-
We can all agree that coding harnesses are absolutely ABSURDLY effective, and when trying to get things done quickly - are the way to go.
HOWEVER I think that chat interfaces are underrated: for one key reason - they make you learn.
If an agent takes your prompt, adds a feature, sees that it errors, then fixes it - you didn’t learn anything from the debugging. Nor did you learn anything from the implementation.
A chat interface necessitates a bunch of annoying copy pasting, manual debugging, and pain. But I think that’s a better way to learn.
-
In want of web builds
Recently I’ve been planning my next major game development project. It’s going to be a "relatively" simple grid based RPG with a few more systems to make it interesting.
A year ago I would’ve reached for the "Godot" toolbox - but recently I’ve been wanting to branch out and try something lower level with a framework/library.
I haven’t actually "made" anything per-say yet. I tried making something for raylib jam 6 but I only could give it like 3 hours, when I should’ve gave it 20 or so. As a result, I spent some more time on learning engine-less game development.
The big thing that’s killing me with these frameworks is just How many there are. I feel like when I look to learn a new skill or area of development, and I try to see what libraries are used, there’s always a few that are interchangeable, but people feel strongly about them.
This is where AI comes in handy. I ask "Whats the most recent" and "What are some things it lacks compared to the other one" and it saves me like 30 minutes of searching.
Anyway, the point of this article is that I really really need web builds. That’s what matters the most to me. To just throw this game onto itch.io and call it a day. However, a lot of ones in my preferred language (Python) don’t work, and Odin Raylib 6 to web is irritating to me. It’s just too many steps - and I don’t actually know what emscripten is doing behind the scenes.
Apparently Pyray + Pygbag works, and I’ll look into it later.
It’s interesting, because a couple of years back I looked into Ebitengine - and it actually solves this problem perfectly. Yay.
- reccomendationsforjuniors
-
Nvidia DLSS
A while ago Nvidia (And today, in fact, but I didn’t plan to write about that) released a demo for DLSS. It’s a tool that basically paints over frames in the game, and makes them look hyper-realistic (I’m simplifying, and the tech is really cool - but that’s a lot of it).
This has been met with an unimaginable wave of backlash. A lot of people think it’s betraying a lot of the artistic aspects of game design and history.
In fact, it’s a perfect microcosm of a lot of the concerns surrounding AI and its impact on human expression.
I just want to bring up a sobering anecdote.
I had a set of friends in college who weren’t gamers, and the only thing they really played was maybe 2K - and Mortal Kombat.
I showed them Guilty Gear Strive. They were not impressed. They said the game looked bad. Worse than MK.
This is hilarious to those who are in games who are familiar with how these two franchises are seen in terms of their artistic achievements.
Guilty Gear Strive is seen as one of the greatest accomplishments in 3d animations - in terms of personality, gameplay visibility, performance and more.
Mortal Kombat is seen as less impressive, relying on mostly utilizing motion capture (sometimes to an overwhelming degree) and not going as crazy on the styling compared to other fighting games. (The fatalities are technical masterclasses, make no mistake).
The average person - even those who play games - don’t really care about that.
These people don’t care about the technical aspect. They just want something that just "looks good" - which usually means looking realistic.
I think that’s why so much AI art is the same.
Note: No one is stupid or less capable because they don’t care. I’m not a car person. I make a lot of mistakes that car people get mad at me for. But they also understand I’m more of a tech guy instead of a
I wonder how many stylish games will see an uptick in sales because people won’t be put off by the visuals. Is it really such a bad thing that people can pick up "Hi-Fi Rush" and make it look realistic? I don’t know the answer to this.
- pairmentoring
-
2 monitors is the PERFECT way to handle development in the modern age. One main monitor for editing code, one side one for looking at notes/documentation and previews. When I see peoples setups with 5 monitors, I just roll my eyes. There’s no way you can pay close attention to monitor 5 and ignore 1-4. It’s in your vision, taking away your focus. It’s the worst with these people who have their monitors set up unergonomically, where they have to twist their bodies.
I used to be one a 3 monitor guy. My home setup had a very small table, and it was quite awkward. I’m speaking from experience.
I mostly do 2 monitors.
I’m also very frequently a 1-monitor guy as well. There’s a lot of times when you don’t need to have something on the second monitor - and you just need to lock in. For coding, and writing, it’s ideal.
- Rolling my own static site generator
- Google Kaggle challenge =
-
CICD painline (as opposed to pipeline).
It’s annoying when you’re working on something, and it seems like it works (or it works on your local) but for some reason it fails in a later deployment step.
But it becomes worse if the iteration times are long. 3 mins editing code → 30 seconds to push to main → 90 seconds for the worker to start → 5 minutes before you can see the result.
And it becomes painful when the error handling logs are incomprehensible. Something fails and you either see a short, pithy and useless "error in service" with no information, or you see an overly verbose stack trace that is somehow just as useless.
The CICD painline.
- Competence and confidence