How it works
Imagine you want to use jQuery in one of your projects. Therefore you build an integration package, say
js.jquery, which contains the
jquery.js file. You also add a
resource.py which would look like this:
from fanstatic import Library, Resource library = Library('jquery', 'resources') jquery = Resource(library, 'jquery.js', minified='jquery.min.js')
To finish up, you state the version of the downloaded
jquery.js in the
setup.py and upload the package to PyPI.
setup.py. To use them, you write
js.jquery.need() inside any view. Of course you can also add a
resource.py and declare a local resource with dependencies:
import fanstatic import js.jquery library = fanstatic.Library('custom', 'resources') my_awesome_js_code = fanstatic.Resource( library, 'js/my_awesome_js_code.js', minified='js/my_awesome_js_code.min.js', depends=[js.jquery.jquery])
Every time you call
my_awesome_js_code.need() inside a view, it will also load jQuery.
my_awesome_js_code.need() inside a view.
The preparation described above can be tedious, but the assumption is, that someone else has built the integration package before, so you can reuse it. Installing a dependency therefore is as easy as adding a line to your
setup.py and declare it as a dependency of one of your resources.
Fanstatic makes it a lot easier to update external libraries, since you only need to increase the version number in your
setup.py. You can also make sure to use the same version across many projects, e.g. by sharing some buildout configuration across projects.
And besides many small nice features, Fanstatic can also create bundles, i.e. it will merge
my_awesome_js_code and all of its dependencies into one big file and deliver it to the client, rather delivering all files separately. This usually makes the initial page load much faster.
Despite all the benefits Fanstatic offers, we were getting more and more frustrated with it. The main reason is, that we feel a decline in the activity of the Fanstatic user group. With fewer people using Fanstatic, we often have to build the integration packages ourselves. This eliminates one of the main benefits, i.e. that we could reuse the integration packages of others.
For example, in a big project we did last year, we used integration packages for js.jquery, js.classy, js.chosen and many more. Those are pretty popular packages and the integration worked out of the box.
However, even those packages have issues: The integration package for jQuery does not offer version 2.0 or higher and Chosen is only available in a single (outdated) version. We often stumbled upon similar issues with other libraries, e.g. the integration package for js.modernizr is more than 2 years behind.
In addition, some newly developed libraries may not be available at all. Of course we could write the integration packages for those or update the jQuery integration package. But this would mean to find out were the code for an integration package is hosted, write a pull request and wait for a merge, as well as a new release on PyPI. In case the owner is not active anymore, you are out of luck and must add a duplicate package to PyPI. Tedious and unsatisfactory.