Modern web development is complicated and so are the tools to solve those problems. If you want to something even slightly complicated for your side project you fight dozen build pipelines, config files and CLIs.
“Tool X is all you need.” —Creator of Tool X
“Well, Tool X is too complicated, so here’s Tool Y.” —Creator of Tool Y
“But Tool Z is the real answer (just not for beginners).” —Creator of Tool Z
Meanwhile I’m in the corner, sobbing gently, because there is no perfect tool, only varying flavors of pain and suffering.
For a hobby project, configuring an environment is the last thing I want to deal with. I just want to push code to Git and watch it run. Vercel and Netlify get you most of the way. Unless you’re building anything other than a React app, then you are screwed.
Yes, a single Dockerfile can cover most deployments.
But I like to avoid it because I hate the fact that it treats my system's
memory like it's the Pacific Ocean when I am here sitting on what is basically
a pond. So my options for web are limited going "old-school" using
nginx and exposing ports directly, which isn't too bad but is some
configuration. Or fight the docker whale which will drain my poor
system.
Lately, I've found a solution that is not a pain to deal with and doesn't require me to learn a new language or framework. Just get AI to deal with it.
"Hey (AI model) my web scraper deployment isn't working with bun (no i shall not use npm or node - I do not care) because headless chromium is a son of a bitch. Figure it out." - Me
I go to sleep peacefully in the meantime.
I wake up to a message from the AI model saying "I've fixed your web scraper deployment. It's now working with bun." - AI model
Bliss. My silicon DevOps intern works the graveyard shift so I don't have to.