I'm writing a django application in my spare time for a footy-tipping competition we're running at work. I figured I'd use this time wisely, and get up to speed on virtualenv, pip, packaging, django 1.3, and how to write an easily redistributable application. So far, so good.
I'm up to the packaging part. A lot of the django apps on GitHub for instance are mostly bundled (roughly) the same way. I'll use django-uni-forms as an example.
An assumption I'm making is that the
setup.py are the only required pieces that pip needs to do its job. Is that correct? What other components are necessary if my assumption is wrong?
Are the required packaging files generally generated, or are they crafted by hand? Can dependencies be described and then installed also? My application depends on
django-uni-forms, and I have it listed in a
requirements.txt file within my app which I used to install the dependency; but is that something that the packaging system can take care of?
What are the steps I need to follow to package my application in such a way that pip will be able to install it and any dependencies?
setup.py should be sufficient.
This blog post really has some good information on this topic: Packaging a Django reusable app
And here's another good, detailed overview that helped me a lot: Python Packaging User Guide
Especially the tips to get your static files (templates) included are important as this might not be obvious at first.
And yes, you can specify required packages in your
setup.py which are automatically fetched when installing your app.
install_requires = [ 'django-profiles', 'django-uni-forms', ],
Obviously now we have two places where dependencies are defined, but that doesn't necessarily mean that these information are duplicated: setup.py vs requirements.txt
With this setup your package should be installable via
As Pierre noted in the comments, there's now also a relevant section in Django's official documentation: Packaging your app
And then there is this "completely incomplete" guide, which really gives a great overview over packaging and uploading a package to PyPI: Sharing Your Labor of Love: PyPI Quick And Dirty