Here's an issue (public, so we can ask for contributions to it from our readers) to record the bits and pieces left to do on following the introduction of py.cafe to our docs' examples.

• Change the requirement in hatch.toml when py.cafe release their mkdocs plugin (separate issue, stored as 1204 internally)
• Update the code examples in the data.md how-to guide to use py.cafe once we can solve how to illustrate data handling without creating multiple projects in a profile to store the iris.csv dataset.
  • We considered using the function to create the dataframe directly for this example but disregarded it because it doesn't illustrate data loading,
  • We could code the data download from storage in this GitHub repo
  • We could make a bunch of projects like I did for other examples using custom css, for example, but there are a lot and it results in code example duplication, so it's not the optimum solution for any of the examples.
  • Best longer-term solution is to either change the example to call an API to get a dataset (which varies over time, as that's what the example is about) or to wait to see if we can work something out with py.cafe to add files to projects dynamically via the plugin (a bit like we specify requirements files as extra-requirements)
• If a solution arises whereby we no longer need to create examples and link to them (rather than use the plugin) we can delete the examples I've created so far. But long-term, if we have these examples as part of our documentation, we should store them under a new profile in py.cafe that uses a shared email account so the team can update those projects as needed, rather than use my login.