After finishing the official quick start, I wanted to try Next.js out with an actual project.
Soon after I started web programming, I used plain HTML and CSS (and a lot of ChatGPT) to put together a web version of John Selden’s Table Talk. The process took a day or two around new year’s eve and new year’s day.
8 months on and with Next, the process has got a lot smoother.
Unsurprisingly in hindsight, it was the text wrangling that took the most time. I already had it easy: Project Gutenberg provides a single html file, which each chapter separated enclosed within a <div class="chapter">. This makes things a lot simpler than when I was using a plaintext file converted with pandoc.
Still, it took some time for me to get the bash script right, even with all the plethora of AI tools supplied by PromptBros. My focus was on comparatively old tools such as sed and awk: perhaps I would have been better off using some modern and more familiar tools.
Still, if other Project Gutenberg books follow a similar format, I may be able to reuse the scripts I already have.
Once text-wrangling is done, the Next.js’s out of the box features really shines:
- Routing can be automatically done with arranging a directory structure
- The css can be configured in one go.
- Deployment with Vercel is seamless.
A minor snag towards the end was the fact tht Next.js routes are absolute by default: which means if you nest the directories further in or further out, special steps have to be taken, e.g.
const pathname = usePathName()
<Link href={`${pathname}/chapter/1`}>Chapter 1</Link>
I wonder what the design philosophy behind this is. Was it an accident of history, kept on because of the need to maintain backwards compatibility?