When I worked in an E-commerce company, performance was important although to the developers dismay not very important to the business. This changed once we started using Google’s Impact Calculator, it provided a real value to what load time could do. One night I couldn’t sleep, so I was browsing Twitter and I seen a Tweet that really stood out to me.
I think it stuck with me because of the physical reminder that such am expansive game fitting on just 880KB.
This lead me to think of a website 1.44mb as a (very loose) rule of thumb, so I came up with the idea of “fit on a floppy”. So like many times before in the middle of the night I purchased the domain but luckily for me there was a sale on the .website TLD. So I purchased fitonafloppy.website in April and left it until September when I picked up tools, definitely not the longest I’ve let a domain languish.
I didn’t want to spend much time building it as in terms of projects it was pretty low on my priorities, so to get up and running quickly I built a prototype in PHP. Surprisingly I managed to finish build it within one day. Given the way to code worked essentially just a function with no database calls I realised this is a good use case for serverless architecture. I suspected PHP wouldn’t be supported by AWS Lambda and I was right, so I decided to rewrite it in Python 3.6. Given a chance to do it all over again I would have used Go. From a frontend perspective I took lessons learnt from previous projects and used a generator called Yeogurt to setup a basic static site. I did however get to try Nunjucks and Sass, that being said I will stick to SCSS in the future.
Normally I stick to using Github Pages for static hosting but since I am already using Lambda I decided to use AWS for all my infrastructure. So for static web hosting I used S3 and for CI/CD I decided to give CodePipeline a try. I think CodePipeline is the 6th CI/CD I’ve used, I am on the fence about the whole containerization approach but it is growing on me. I definitely do like the visual representation of the pipeline though.
Overall the project went quite smoothly, I kept the specifications tight which allowed it to not drag on for too long. I was able to experiment with a few different things which helped me develop, most notably serverless architecture. As I look back on this project, I am glad I did it and I hope I can use it to develop better websites going forward.