Decker VS advent_of_code

Everything in web-decker is drawn to a canvas using a custom software-rendered imgui-style UI toolkit. I considered this approach essential for maintaining fidelity across platforms and implementations, but it does come with notable accessibility and performance tradeoffs.

There is a native version of Decker implemented in C using SDL2. It's fairly easy to build from source for Linux[0], and I offer periodic Windows and MacOS builds on itch.io[1].

[0] https://github.com/JohnEarnest/Decker#building

[1] https://internet-janitor.itch.io/decker

Part 1 had a massive delay since i found a parser bug, but otherwise this was smooth sailing. Lots of off by ones, and luckily the example output was very helpful with that.

Here's the relevant extract from my (recursive) solution. Full code on GitHub.

Like others, I used lcm. Here's an extract of the solution, omitting parsing and main(). Full solution on GitHub.

Solutions to parts 1 and 2 nearly identical and pretty much worked first time. Nothing clever here. This was far simpler than I thought it would be. I probably spent most time trying to think of a mathematical way of getting all points between p1 and p2. In the end, I just constructed two ranges. Full code here.

Full code on github.

The core of my solution below using numpy (because I'm learning it!). See github for full code.

Python 3 solution using numpy. I got held up because I assume (but know better) that numpy arrays are [x,y] instead of [row, col], which is [y,x].

Below is the new solution, which works for part 1 and 2. The full code is on GitHub.

These are the key functions. The whole code is on GitHub

Part 2 (extract shown below) was a matter of starting with the low points found in part 1 then recursively looking around for relevant points. Full code on GitHub

Code is on GitHub.