The fork allows installing numpy before installing pip requirements in order to fix a bug with the install of
statsmodels. The numpy install is triggered by a file
pre-requirements.txt. This file has the same format as a pip requirements file.
An IFTTT recipe is set up to alert us when the upstream repo has changes for merging.
Recommended web frameworks include Django and Flask. The recommended webserver is Gunicorn. There are no restrictions around what software can be used (as long as it's pip-installable). Web processes must bind to
$PORT, and only the HTTP protocol is permitted for incoming connections.
Python packages with C dependencies that are not available on the stack image are generally not supported, unless
manylinux wheels are provided by the package maintainers (common). For recommended solutions, check out this article for more information.
Deploying a Python application couldn't be easier:
$ ls Pipfile Pipfile.lock Procfile web.py $ heroku create --buildpack heroku/python $ git push heroku master … -----> Python app detected -----> Installing python-3.6.6 -----> Installing pip -----> Installing requirements with Pipenv 2018.5.18… ... Installing dependencies from Pipfile… -----> Discovering process types Procfile declares types -> (none)
requirements.txt must be present at the root of your application's repository.
You can also specify the latest production release of this buildpack for upcoming builds of an existing application:
$ heroku buildpacks:set heroku/python
Specific versions of the Python runtime can be specified in your
[requires] python_version = "2.7"
Or, more specifically:
[requires] python_full_version = "2.7.15"
Or, with a
$ cat runtime.txt python-2.7.15
Runtime options include:
Copy the snippet above into CLI.