hey,
this week ended up being more about unlearning and relearning than shipping new features.
i had a bit of a reality check with vibe coding. when i switched to supabase for my new project, i wanted to see if i could build it without really knowing how supabase worked. for the first page, it felt amazing: instant progress. same with the second. but by the third page, i hit a bug. every time i asked ai to fix it, it just shifted the problem somewhere else. when i finally looked at the code myself, i realised the issue: i didn’t actually understand what was going on.
the next day, i wiped everything. i went back to zero and decided to learn supabase properly. i rebuilt the first page by hand. it took way longer - around 15 minutes instead of one - but this time, i understood every line. more importantly, the joy of writing code came back. it made me think a lot about instant vs delayed gratification. vibe coding feels fast at the start, but it’s fragile and opaque. writing code myself is slower, but it teaches me real concepts and makes me a better programmer long-term. from now on, i’m sticking to writing my own code, using ai as a helper - not a crutch.
huge shoutout to the supabase docs too. they made learning the basics genuinely easy, especially auth. magic links were almost effortless to set up, and it honestly felt cleaner than a lot of oauth flows i’ve used before.
because i spent this week learning supabase, sql, table design, and querying properly, i’ve pushed my “3-week full mvp” plan back by a week. but it feels like the right tradeoff. i now have a much stronger foundation, and future progress should be faster and cleaner because of it.
also… my opinion on sql is changing. i used to love mongodb and refuse anything else, but i’m starting to see why people swear by sql. the structure, the guarantees, the power - it’s growing on me. as my dad always says: sql can do everything nosql can do, and more.
with a clearer understanding of the stack, i’m going to make a refreshed, more realistic todo list today, properly tailored to how i actually work. if all goes well, i’ll start building properly again tomorrow.
this week wasn’t flashy, but it mattered. fewer shortcuts, more fundamentals. slower now, stronger later.
thanks for following along,
vulcan