Jian

Open source · 7 Nov 2022 · 3 min

Hacktoberfest 2022

A month of open source, one small word cloud generator, and the parts of it that were not the word cloud.

Hacktoberfest is a month-long celebration of open-source projects, their maintainers, and everyone who contributes to them. I have taken part since 2017.

Five Hacktoberfest t-shirts, dated 2017 through 2021, laid out in a row on a light grey carpet.
Five years of Hacktoberfest shirts.

This year I did something different. After a stretch of not feeling well, and only recently feeling better, I wanted to finish something of my own instead of contributing to someone else's — so I built a word cloud generator. It looks like a word cloud generator. The interesting parts are underneath it.

A word cloud of the Wikipedia article on Australia. AUSTRALIA is by far the largest word, with Australian, area, population, Commonwealth, country, largest and world next.
The cloud it produced from the Wikipedia article on Australia.

What is underneath it

  • gh-pages deploys it, with no build step worth the name.
  • d3 and d3-cloud draw the cloud itself.
  • Bootstrap 5 handles the responsive layout, so I did not hand-write flex boxes.
  • Web Workers do the text processing off the main thread, so the page stays usable while it works.
  • Natural-language libraries handle the tokenising and lemmatising, rather than my own regex.

What it taught me

  • Bootstrap 5 is still pleasant to use, and dropping jQuery as a dependency made the library much lighter.
  • d3-cloud needs a small patch if you want more than one cloud on a page, or it warns. Starting again, I would use the Chart.js word cloud plugin instead.
  • Web Workers are the right tool for anything that would otherwise freeze the page — the work still takes seconds, but the page can say so.
  • The HTML <select> element cannot hold child nodes, so putting images in a dropdown takes more work than it looks like it should.

Once the base worked, I opened issues and labelled them for contributors. I built it in a hurry and plenty of it could be better, which was deliberate: finish it first, then let other people improve it.

The part that stayed with me

Being away from this field for a while means being behind in it, and the field does not wait. The word cloud was never really the point. It was a way of checking that I could still learn the new parts on my own, which after a long illness is a question worth answering rather than assuming.

The other thing it settled: health is part of the toolchain. Learning needs energy and morale behind it, and neither is a given.

The generator is here and the source is on GitHub. Issues and pull requests are welcome.

Bye.

Written in 2022, converted from the old Hugo site and lightly edited with an AI assistant.

Comments

Discussion lives on GitHub — you'll need a GitHub account to post.