I wonder if we could achieve a thousand likes on this comment of yours?

Black and Decker

One thing with the if can_harvest() is that the way youre doing it now you just skip over whatever isnt ready even though it will probably be ready soon so it will be at full growth for a while wasting growing time. It also usually works out in a way where you end up getting to the other crops faster that also havent been grown yet. So my conclusion here is that its better to wait until its ready for harvesting in that spot. Now I think the game wants you to use the do_flip thing there but what actually works best in my experience is doing this: while not can_harvest(): quick_print("waiting") harvest() Who knows maybe you end up doing something like this later in the video its so long :) but if not i hope it helps! edit: you do have to check can_harvest but you dont have to with planting pumpkins, if there is already a pumpkin you dont use seeds and nothing happens.

We need Rusty's Retirement python coding

it says 2 days PAY UP 😅

when i was little, like 18 years ago, we used robot karol in school for a short time. pretty similar concept but you only can pick up and place blocks and markers. even has a free online version

I work tech in a school district and actually showed this to our department who looks into software to use.

That's how he gets you to buy the games.

Welcome to coding. "This should work. Why doesn't it work? What did I even do now? Why does it work now?! Don't touch anything."

It works on my local...

This is LITERALLY EXACTLY WHAT I THOUGHT OF TODAY haha. Gonna build that next episode

@@OlexaYT Haha amazing! Very much looking forward to that one then. There's some cool optimisations you can do in the goto() to make it decide whether to wrap around or not :) Good luck and loving this series!

Its weirdly actually not that much more efficient, I have that set up and compared them and you get maybe 25% more every minute. Which is great but stil weirdly underwhelming

​​@@lordthor5951A 25% gain might be quite significant. Let's say that an improvement in something reduced the number of cycles from 5 to 4; that's approximately a 25% gain. In this case, the initial loop and planting take quite some time, which is the same either way. That means the actual gain except the initial loop is even more significant than 25% gain. Even though the drone does not teleport and has to pass the grids, the efficiency gain is less significant. Also, it takes a while for a failed pumpkin grid to regrow. The drone keeps moving to all incomplete grids until they are completed more efficiently, but it's still limited by the growing time.

@@only2sea to put it more into perspective, my current speedrun time for the speedrun function in the game is around 40 minutes, I think i spend 4? minutes of that time on pumpkins. Compared to the naive solutions on the other harvests it also makes it pretty clear how its not that big of an improvements, my maze solution is shitty and like a third the speed of the fast maze solvers. My Cactus sort is pretty close to perfect and maybe double the speed of some of the slower but still competent cactus solves. The 25% on pumpkins are nice but weirdly not that great of an improvement. (Also for talk about the initial loop the 25% is based on pumpkins per minute, on a 10x10 grid, you go roughly from 80k to roughly 100k, the initial loop doesn't really matter in this)

Thank you, I needed to hear this. I actually do drafting for a living, but have some coding I've done to help be more efficient. The policy vs mechanism point really sounds helpful for how to approach a rewrite that I've been spinning my gears on.

pretty sick

Check description!

I don't think that changes anything as its a while loop and will only check if checks is = total_grid at the start of a new loop, which only happens at the top left The idea is good and it would make it more efficient but i think implementing that would require more changes

I think what you want to do is to count up how many pumpkins you have seen since the last planted one? Of course that'll still go through the entire field unless you make a special wait and check loop for the last pumpkin that was missing.

I honestly want to set up doing pumpkins like that, and then setting up a timer to see which way is faster. Bc I thought about that too but the growth just seems too slow

@@OlexaYT Idk when it gets unlocked, but at some point you get a new menu called "Stats" that measures your collection over the last 60s. That does what you want to.

Aren't the functions of the game like classes? I have no clue about coding, but I thought a class is simply a collection of code that always requires predefined variables. If no, what's the difference?

@@Solanaar Disclamer : Junior software engineer here so my knowlege and expirience is limited. But my 2 cents on this is Class is just a chunk of code that creates an Object in memory. Now objects are useful because they have their own variables and functions that you can access when you need. Think of it as an item in RPG, that item has it's stats (variables) and can do things like attack (which you would use do_damage function there). Drone for example is an object, it has its own stats (speed for example) and functions (harvest() for example). Coding for drone in the background of the game would be a class. Not sure what Olexa was thinking to do with classes but my guess would be to define the grid as class and then do some stuff with that, or perhaps define pumpkin as a class and then track how big it is untill it reached desired size. I guess. I dont know, my young junior programer brain is probably not developed enough to comprehend XD

@@OlexaYT It gets more effective the larger your field size. The time and/or fertilizer cost scales linearly with field size, but the harvest amount scales exponentially. Eventually the exponential term dominates and the linear term's contributions can be ignored. I'm not saying it's the optimal solution, but it's good enough for for me to not want to optimize further.

Mazes were really hard for me, but technically if you're willing to wait a bunch of time for every solution, you can solve them by making a while loop that moves the drone in a random direction and then stops when it's on the treasure

i just made my drone stick to the RHS as much as possible. similar to how i would solve a maze irl. only tricky thing is knowing the direction you came from as that affects what’s on the RHS. but you can do it with just lists. it’s not the MOST efficient but it’s not random and it does it as quick as i could think of

I'm stuck at mazes currently, I'll clean up the rest of my code before trying that one

The maze is damn complex. So pen and paper design a flow chart first, then convert into pseudo code, then write the real code. Otherwise there are just too many unexpected things to think about...

Yeah, the Maze was horrible. A naive "LHS' (or RHS) rule will just get you stuck in funny corridors and make no progress. I had to make about 3 distinct attempts before I finally got something semi-decent and actually reliable. I genuinely wish there was a can_move(dir) I could use to test if a path is blocked, instead of doing the solution I settled on of moving them moving back to test for forks in the path. For my last attempt, I made a recursive solver. Basically every time I detect more than 1 path (excluding the direction the drone came from), it will recurse the solver down each path. If there's 1, it just travels. And if it reaches a dead end, it just undoes all the steps that specific iteration did (by popping a list), so the caller can try the other forks. It still got stuck in loops, so I tracked the already visited tiles in a set once i unlocked those... There is another possibly faster one I want to try, but it basically would just make every step recursive, instead of only when reaching a fork. There's also some funny tricks you can do with dictionaries to make a path resolver, if you refertilise the treasure to solve the maze again (since you know where it warps to), but that requires contingencies if the maze moves walls.

this game was my first introduction to coding and it took me 5ish hours of failure to get a working code for mazes

@@starwheels8648 I remember spending 5 hours on a level in Shenzhen IO then deciding the game wasn't for me. So far I like how this game gives a bit more freedom and simultaneously has something very close to the real coding language it's based on.

I think that’s definitely an approach. I kinda like the all or nothing more so I can specialize quickly in certain crops but it’s clear there’s many routes to success!

@@OlexaYT With the ability to clear the grid so quickly, why choose anyways? 😂 I should build out more specialized functions for targeted upgrades, I've just been farming everything at once in the background while I read on my other screen

Literally same!

JavaScript 🤮

I think it’s less efficient because there’s a LOT of dead time waiting

@@OlexaYT possibly but imagine you have a big board. you are doing a full cycle every single time there is a failed check. your goto solution still has to be the best tho

Beauty of code, lotta solutions for something!

Definitely an issue for both, it’s just something I’m not too concerned about. It’s an extremely minimal concern in the long run

Yup, it’s a good catch. It’s fixed in next epi bc I redid that function

Yeah the plantable grid is not wrapping even though the drone can wrap.

I think you’re overcomplicating things haha

@@OlexaYT most like yes

@@OlexaYT most like yes

Yup, this is basically the only way. I have an idea but it’s super ugly haha. Might write it offline and show it to you guys next time

I think a minX/maxX minY/maxY bounds narrowing would probably do it well enough, scanning smaller subsets of the farm.