
2 (fast) $ hatch use -n old
3 (old) $ exit
2 (fast) $ exit
(my-app) $ exit
$or use them as usual:
$ python -c "import sys;print(sys.executable)"
/usr/bin/python
$ hatch use my-app
(my-app) $ python -c "import sys;print(sys.executable)"
/home/ofek/.local/share/hatch/venvs/my-app/bin/python
(my-app) $ hatch use fast
(my-app) $ exit
(fast) $ python -c "import sys;print(sys.executable)"
/home/ofek/.local/share/hatch/venvs/fast/bin/python
(fast) $ exit
$And so much more!
Hatch is distributed on PyPI as a universal wheel and is available on Linux/macOS and Windows and supports Python 3.5+ and PyPy.
$ pip3 install --user hatchor simply pip if that already points to a Python 3 version.
If hatch doesn't work on your system immediately after that, please
run this command then
this command.
After the first installation, you may want to run hatch config --restore to
ensure your config file is available.
- meta
- start using AppVeyor
- next to the snake ascii art, put a
hatched egg (blocks1.0.0)
- issues
- I really, really need help with
this
(blocks
1.0.0). Whoever solves this will get all my BTC. Occasionally added to, never decreases.
- I really, really need help with
this
(blocks
- project creation
- Commands
bench- use https://github.com/ionelmc/pytest-benchmark (blocks1.0.0)python- installs the desired version of Python. will work on each platformstyle- maybe not needed. use https://github.com/PyCQA/flake8docs- maybe not needed. use https://github.com/sphinx-doc/sphinx/
Hatch is distributed under the terms of both
at your option.
- All the people who work on Click
- All the people involved in the Python packaging ecosystem
- All the people involved in the pytest ecosystem
- Ned Batchelder, for his Coverage.py is the unsung heroic tool of the Python community. Without it, users would be exposed to more bugs before we are.
- Te-jé Rodgers for bestowing me the name
hatchon PyPI
Important changes are emphasized.
- You can now
usea new temporary virtual env via the-t/--tempoption!!! - Pretty terminal colors {^.^}
- Nicer self updating for Linux.
- Upon project creation, it is now possible to automatically install it locally (editable) in a virtual environment.
- If the
releasecommand is invoked with no path information, it will use acurrent_directory/distdirectory if it exists before defaulting to the current directory. - Fix/Change: Shedding now uses
/as a separator instead of a comma for names.
envcommand is now quiet by default (option removed) and option-v/--verboseaddedenvcommand option-l/--listcan now stackbuildcommand is now quiet by default and option-v/--verboseadded. Also, it now shows what files are inside the build directory afterward.- Resolving user supplied paths for options is now more robust.
- Fix: using virtual envs no longer uses an abundant amount of CPU
- Significant improvements to documentation
MANIFEST.innow considers users' files fromextrasconfig entry
- Change:
eggcommand is nownew - Removed
download_urlattribute fromsetup.py. See: pypa/packaging.python.org#264
- Initial release