A central question in supporting catagorical data is which libraries and how?

Matplotlib wants to avoid conditional supports because they are messy and can lead to unreasonable quote bloat quite quickly. And it puts the onus on matplotlib to maintain that the conversions work throughout api changes in the client side library (such as pandas or xray). Instead, in discussion with a pandas developer, the plan is to write the conversion code such that it lives in the library providing the data (such as pandas) and that code is registered with matplotlib.

Because the quite amazing Pandas already supports a lot of plots of its catagorical data, it makes for the perfect library to test out this plan. If this works, it will also hopefully yield a guide to adding support for your favorite library. So where am I starting?

I’m reading the following lovely pieces of code as a framework:

And I’m documenting my task list on the wiki for my fork of matplotlib so that I can find it and my mentors and I can stay on the same page. Going by our weekly meetings, my todo says:

1) Write tests modeled on the matplotlib ticker tests

2) Get tickers, formatters, and locators working natively with pandas catagorical dataframe

3) bundle it up in a matplotlib pull request

4) after mpl dev discussion, pitch it to pandas as a PR