I've become a bit performance obsessed. I often lament that the worst part of working on something you are passionate about is that first impressions with the software are everything and that if someone isn't hooked visually in about 5 seconds, you're not going to get them interested in your work.
To that end, I've been performance obsessed in web components, HAXTheWeb , and all server configuration areas. I was able to take our scores previously from pretty bad to decent on Google Lighthouse, just by making some tweaks to how things rendered, timing, etc.
I neglected one factor in all my previous testing though... I wasn't actually using a live, running, out there in the open web server. It was always local development or github pages, which can only go so far on their own. I expected to get the same results when I pushed things up to a live server and ran some tests. Boy was I wrong...
These rather insane scores are showing an upcoming SaaS version of HAXcms that I'm currently building. Other team members reactions to playing with the SaaS offering recently have ranged from "I love you" to "I can't believe it's this far along" to "why is it this fast? This is stupid fast" to utter disbelief.
@hey__mp opened our test site on his phone, on rural wifi (so DSL, and not good DSL) and had the site in 1 second. This HAXcms in production, is stupid fast. I decided to shoot a video talking about how we were able to achieve these stupidly fast scores given the following constraints we intentionally place on ourselves:
So given these constrains,... how the hell were we able to achieve these scores? I encourage you to watch and find out.