Last Week In Code -001-
Last week in code -001-
The beginning of a new chapter
I find I always have a whole lot to talk about at the end of every week. And that most of it is from the tech space. I attend this Software Club meeting every Thursday where I try to use some of that time to talk about the latest developments with people who get it. We exchange opinions on things we’re building. You know how that is. So I was like, how about I create something here that’ll be an outlet for that? So I did. Like many things I’ve started, I can’t be sure this will last. But I won’t try to preemptively kill it either. May the Force be with it. This week, I’m shipping the skeleton and letting the structure emerge naturally as I pump in more episodes. My consumption is personal, so what I include might skew toward my interests. Beware :)
Updates From me
A new way to do project management
So I’ve been moving my project management from scattered TODO lists to GitHub issues. And when I finally got used to it, I wondered why I hadn’t done that earlier. I feel It’s way better than what I was doing. Using a todo list that is. The ease of centralization for issue tracking being one among other reasons. Every time I think of an improvement, every time I notice a bug while using my apps, I can always just add an issue to their corresponding issue pages. And there’s labeling too. That labeling system is actually how I got my first PR. If you are like how I was, not knowing much about this feature github offers, here’s an article that could get you started. click. So as part of a larger app I had to create a pomodoro app. I got the basics down then moved the list of all I hadn’t added to the project’s issues page. I labelled most of them ‘good first issue’. And next day someone sent in a PR asking to work on one of the issues. There’s a lot of talk online about how to get your pull requests accepted. But I’d never been in the reviewer position before. Never felt that side of it.
Can’t do it – don’t delegate it: Non-mainstream AI advice
So I beat myself up, the 1000th time, for choosing to use AI for something I didn’t understand. In one of my projects, a mostly CLI driven app, I wanted to move from CLI to GUI. To build a django front end. I was feeling lazy and wasn’t ready to put in the time so I gave the whole task to Grok. Hours of prompting later, I ran the code it generated and almost drowned in bugs. Bugs arising from concepts Grok had used I that didn’t understand. The lesson hit hard:
Don’t assign AI tasks that you can’t do yourself.
I’ve seen myself go through the same process over and over but still wonder why I keep coming back. A youtuber (Brodie Robertson) sums up my point nicely here.
In other news last week
AI & LLMs
Data Privacy & Cybersec
A challenge for you
Can you tell how much of the above post was GPT? Until next week. Yes, I should see you next week. That’s the plan.. May the force be with you