Game DevLog
@devlog.gruen.lol
Making a game. Need a place for notes. Let’s try using a PDS.
Author ⇨ @gruen.us
Author ⇨ @gruen.us
Progress:
- Rope is largely functional (need to sharpen some of its segment creation/deletion logic)
- Collisions are easier assuming spherical cows of uniform density
- Vastly improved rope controls for player
- (Almost) workable collisions for player. Bugs abound, but not getting in the way.
- Rope is largely functional (need to sharpen some of its segment creation/deletion logic)
- Collisions are easier assuming spherical cows of uniform density
- Vastly improved rope controls for player
- (Almost) workable collisions for player. Bugs abound, but not getting in the way.
January 4, 2026 at 4:21 AM
Progress:
- Rope is largely functional (need to sharpen some of its segment creation/deletion logic)
- Collisions are easier assuming spherical cows of uniform density
- Vastly improved rope controls for player
- (Almost) workable collisions for player. Bugs abound, but not getting in the way.
- Rope is largely functional (need to sharpen some of its segment creation/deletion logic)
- Collisions are easier assuming spherical cows of uniform density
- Vastly improved rope controls for player
- (Almost) workable collisions for player. Bugs abound, but not getting in the way.
Also, an awesome part about having an LLM handy is that it knows how to do all the math I've once forgotten.
It is still shit at choosing the right math to use, but it is very good at putting it into code.
It is still shit at choosing the right math to use, but it is very good at putting it into code.
January 3, 2026 at 2:30 AM
Also, an awesome part about having an LLM handy is that it knows how to do all the math I've once forgotten.
It is still shit at choosing the right math to use, but it is very good at putting it into code.
It is still shit at choosing the right math to use, but it is very good at putting it into code.
Focusing on rope wrapping and collisions this go around. Since roping is the core mechanic, it is critical to get it right while being performant. So, spending time with it.
Also changed to dark mode because it's no longer a daytime project but a nighttime one.
Also changed to dark mode because it's no longer a daytime project but a nighttime one.
January 3, 2026 at 2:26 AM
Focusing on rope wrapping and collisions this go around. Since roping is the core mechanic, it is critical to get it right while being performant. So, spending time with it.
Also changed to dark mode because it's no longer a daytime project but a nighttime one.
Also changed to dark mode because it's no longer a daytime project but a nighttime one.
Back on my bullshit. (Using an older laptop to test some collision detection algorithms... and based on the stutters, clearly getting it wrong.)
January 1, 2026 at 3:37 AM
Back on my bullshit. (Using an older laptop to test some collision detection algorithms... and based on the stutters, clearly getting it wrong.)
Gemini 2.5 Pro was able to recreate about 2 hours of work in 3 prompts (and about 20 minutes of hand-debugging). Kinda neat.
Very broken code, but I can see this being very useful for testing out some ideas before premature optimization once the rest of the basic game physics are up and running.
Very broken code, but I can see this being very useful for testing out some ideas before premature optimization once the rest of the basic game physics are up and running.
March 28, 2025 at 6:57 PM
Gemini 2.5 Pro was able to recreate about 2 hours of work in 3 prompts (and about 20 minutes of hand-debugging). Kinda neat.
Very broken code, but I can see this being very useful for testing out some ideas before premature optimization once the rest of the basic game physics are up and running.
Very broken code, but I can see this being very useful for testing out some ideas before premature optimization once the rest of the basic game physics are up and running.
So, annoyingly, my stand-in game engine (ebiten) positions based on object center whereas my stand-in physics engine (chipmunk) positions things based on upper-left-corner.
Also, the Y axis direction is flipped. (Positive is down in ebiten-land and up in chipmunk land.)
Also, the Y axis direction is flipped. (Positive is down in ebiten-land and up in chipmunk land.)
February 16, 2025 at 3:23 PM
So, annoyingly, my stand-in game engine (ebiten) positions based on object center whereas my stand-in physics engine (chipmunk) positions things based on upper-left-corner.
Also, the Y axis direction is flipped. (Positive is down in ebiten-land and up in chipmunk land.)
Also, the Y axis direction is flipped. (Positive is down in ebiten-land and up in chipmunk land.)
Got collisions re-working. That was fun.
Now, instead of moving forward on that solution, going to use a well-tested library as people much smarter than me have already done this work: github.com/slembcke/Chi...
Now, instead of moving forward on that solution, going to use a well-tested library as people much smarter than me have already done this work: github.com/slembcke/Chi...
GitHub - slembcke/Chipmunk2D: A fast and lightweight 2D game physics library.
A fast and lightweight 2D game physics library. Contribute to slembcke/Chipmunk2D development by creating an account on GitHub.
github.com
February 7, 2025 at 6:37 PM
Got collisions re-working. That was fun.
Now, instead of moving forward on that solution, going to use a well-tested library as people much smarter than me have already done this work: github.com/slembcke/Chi...
Now, instead of moving forward on that solution, going to use a well-tested library as people much smarter than me have already done this work: github.com/slembcke/Chi...
The joys of coding with LLM support.
While the entire game logic can (currently) fit within a tiny context window, it doesn't seem to retain any context about how bounding boxes are drawn. (Corner vs center positioning.)
Time to refactor (the refactor) by hand.
While the entire game logic can (currently) fit within a tiny context window, it doesn't seem to retain any context about how bounding boxes are drawn. (Corner vs center positioning.)
Time to refactor (the refactor) by hand.
February 7, 2025 at 5:52 PM
The joys of coding with LLM support.
While the entire game logic can (currently) fit within a tiny context window, it doesn't seem to retain any context about how bounding boxes are drawn. (Corner vs center positioning.)
Time to refactor (the refactor) by hand.
While the entire game logic can (currently) fit within a tiny context window, it doesn't seem to retain any context about how bounding boxes are drawn. (Corner vs center positioning.)
Time to refactor (the refactor) by hand.
Here’s an idea: code the engine such that it overlays over whatever content you’re viewing.
If it’s a streamer, maybe allow players to play each other who would otherwise be “in chat”.
There has to be a way to engage an audience in a game while watching a game.
If it’s a streamer, maybe allow players to play each other who would otherwise be “in chat”.
There has to be a way to engage an audience in a game while watching a game.
February 5, 2025 at 1:44 AM
Here’s an idea: code the engine such that it overlays over whatever content you’re viewing.
If it’s a streamer, maybe allow players to play each other who would otherwise be “in chat”.
There has to be a way to engage an audience in a game while watching a game.
If it’s a streamer, maybe allow players to play each other who would otherwise be “in chat”.
There has to be a way to engage an audience in a game while watching a game.
More research on collision detection:
My initial thinking was to have everything in the world be a spline or a polygon so you could get really excellent, high-fidelity interactions between players, utilities, and the world. BUT, turns out that function/spline based collision detection is expensive.
My initial thinking was to have everything in the world be a spline or a polygon so you could get really excellent, high-fidelity interactions between players, utilities, and the world. BUT, turns out that function/spline based collision detection is expensive.
January 31, 2025 at 3:23 PM
More research on collision detection:
My initial thinking was to have everything in the world be a spline or a polygon so you could get really excellent, high-fidelity interactions between players, utilities, and the world. BUT, turns out that function/spline based collision detection is expensive.
My initial thinking was to have everything in the world be a spline or a polygon so you could get really excellent, high-fidelity interactions between players, utilities, and the world. BUT, turns out that function/spline based collision detection is expensive.
Required viewing.
Convex Polygon Collisions #1
YouTube video by javidx9
www.youtube.com
January 31, 2025 at 1:33 PM
Required viewing.
Shared a bunch of my design thinking to a bunch of 30-50 year olds. Granted, these are all tech nerds, but every single one of them got my references/influences I was considering incorporating.
Nostalgia is a powerful drug.
Nostalgia is a powerful drug.
January 21, 2025 at 6:33 PM
Shared a bunch of my design thinking to a bunch of 30-50 year olds. Granted, these are all tech nerds, but every single one of them got my references/influences I was considering incorporating.
Nostalgia is a powerful drug.
Nostalgia is a powerful drug.
Found myself thinking about startup screens, entry menus, and the like.
I get that there’s some fun in the build-up to being able to play the game while the game loads… but if the game engine is lean enough, there’s… not much to have to load.
Why not just go straight to game rather than mash ESC?
I get that there’s some fun in the build-up to being able to play the game while the game loads… but if the game engine is lean enough, there’s… not much to have to load.
Why not just go straight to game rather than mash ESC?
January 12, 2025 at 2:40 PM
Found myself thinking about startup screens, entry menus, and the like.
I get that there’s some fun in the build-up to being able to play the game while the game loads… but if the game engine is lean enough, there’s… not much to have to load.
Why not just go straight to game rather than mash ESC?
I get that there’s some fun in the build-up to being able to play the game while the game loads… but if the game engine is lean enough, there’s… not much to have to load.
Why not just go straight to game rather than mash ESC?
Haven’t had a moment to code this out, but current thinking for frame-by-frame collision detection:
- all world entities report pos, vel, type for current AND intended next position solipsistically
- collision detection checks for intersections and course corrects entities by reassigning pos, vel
- all world entities report pos, vel, type for current AND intended next position solipsistically
- collision detection checks for intersections and course corrects entities by reassigning pos, vel
January 6, 2025 at 12:01 PM
Haven’t had a moment to code this out, but current thinking for frame-by-frame collision detection:
- all world entities report pos, vel, type for current AND intended next position solipsistically
- collision detection checks for intersections and course corrects entities by reassigning pos, vel
- all world entities report pos, vel, type for current AND intended next position solipsistically
- collision detection checks for intersections and course corrects entities by reassigning pos, vel
Today: Total failure.
Reframe: Learned that my player inputs and movement intents need a bit of a refactor to accommodate the high-precision, high-speed collision detection I’m going for.
Tomorrow: refactor!
Reframe: Learned that my player inputs and movement intents need a bit of a refactor to accommodate the high-precision, high-speed collision detection I’m going for.
Tomorrow: refactor!
December 31, 2024 at 5:20 PM
Today: Total failure.
Reframe: Learned that my player inputs and movement intents need a bit of a refactor to accommodate the high-precision, high-speed collision detection I’m going for.
Tomorrow: refactor!
Reframe: Learned that my player inputs and movement intents need a bit of a refactor to accommodate the high-precision, high-speed collision detection I’m going for.
Tomorrow: refactor!
Collision detection work is fun I’ll tell you what.
December 31, 2024 at 4:20 PM
Collision detection work is fun I’ll tell you what.
Really liked the jettison/propelling mechanic. Incorporated that in.
Instead of just cutting the rope and letting momentum carry you through the air, player can propel off in the direction of the rope. Will play with this force and see how it works out!
Instead of just cutting the rope and letting momentum carry you through the air, player can propel off in the direction of the rope. Will play with this force and see how it works out!
December 31, 2024 at 1:58 PM
Really liked the jettison/propelling mechanic. Incorporated that in.
Instead of just cutting the rope and letting momentum carry you through the air, player can propel off in the direction of the rope. Will play with this force and see how it works out!
Instead of just cutting the rope and letting momentum carry you through the air, player can propel off in the direction of the rope. Will play with this force and see how it works out!
Today:
Got a demonstrable core mechanic coded. Nice. Still need to tune, but that will come later.
Refactored the already-difficult logic in preparation for more complicated logic.
Wrote out some TODOs before I forget; will be a few days before I get a nice chunk of time to work on this.
Got a demonstrable core mechanic coded. Nice. Still need to tune, but that will come later.
Refactored the already-difficult logic in preparation for more complicated logic.
Wrote out some TODOs before I forget; will be a few days before I get a nice chunk of time to work on this.
December 28, 2024 at 11:09 PM
Today:
Got a demonstrable core mechanic coded. Nice. Still need to tune, but that will come later.
Refactored the already-difficult logic in preparation for more complicated logic.
Wrote out some TODOs before I forget; will be a few days before I get a nice chunk of time to work on this.
Got a demonstrable core mechanic coded. Nice. Still need to tune, but that will come later.
Refactored the already-difficult logic in preparation for more complicated logic.
Wrote out some TODOs before I forget; will be a few days before I get a nice chunk of time to work on this.
This was surprisingly annoying to get working. #maths
(Player input forces more aggressive than will likely make the final cut, but it makes it easier to stress-test.)
(Player input forces more aggressive than will likely make the final cut, but it makes it easier to stress-test.)
December 28, 2024 at 10:11 PM
This was surprisingly annoying to get working. #maths
(Player input forces more aggressive than will likely make the final cut, but it makes it easier to stress-test.)
(Player input forces more aggressive than will likely make the final cut, but it makes it easier to stress-test.)
Idea: discarded ropes should persist on-level. (Especially if unsuccessfully attached.)
Hilarious.
(Do discarded ropes pile up and fuse with level terrain? Maybe!)
Hilarious.
(Do discarded ropes pile up and fuse with level terrain? Maybe!)
December 28, 2024 at 3:10 PM
Idea: discarded ropes should persist on-level. (Especially if unsuccessfully attached.)
Hilarious.
(Do discarded ropes pile up and fuse with level terrain? Maybe!)
Hilarious.
(Do discarded ropes pile up and fuse with level terrain? Maybe!)
Busy day—started from scratch
Scrapped all prior work to restructure project. (Previously a single main.go file… opted to rebuild it on ebitengine patterns.)
Got halfway.
Added a camera.
Also, seems I will still need to spoon-feed LLMs on physics and collision logic. (4 hours lost.) Witness:
Scrapped all prior work to restructure project. (Previously a single main.go file… opted to rebuild it on ebitengine patterns.)
Got halfway.
Added a camera.
Also, seems I will still need to spoon-feed LLMs on physics and collision logic. (4 hours lost.) Witness:
December 28, 2024 at 12:26 AM
Busy day—started from scratch
Scrapped all prior work to restructure project. (Previously a single main.go file… opted to rebuild it on ebitengine patterns.)
Got halfway.
Added a camera.
Also, seems I will still need to spoon-feed LLMs on physics and collision logic. (4 hours lost.) Witness:
Scrapped all prior work to restructure project. (Previously a single main.go file… opted to rebuild it on ebitengine patterns.)
Got halfway.
Added a camera.
Also, seems I will still need to spoon-feed LLMs on physics and collision logic. (4 hours lost.) Witness:
December 27, 2024 at 9:11 PM
Oh! Worth mentioning: most of my working knowledge on the core tenants of video game software and hardware engineering can be attributed to onelonecoder.com and gamemakerstoolkit.com.
Thank you, gentlemen, for the excellent content.
Thank you, gentlemen, for the excellent content.
December 27, 2024 at 5:27 PM
Oh! Worth mentioning: most of my working knowledge on the core tenants of video game software and hardware engineering can be attributed to onelonecoder.com and gamemakerstoolkit.com.
Thank you, gentlemen, for the excellent content.
Thank you, gentlemen, for the excellent content.