I'm writing this half-asleep so the I'm not sure how congruent this is.
Was working on the browser today, mainly reading the response and making sense of the data. Splitting the response into it's various parts. Spent the most part of the evening yesterday and most of today. Today, I was like I have the raw bytes, what if for now I just kept the headers and body as bytes, just like that the problem was solved for now. I've still got to figure out how I want to deal with headers. Will most likely do what the standard library does. It seems like the simplest approach.
I was working on the simplest HTML renderer, one which removes the tags and prints the contents as is, to the stdout. I was quite surprised that I couldn't pass the raw bytes to a method. It's such a straightforward thing, or so I thought. I'm still not sure what the mistake is. I'm working with gdb but there are still some simple things I need to understand about how to use gdb, and I'll get there. It's a little bit frustrating but also fascinating to know that the things you assumed should just work, only works because someone made it so. It's one of the joys and surprises of low level programming I suppose. It could very well could be a skill issue. I'm also glad that Claude cannot answer it either, it's answers were so bad today that, I've quit asking it for help when it comes to Zig.
I just canceled my Claude subscription
Just as I was writing this article. I've canceled it before, but as I was writing the last paragraph I was trying to think when was the last time I found it useful. It was when I wanted a SQL query about three weeks ago. I might re-start the subscription if I really feel like I'm slowing down. But I'd like to do an experiment, this one might be quite uncomfortable. But let's give it a try. The experiment is to try and learn the tools that I use Claude for. Just off the top my head, they would be to get better at Lua and SQL. Hopefully, I don't come across any Neovim config that I have to do for a while. As, I don't want to have to learn too many things at once. A little bit nervous about the idea of not using Claude. The main concern is that, am I becoming one of those people who reject all new technology? My rationality is that, if mastery is better and more profitable over time then this is a good move. I'm not sure if I want to tell anyone about it, for now.
It's strange but the immediate feeling is that I'll be stuck on things forever. Goes to show, how much you can come to rely on a tool. I'm being silly, people have been programming without even the Internet, reading code from magazines. I can do this. See how long I can keep this up. I guess this marks Day Zero of the no AI experiment.
Since, most people are sold on the AI idea, especially developers, I'm feeling skeptical about the whole AI thing. Most people are always wrong, when it comes to gauging value. The only reason for me not being completely sure is that, I don't really know, I'm not on the forefront of AI development. Also, I came across the idea of Analog computers, it might solve the issues of scale but it is mostly useful once the training is done. But as far as bets go, this is the one I want to make. I'm betting on competence and excelling at my craft. To get as good as I can. It's my only rule in life, live life on my terms and be willing to risk it all. Especially that I have the luxury of not having any dependants. I Might not always have this luxury, so might as well make all the bets I can make. I'm not a gambler per say, these are long term bets with some of the risks accounted for. I've made my single decision.