Bringing Back `whits.io`

Dollar short and a day (or 1,461) late, but I’m finally bringing my old website back online. I’ve never really been one for front-end webstuff, but I am very fond of rambling so I’ve been itching to get it working again. It’s just a hugo static site (congo theme) with my patent migraine backend infrastructure behind it.

There’s not much to add; for the past few years my website has just 302’d to my GitHub profile, but now runs on a (admittedly tiny) DigitalOcean Droplet running Nginx backed by Certbot-LetsEncrypt. Kinda an anachronism now, certainly reminiscent of what I was doing a decade ago. I did not want any credentials stored, and because the website’s repo is public site updates happen according to a droplet cron job. It’s not modern and event-triggered like it probably should be, but it’s simple, safe, and it works, which is all that matters in the end.

The cron job runs a shell script which does the following:

1. Check the most recent commit hash of the repo
2. If the most recent hash is equivalent to the last checked hash (stored in ./main.hash) then the program exits and waits to be triggered again.
3. If the most recent hash is different from ./main.hash, then ./main.hash is deleted and the repo is pulled to the latest commit in branch main.
4. hugo is called to compile the website into .../public/
5. The data of /var/www/whits.io/public/* is deleted
6. Ownership of .../public is changed to www-data:www-data
7. The contents of .../public/ is moved into /var/www/whits.io/public/
8. The new most recent hash is stored in ./main.hash, and the process repeats.

The script used is available on the site’s GitHub repo

It is worth noting that due to how old Debian is, Go had to be installed via backports and Hugo from the .deb from their GitHub releases page (specifically the extended version for webp support).

Yes this could just be a GitHub Pages website, but where’s the fun in that? (real reason: I forgot that existed before I was in too deep)