plangs!
Emmanuel Oga

Releasing v3 — source available soon

The site you are looking at is new. Plangs v3 replaces the version that had been serving this domain since early 2025, and it is a rebuild rather than a reskin: different data model, different rendering, different everything underneath.

What actually changed

The old site was a single-page app. One HTML shell was sent for every URL and JavaScript decided what to draw. That is why the old /python and the old /about were byte-for-byte the same document until the browser got to work.

v3 is a static site. Every page is built ahead of time and served as itself, so a language page arrives as a page about that language. Links work without JavaScript, the browser’s back button behaves, and search engines get real documents instead of an empty frame.

The data is the point

Plangs is a database first and a website second. The database is now a folder of YAML files — one per language, tool, library, license, paradigm, tag and platform — and that folder is the canonical source. Nothing is authored in a CMS or hidden in a hosted table. Every fact on this site is a line in a file that can be read, diffed, and argued with.

That has a few consequences worth knowing about:

  • Every page has a machine-readable twin. Add .md to any URL and you get the markdown behind it.
  • Agents get a map. There is an /llms.txt describing the site for anything trying to read it programmatically.
  • Relationships go both ways. Languages influence other languages, and both ends of that edge are queryable, which is what the influence graphs on each language page are drawn from.

Language pages now carry two things the old site did not. Trend sparklines draw from Languish and PYPL, so you can see where a language’s attention has been going rather than just where it is now. And the influence graphs are computed from the dataset’s own edges — laid out at build time, not fetched — which is why they load instantly and why they are the same for everyone.

Source available soon

The code is not public yet. It will be. There is some tidying to do first, and I would rather publish something a contributor can actually run than dump a repository and leave people guessing.

Data corrections are the most useful thing anyone can send once that happens. Plangs started from Wikipedia, some AI prompting and a lot of manual fixing, and the mistakes that remain are exactly the kind a person who knows one language well will spot in about four seconds.

Thanks for visiting

If something looks wrong, it probably is. That is what the next post will be about.