Custom Context and more

• Added the option to pass a custom Context class to cleanly override any methods and provide your own implementation for things like pyplot.
• Added a metadata parameter to Show where you can put a dictionary with per-slide keys, whose data will be available in each ctx instance.
• Lots of changes to the server UI.

Fix dependency issues

Merge pull request #50 from apiad/develop

v20.2.1

Because we forgot

Merge pull request #49 from apiad/develop

Develop

Better server

Go to http://auditorium.apiad.net to see.

Publish command

This release brings two new commands: publish and server, which allow you to host a slideshow with a public URL while still running in your computer.

Better verticals

This release improves the API for vertical slides. Right now, instead of the old nested functions, simply use a main function @slide decorator, like this:

@show.slide
def main(ctx):
    # ...

@main.slide  # <-- see here
def vertical(ctx):
    # this is a vertical slide under `main`

Port to FastAPI

This release brings a breaking chance. We ditched Sanic in favour of FastAPI. Sanic is great, but is way too large a dependency, especially for static hosting at https://now.sh. With FastAPI we get a smaller install without support for the run command and an optional larger install (with [server]) that add uvicorn and a better support for now.sh.

Switching to CalVer

We are switching over to Calendar Versioning scheme for good!

Ready for now.sh

This release adds a few tweaks here and there to make auditorium render generate now.sh compatible source. This has allowed us to move the demo over to auditorium-demo.now.sh and released us from the burden of manually hosting a working version of the demo.

This also means that now, by look at demo/now.json you can get a pretty good idea of how to setup a static version of your slideshow that works in now.sh with backend included.

Extended compatibility

This release adds compatibility with Python 3.7 and 3.8, as well as a new development environment in which all targeted Python versions are easy to test.