ah, I exclude loading and reading the file. but since you are pasting it from pasting into the terminal, that is alright.

My main gripe is that I am looking at the performance of the algorithm/functions over the performance of the disk access and read, the startup/end overhead. Python is notorious in having overhead during startup for loading the code and before execution occurs. Why should I measure the performance of the language too harshly? I rather look at how my code performs. In windows, the python overhead adds 30-40 ms, while on Linux, it performs faster with only an overhead of consistent 20 ms. Though, that is just without importing heavy or many libraries. If startup is a concern, then a precompiled non-interpreted language is a better option.(along with the other benefits) This is my reasoning for only measuring my algorithm. I do include parsing the input as that is part of the challenge, but I do see there are reasons not to do that. when you are looking for code that is performant, you want to scientifically remove too many variables. If you are to reuse some code, lets say the input to that function is already parsed and then you just want performance. However, I do measure my parsing because for the AoC, I want to think about what would be faster to parsing and what is a bad parsing.

For AoC, I find a language overhead is not part of the challenge. we should rather learn new languages when we want or use what is comfortable. however, languages like Uiua with a lot of specialty functions is just not worth measuring performance as the main code is just a simple “function call”

I am sure there is a python package/module that includes a fast path finder, too. I just want to challenge myself mostly to learn instead. however, I am finding I would need to start learning rust instead, because my python skills are starting to plateau.