First pass at adding an atom feed

[pipwilson]

Steve Messer mentioned that he had tried to add a feed for the blog, but not quite got there. I thought I'd have a quick pass.

This is a naive implementation which was intended to be minimally invasive or overly implementation specific. It adds a feed for every _list.md, even ones where _list.md is the only file in a subdirectory of content.

It reuses the data created during publication of a list as its input - this means a feed should always match the relevant list page.

The only customisations I had to make were exposing the parent_path and a publish date to the template in order to get absolute URLs and make some of the feed values valid. It's possible that parent_path is passed to the template already but I couldn't see it.

I have not committed the generated changes to files in docs/ since I think the GitHub Action is supposed to do that, but they are all published correctly locally during a make render.

[stevenjmesser]

That’s really kind of you, thanks for contributing, @pipwilson! I’ll share this with the team so that we can get it reviewed.

[pipwilson]

It's the first time I've ever looked at this repo or used jinja, and several years since I last used Python, so I'm sure they'll suggest a better way forward, but it was fun :)

[colmjude]

Thanks @pipwilson

I've made a couple of tweaks to the output to reduce the validation issues/warnings from https://validator.w3.org/feed/check.cgi.

There are still a couple more, but we can fix them over time. Going to merge now 👍

[pipwilson]

Ah that's great, thanks - I used the validator when I started but didn't check it again at the end. FWIW I blogged about this at https://philwilson.org/blog/2024/03/rss-for-fun-and-the-public-good/