Ideas to discuss in grooming session about docs

[stichbury]

This is a list of everything I've been asked for recently

• load .env in settings.py for aws credentials and use templating in credentials.yml and the catalog.

  • Expand the docs on credentials in AWS

• test examples +1

  • YES but predicated on how we recommend unit testing to users. We need a tech design session on this

• deployment of a model with fastapi or streamlit + model versioning

  • how to serve a web application powered by a kedro project. touch on what other teams in Labs have done. This could instead be a blog post. See https://blog.streamlit.io/using-chatgpt-to-build-a-kedro-ml-pipeline/ for blog post ideas.

• Great expectations example code

  • Pandera 0.16 has been released so maybe we could favour this (but we don't have any extension yet)
  • We could add a warning to the pages right now and ask for contributors

• Need more examples or videos on the airflow integration

  • Ask Simon Brugman about this as he's been making PRs about airflow. Otherwise reflect that it's not recently been updated.

• Need more on packaging.

  • We have at least one page about this, so maybe it's not easy to find or incomplete. One to review.

• Kedro context better explained

  • This is something we could improve by pulling together what we already have. Are they actually talking about KedroSession? Would probably reduce the number of bug reports when people try to use Kedro cli from python. High priority. It may be more sensible to prioritise a fix instead to Click rather than use docs as a workaround. See Consider using run(standalone_mode=True) to fix packaged Kedro project getting a sys.exit  #2682 for more.

• Add a section on how to use the Python API for each part of the kedro project lifecycle (E,g, new project, create catalog, add/modify pipelines, run the project, and package the project)

  • Jo to follow up on this one.

• Extend spaceflights to add namespacing, config templating, S3 as a filestore, various deployment options, pipeline templating (need to ask the survey participant about the last one)

  • Base this on the frequency of questions we are asked. People are always asking about S3 so this is definitely one to show. How to share a project? How to deploy?

• Would be good if there are best practices for how pipelines should be structured, or at least common pitfalls to avoid. (e.g. what are some ways to structure (or have naming conventions for) parameters)

  • This one is needed

• Kedro's tips and tricks, for example I didn't knew that you could pass parameters when creating a pipeline

[stichbury]

I've transferred all of these into existing or new tickets, so closing this now.