On making websites like it's 1999
I burnt out of software engineering in 2024, had a brief burst of energy where I learnt how to make iOS apps, and then promptly completely crashed out for a year and wrote nothing. In a variation on the standard software engineer career path, I started selling vintage film cameras in an antiques market instead.
So it had been a year without writing code of any substance, and I found myself needing a website for my professional photography endeavour. To pick a framework, a frontend system, a hosting platform.
Well, fuck that. I'll see CSS precompilers in hell.
I settled instead on trying out the usual photographer route, influenced by fellow YouTube sellouts; Squarespace. Which, it turns out, has a couple of issues:
- Everything about it runs like shit
- It generates invalid HTML which, in a bizarre turn, causes Bing to refuse to index it
Leaving aside a Microsoft product suddenly being pro-standards-compliance, that I don't generally disagree it's a bit daft to ignore non-standards-compliant webpages as a search engine in 2025, and that it was probably a theme issue; the overall experience was not great. I just wanted a quick, simple website that didn't require me to confront my software demons. I did point out #2 to customer support but hey, what do I know about software.
Anyway, I ended up rebuilding the website from scratch. And when I say scratch, I mean HTML & CSS.
No frameworks, no compilers, no nothing. Maybe a normalise stylesheet, for old times' sake. A call on an old friend.
Which lasted until I realised I needed multiple pages and Server-side Includes aren't really a thing any more, and I shoved it into a simple Sinatra app so I can avoid duplication. But still, the intention was there.
And how lovely that time was. I don't think I've enjoyed making a website that much at any point in the last 20 or so years. How refreshing to not have to worry about whatever the fuck the current flavour of the month is in the JavaScript world. How pleasant it was to build a website like it was 1999 but using flexbox instead of tables. There has been some progress, thankfully.
The stack ended up beautifully simple and it's one I'm quite happy with. I prefer Sinatra to static site builders as it gives me flexibility, and after 15 years of professional Ruby it's nice to fall back on something comfortable and minimal that I never grew to dislike.
The full stack is:
The server is practically overkill, but hey - I enjoyed this so much, I might make another website and host it on the same box. Deploying to a single server like this might actually be one of my favourite parts of the process, as I can't remember the last time I managed to avoid containers. Perhaps one of my biggest takeaways from this experience has been that I'd rather cock something up myself than learn yet another PaaS. For reference, deploys are handled using Tomo, which is alright.
Anyway, the thought of working professionally in software again still fills me with dread. What if someone asks me to integrate AI into what I'm working on again? Where will I hide the body?
But I am overall happy with how it turned out. And if you want to either hire me to photograph something for you or you'd like to point out in what ways the website is broken, head on over to mayday.wales.
Now I just need to decide if I'm back enough to build my own booking system, so I can chuck the Acuity system in the bin as well.
Maybe I could treat myself with a little Elixir.