advent-of-code

[woven sable]

bc there's no benefit for the ||J's to be different if there's more than one||

[noble skiff]

i mean it's a lot simpler than that, ||all the jokers should always be the same as your most frequent card||

[dense osprey]

oh yea but then it would be greedy and not purely naive anymore

[ripe meteor]

This is the one thing i checked actually

||there werent any||

[woven sable]

true

[noble skiff]

it's pretty easy to convince yourself that's true; imagine yourself with any given hand, add a joker. what the joker should be is always the same ||the most frequent card you already have||

[simple raven]

then you still need to ||sort to always choose the bigger of the two in the case of KKTTJ|| right?

[noble skiff]

no, because it doesn't matter

[dense osprey]

yea you can just prove this by showing that it wouldnt make sense for all cases

[simple raven]

oooooh it doesn't because ||we are not replacing|| omfg I made it soooooo much harder for myself in so many ways with my first solution...

[noble skiff]

yeah

[dense osprey]

im too lazy to implement this tbh

[noble skiff]

I think it's pretty deliberately designed to be this way; to actually make the joker substitution very simple

if there were real poker hands, or if the joker actually became teh card for tie breaking purposes then there would be at least some logic around choosing what to make the joker

as it stands there's zero; I never actually explicitly choose what the jokers will be in my code

[woven sable]

oops idk why i went the naive way

whb i could've j done that

[dense osprey]

why were todays and yesterdays problem so easy compared to the problems before

[torn sapphire]

the questions are all laid out by a single dude, and noones perfect at laying out problems 🤷

[noble skiff]

if you consider that day 5 part 2 worked with brute force, i'm not sure there really was a big difficulty spike

doing day 5 part 2 "properly" has been easily the hardest thing so far but it's not necessary

[woven sable]

u could brute force it??

[torn sapphire]

according to the stats page, day 5 does have less 2-star completions than day 6

[woven sable]

i got a java heap error 💀

[simple raven]

witness my silly solution
||```py
def replace_jokers(self, hand_type, hand):
if 'J' not in hand:
return hand_type, hand
card_count = Counter(hand.replace('J', ''))
#case 3333J
if hand_type == '4':
hand_type = '5'
#case QQQJA
elif hand_type == '3':
hand_type = '4'
#case QQQJJ or JJJ22
elif hand_type == 'fullHouse':
hand_type = '5'
elif hand_type == '2':
#case KKTTJ
if all(x[1] == 2 for x in card_count.items()) and len(card_count) == 2:
hand_type = 'fullHouse'
#case KTJJT
else:
hand_type = '4'
#case 2388J or 23J8J
elif hand_type == '1':
hand_type = '3'
#case 2348J
elif hand_type == 'highCard':
hand_type = '1'
return hand_type, hand

[dense osprey]

yea but i dont think that the bruteforce solution was the way it was supposed to be solved

[noble skiff]

although I guess to brute force 5p2 in python you need to at least use multiprocessing, or do the clever/||reverse|| brute force

yes, I agree

[dense osprey]

while todays problem is supposed to be solved in a naive/greedy way

[woven sable]

what's greedy 💀

[noble skiff]

you can't pick the difficulties perfectly

[royal gull]

Yah, my d5p2 was brute force.

[woven sable]

i bet u tmrs gonna be a pathfindinh or flood fill problem

it's always around this time

that we get the first one

[noble skiff]

also in the end I think difficulty is more oriented around completion than doing it the way you're "supposed to"

[dense osprey]

wait how do i do multiline spoilers

||x
x
x||

[simple raven]

put the lines before and after the backticks

[royal gull]

Spoilers should really go in the daily spoiler channel

[noble skiff]

i think that part of the design in 5 is that the actual input size isn't huge (depending how you look at it), which gave brute force a chance to work.

[woven sable]

are u talking about a diff day

oh wait nvm

the size itself wasn't big

but the numbers were hugee

made brute forcing each ||individual seed|| hard

[royal gull]

Can you put this in the daily solution thread? Not here plz

[noble skiff]

greed just refers to algorithms that make the best "immediate" move.
Like when you try to find a minimal spanning tree for a graph, there's a greed solution to always pick the cheapest path that connects an unconnected node.
it's not obvious by inspection that this "greedy" algorithm is optimal, but it actually is

[dense osprey]

ok

[zenith elbow]

a technique where you choose to do something that gives you a better answer first. like in finding elements in an array that add up to a sum, you could say to pick the largest number smaller than given sum and go from there

[woven sable]

ohh

i see

[noble skiff]

there isn't really a good example yet this year of a greedy algorithm, except maybe in a trivial sense

[woven sable]

i struggled w part 1 bc i spent 10 min tryna make a custom comparator in kotlin bc u can't compare lists

[noble skiff]

unless i'm forgetting something

[woven sable]

and then i j manually checked each index

[zenith elbow]

well the problems so far are all doing some processing for each input rather than eg finding an optimal way to arrange inputs. tbh for day 5 i considered ||bruteforcing from highest mapped location and reverse map it|| but couldn't think of an efficient way to handle ||unmapped locations||

[noble skiff]

yeah, there hasn't really been any optimization problems which is where you see greedy

[runic drift]

Can anyone look at my day 5 solution and tell me why it’s not working

[muted aurora]

i can try to have a quick look, send it

(in https://discord.com/channels/267624335836053506/1181459209422917683)

[runic drift]

I’m what?

It says #Unknown @muted aurora

@muted aurora just sent it

[dense osprey]

nvm part 2 required me to only add like 10 lines to my code

huh how did this code even get the correct answer

[rancid creek]

how hard was day 5? chart of every top 100 submission for every year in aoc:
https://www.maurits.vdschee.nl/scatterplot/

[dense osprey]

this code shouldnt work for the hand JJJJJ

but it somehow does

[simple raven]

so like 15 last year, 8 or 16 2 years ago and 16 or 19 3 years ago, cool

[zenith elbow]

what's with the day 1 spike in 2015 I wonder

[modern terrace]

i assume it's related to AoC 2015 being the first aoc event

[jovial jay]

what with the 2018 outliers?

actually, too afraid to think about it, it's probably bad, very bad

ok, i see 2015 outliers, now i need to take a check at that problem

2015 day 19 doesn't look too bad. still nothing like day 5 of this year i think

[honest nacelle]

I ended up ||counting up one by one from 0|| lol

[noble skiff]

Worth noting for anyone that looks at this that they don't keep the y axis scale fixed

So the plot is extremely misleading

Although day 5 certainly still seems to be an outlier

[woven sable]

u can fix the y axis

[mossy basin]

why axes?

[honest nacelle]

This is pretty cool, watching a leaderboard dev solve in realtime. Are there others who do this higher on the leaderboard?
https://www.youtube.com/@nthistlethwaite

[rustic cobalt]

https://www.youtube.com/channel/UCuWLIm0l4sDpEe28t41WITA

[jovial jay]

Can we do AGC assembly for next roulette?

[honest nacelle]

never would've expected a leaderboard dev to use idle

[honest nacelle]

danke

[rustic cobalt]

pretty amazing stuff

what if I want to read the little elven stories?

..or maybe they do..

[jovial jay]

Guess I'm bad at jokes

[honest nacelle]

Honestly yeah, I def get the alure of the leaderboard, but knowing that I'll never crack it, I like going at a liesurely pace and enjoying the story

[minor cave]

Language Roulette: Day 8
The language is ... Bash

[mossy basin]

oof

[magic oxide]

damn bash

hope today isn't too hard

[mossy basin]

gonna bash your keyboard

[honest nacelle]

f

[earnest forge]

Good luck everyone

[honest nacelle]

🫡

[earnest forge]

you know, bash might be less painful than prolog lol
I'm not doing the language roulette but still

[honest nacelle]

Who has a prolog solution up? I wnna see it

[zenith elbow]

!! maybe I should start doing the language roulette from today, good thing it isn't bourne shell

[shut trellis]

<@&518565788744024082> Good morning! Day 8 is ready to be attempted. View it online now at https://adventofcode.com/2023/day/8. Good luck!

[magic oxide]

ooh nice

finally one of these problems

glhf leaderboarders

[honest nacelle]

I can tell this is gonna be a CPU-killer in part 2..

[pine tiger]

what are the odds

same rank both parts

[feral hazel]

31 / 10 wooo

[jovial jay]

Leaderboard is now filled?

[feral hazel]

spoiler channel isn't up?

[pine tiger]

not yet

[jovial jay]

I too like going at leisure rate. I find problems hard to understand at first, and I'm challenging myself to use a non-generic language to solve it.

Closer to solving day-5 now 🙂

[timber glacier]

how. part 2 is insane

how many steps did your part 2 take

[pine tiger]

it's just ||lcm||

[timber glacier]

how

[untold briar]

Can someone pls explain this part for me

Of course, you might not find ZZZ right away. If you run out of left/right instructions, repeat the whole sequence of instructions as necessary: RL really means RLRLRLRLRLRLRLRL... and so on. For example, here is a situation that takes 6 steps to reach ZZZ:

[pine tiger]

loop around if you reach the end

[mossy basin]

my best finish

[untold briar]

I see clearly that you just need to do LR

How did that become 6 steps?

RR works too

[timber glacier]

the second exmaple is 6 steps

the first example is 2 steps

[untold briar]

Yeah and the 2ns example confuses me

Start at AAA, go either left/right to reach BBB

[pine tiger]

||do you have to consider all the nodes at once?||

[untold briar]

Then go right to reach ZZZ

How did that become 6 steps?

[timber glacier]

isn't that what it's asking though. Make sure all end with Z

[honest nacelle]

If the instructions in the first line don't get you there, you repeat them until they do

[pine tiger]

i just realized i made a crazy assumption that worked out

wait how did that work what???

[timber glacier]

bro

what

[untold briar]

Hmm yeah I would punch the guy who made this map

[timber glacier]

idk what you're trying to do with ||lcm||

[untold briar]

Why do you make me take 6 turns in a desert when I can get out of it in 2 turns

[compact wasp]

i started late again oops

[timber glacier]

my code has already gone through 1,000,000,000 steps and more for part 2

[woven sable]

i could've gotten on the leaderboard lowk for part 1

too lazy to open my comp th

[timber glacier]

same

[woven sable]

so i j read the prompt

[timber glacier]

i was too slow to understand

[woven sable]

i'm worried ab part 2 tho

[compact wasp]

and i completely forgor to do it yesterday

[woven sable]

ik it's some brute force thing

tired of these problems 😭

[honest nacelle]

I kinda like them, make you think a bit

[compact wasp]

well… it's ||lcm||

[timber glacier]

bruteforcing is taking forwever, I don't think it's a valid way of doing it

[woven sable]

some of them are fun fs

but day 5 was so bad for me idk why

[timber glacier]

how is it ||lcm|| though? That makes no sense

[timber glacier]

I still have no clue how to get d5 p2 to work. I got the answer that worked for my input but not the example data for some reason...

[honest nacelle]

I thought I could relax but there's a guy on my private leader board who also has part 1 I'mma panic

[compact wasp]

||order of products of cycles||

[timber glacier]

that's making no sense

[stable shale]

:)

[compact wasp]

you're talking about part 2 right?

for day 8?

[timber glacier]

oh

wait

[compact wasp]

[timber glacier]

OH I GOT IT

[timber glacier]

||lcm|| helped, this didn't...

kik'

[zenith elbow]

if we imagine each ghost to start on a node ending with A, ||the number of steps (let it be X) needed for each of them to get to Z is independent of each other||, and ||once they get to Z, when we continue following instructions they just keep cycling every X steps||

[timber glacier]

too late

[zenith elbow]

:')

[humble copper]

need a general hint for part 2

[timber glacier]

||lcm||

i don't understand how it works, but it does

because how would you know if the ||Z makes that node start all over again instead of going for a few more then starting again||

[humble copper]

i nearly had it cuz ||i realized the lengths for all ghosts to find a Z node was easily divisible by the length of the instructions||, never occured to me it could be more than that

time to implement it without the math lib, shouldnt be too bad

[timber glacier]

why without?

[humble copper]

tryna do no imports

[flat sequoia]

It doesn't work in the general case

the inputs are just generated so that it works

[timber glacier]

thought so

[flat sequoia]

you can just explicitly check for a cycle instead

[covert arrow]

AHHH I FORGOT

I am at a swim meet

[compact wasp]

rip

[covert arrow]

and it just happened that my heat is right when aoc opens

[compact wasp]

i also realized that the code i wrote wasn't working for automatic example parsing 💀

[covert arrow]

I did get a 50 free pb though (29.19)

[zenith elbow]

nice one!

[humble copper]

niceee

[zinc kindle]

Right. Psrt1 was easy. Assumed part2 would involve finding the shortest paths, was glad I was using rust because of the 3rd party pathfinding module. ||I was wrong||
After some time in part2 ||trying to convert it to using a vec and sorting it, then hashset, then finding it was taking too long|| realized I just needed ||lcm of all the times (well, used GCD, realized answer was obviously wrong, replaced with LCM and it worked). Unfortunately, rust doesn't have lcm included, so I needed to import it, and a short function using reduce to use it on all elements||. Still, managed a significanlty better part2 rank than part1 (still not in top 1000 though).

[naive stone]

there's some kind of trick to this isn't there

[covert arrow]

wait how does ||lcm|| just work?

[compact wasp]

eric wanted to be cool

i should've gotten onto the leaderboard but i didn't bc i didn't think ||lcm|| would just work

[pine tiger]

it shouldn't lmao

[zinc kindle]

||my intuition/hope was that the time for start to dest1 would be time for dest1 to dest2. In retrospect, this might have not been guaranteed, but it worked. Also, it turns out that gcd of all times !=1, so that might be related||

[pine tiger]

what's dest1/dest2?

[covert arrow]

oh wait, so ||all the endpoints are in one big cycle||?

[zinc kindle]

||first valid destination reached, second valid destination reached||

[untold briar]

Wow 1st place today got 2 stars in 4mins

[candid brook]

I feel like the input is specifically constructed so that ||lcm|| would work. I was thinking about using ||Chinese remainder theorem|| if the ||initial number of steps to reach Z|| is different from the ||loop size to reach another Z||, but apparently they're the same

[untold briar]

So just 4 mins to do everything, including writing a non-brute forcing solution

[slate berry]

can you elaborate

[covert arrow]

#1182550042091991180 message this is true for all cycles, so it definitely seems intended

[deep ferry]

Day 8 part 2 is actually so stupid

[deep ferry]

it's a hidden assumption

so like how am i supposed to know that it would work? like technically it should have never worked, no?

unless im missing something stated in the problem statement

in that case, please enlighten me

[covert arrow]

you could have figured it out after looking at the values

[deep ferry]

but that's still assuming

i never want to assume anything by looking at the values. ig im too used to usaco being explicit about everything

[honest nacelle]

Oh god

I made it

[covert arrow]

Yup, aoc is a completely different style from usaco and other cp competitions

Almost anything is fair game since you have the whole input (except ai ig)

[honest nacelle]

wait, is there a situation that wouldn't work in?

[deep ferry]

what if there wasnt any cycles

what if it was an acylic graph

what if each "cycle" length was actualyl different

[covert arrow]

I haven't seen anyone's input not do that, so I'm confident it's designed that way

[brazen bridge]

How is it even possible to solve part 1 in 38 seconds?

[deep ferry]

it's like coding it while reading it ig

[untold briar]

Ah yes I love today's problem haha

[honest nacelle]

Ooooh... yeah actually

[untold briar]

It makes brute force totally impractical

[deep ferry]

now i think of it, possibly AI

[deep ferry]

i dont like today's pt2

[honest nacelle]

@deep ferry I see what you mean now, ||I assumed the graph cycled immediately after reaching an end node and that happened to be correct||

[untold briar]

why

[deep ferry]

#advent-of-code message

[pine tiger]

relies on input assumptions

[deep ferry]

yea

[pine tiger]

fellow usaco enjoyer

[covert arrow]

Well, not quite but close enough

[deep ferry]

grinding for the college app

dont get me wrong i love programming

and USACO's not bad

[honest nacelle]

What's off there?

[deep ferry]

but i'd rather be making projects

[covert arrow]

||the number of steps before reaching the loop is equal to the number of steps between the end and the start of the loop, so they cancel out||

#1182550042091991180 message

[covert arrow]

I despise usaco

[deep ferry]

eh W ig

[covert arrow]

And c++ by extension

[deep ferry]

oh

try rust

[untold briar]

Well C++ is very fast

But the syntax, well...

[covert arrow]

I like rust more

[deep ferry]

Rust >>

maybe even better than Python

the one thing i would complain about python is the implicit pass-by-references vs pass-by-value

[untold briar]

Also the code block defined by indentation

Could get messy when your code get big

[deep ferry]

https://www.reddit.com/r/emacs/comments/3m6yhd/fibonacci_indentation/

[untold briar]

Wow, great idea! Prevent anyone from editing your code without great work or great disruption! Job security! 😉

[slate berry]

is there private leaderboard for this server?

[deep ferry]

yes

[slate berry]

code?

[untold briar]

No thanks, I already have enough trouble between tab and 4 spaces 😅

[deep ferry]

i cant find it though

[slate berry]

@minor cave can you help

[deep ferry]

@placid lake ? the bot used to show the link but now it doesnt... hope this is the right person to ask

[untold briar]

See the aoc-faq channel

But there could be other codes?

[minor cave]

run /aoc join

it'll give you one

[deep ferry]

ah

[untold briar]

That Dennis Pham leaderboard is not very active

[minor cave]

we technically have 9 leaderboards, because we have more people than a single lb can fit

[deep ferry]

oh so the &aoc and slash commands are different

[untold briar]

Only 20+ ppl still remain in action

Most just joined and stay dead lol

[minor cave]

it used to be &aoc join and it would try to DM people. But using a slash command and having it be an ephemeral message is a much nicer solution

we also re-use the same leaderboards each year, so there are a few people that joined a previous year and aren't active this year

[deep ferry]

i joined it last year...

pruning inactive members would be nice maybe

[minor cave]

I might go through after this year and prune people who haven't done a single day on any year

but it's also a pita because I need the owners of each leaderboard to do that and that's a lot of cat herding

[naive stone]

i mean, like 80 people just did the first star for today. That's a good amount i'd say

[untold briar]

We are probably in different lb

[naive stone]

yes, there are like 9 of them

[untold briar]

Yeah so in mine, only 48 ever did a start

[covert arrow]

is there an archive of older years' leaderboards that I don't know about?

[naive stone]

oh were you only referring to the leaderboard you're on?

[untold briar]

And 20+ still go on after the disasters that are day 3 and 5

😂

[minor cave]

Not rn, but it has been on my list to do. Maybe I can get it this weekend

[untold briar]

Yeah, I cant see other lb anyway

[naive stone]

i'm actually on a few of them, i'm not sure why

[untold briar]

Where do you get the code for all of our lb?

[covert arrow]

just delete them all and make everyone join again

[untold briar]

You are brute forcing it

😏

[naive stone]

that's not the worst idea lol

[minor cave]

This is the current leaderboard output if anyone's curious https://paste.pythondiscord.com/6SHBR3SN6Y5NFR5NGSMNYGZCVE

Only has top 1,000 because of the paste service size limit

[naive stone]

i was curious, the lb command used to output this link

is there no way for us to get the full leaderboard anymore?

[minor cave]

the amount of cat herding I'd have to do to get the leaderboards recreated 😩

[minor cave]

it's still supposed to, but when we moved to the new paste service the feature broke. A fix for it should be merged in tomorrow

the link I have is from me testing the fix

[naive stone]

ahhh

[untold briar]

Will this page update automatically?

[naive stone]

i think pastes are probably immutable

[minor cave]

nah, but whenever Chris or Bella wake up and merge in the fix, the &aoc lb command will output a new link with updated results

[covert arrow]

maybe have an endpoint on pythondiscord.com redirect to the most recent leaderboard?

[naive stone]

submit a PR :)

[covert arrow]

I have no clue how that wouuld work lol

I might open an issue for it later

[naive stone]

contributing to our repos you mean?

cuz we have great documentation for doing that

[covert arrow]

I know how to contribute, but I don't know how I would implement it

[naive stone]

ah

[minor cave]

If someone wanted to spend sometime and do a cute little page with dynamic results, we store the leaderboard in our redis cache. So theoretically we could pull straight from that and skip the paste service altogether

[naive stone]

that would be cool

[covert arrow]

hmm interesting

[long jacinth]

theres no ||decision making|| in todays task right?

||its all just traversing a tree in a cyclic fashion||

[covert arrow]

yeah, if you're gonna store a url to the paste link it may as well just store the leaderboard itself and display it nicely

[minor cave]

anyway, at 2am it is time for bed

[naive stone]

careful about spoilers

[long jacinth]

isnt that just the assignment?

[naive stone]

i would argue that figuring out the ||cyclic|| nature of part 2 is part of the problem

[long jacinth]

ive only just started reading part 1, so i wouldnt know, if it is, im sorry

[naive stone]

oh gotcha, and no problem!

[slate berry]

just spoiler it lol

[naive stone]

seems you have a good eye for the problem then

[untold briar]

Yeah I got fumbled at the 2nd example of 1st part, ||cause I initially thought we must find the shortest route or sth||

[long jacinth]

maybe i should start to use libraries for this

so i dont have to make my own ||loops and trees||

although, i guess it doesnt take much time

[woven solar]

i don't think this problem would be very expressible in e.g. networkx.

[untold briar]

Wait do we even need those for today's challenge?

[long jacinth]

i dont know, but thats my solution

[naive stone]

i feel like i would struggle to adapt all these problems to networkx. Before i realized it was the wrong solution, I wrote a custom ||bfs|| that of course never finished, but the logic was definitely right

[long jacinth]

i mean, you dont

you can just make it in a || list or a dict, or whatever|| i guess

now that im thinking about it, it probably just needs a ||dict||

[woven solar]

yeah, I represented ||the graph as a [source, destinations] dict|| as I usually do

[long jacinth]

now that im thinking about it, i would probably just use a ||dict || while building my || tree||

guess that makes it redundant

unless its a P2 thing

[naive stone]

i gotta say, it's pretty hilarious how the bot just does not care about spoiler tags. Isn't that right bot? ||tree||

[long jacinth]

spoiler? hardly know'er

[ripe scroll]

|| santa ||

[rocky spire]

I'm assuming you're definitely not supposed to just simulate part 2 right?

[covert arrow]

||yes, you'll have to find a faster solution||

[austere basin]

yo guys, rouhgly how big is the answer to number 2?

like what kinda ballpark is it?

[rocky spire]

Well I have an idea, time to get mathy
though this is gonna be the first time I implement the related algorithm sooo...
To debug hell I go

[austere basin]

bc my programs currently at a billion... should i just stop it or is there hope?

[naive stone]

||there is no hope||

[austere basin]

aw man

what is it, like trillions??

[naive stone]

like ||ten trillion|| actually

[austere basin]

oh shit

welp, 0.2% of the way, LETS DO THISSS

[woven solar]

I dislike how many days of this year are like "this problem is actually much simpler than it would seem because the input is nice on purpose. nothing in the problem description reveals that, you'll just have to look at your input to find that out. If you didn't do that and started writing a general solution, well, sucks to be you!".

I generally like how AoC makes you learn to solve problems quickly, but I would prefer not learning how to solve problems incorrectly to be more quick.

[naive stone]

i'm not sure i understand your critique

i felt like today's prompt foreshadowed what the ultimate solution needed to be

[woven solar]

You mean, the example? I don't think so, as there are days on which the example fails to reveal the problem's complexity (e.g. day 1 this year), so "the actual input won't have caveats that aren't in the example" would be a bad lesson to learn.

[naive stone]

are you saying you would prefer the example contained all edge cases that are present in the real input?

[woven solar]

I think I mostly just want it to be consistent. As it is, if you want to be quick, you have to guess whether the caveats are there or not. E.g. in today's day, it'd be nice if the problem statement mentioned something like ||'the path always leads you back to the start eventually'|| - then it'd be unambigous what the solution is. As it is, today people who solve the general problem instead of a subcase of it (that the problem statement doesn't say it's limited to) will have just wasted their time. I think this is bad - it's teaching a bad lesson in how to generalize from examples to a solution.

[naive stone]

hmm, i'm not sure i agree. I think adding that would have been a big spoiler. You already can see that it does that in the small example. I don't think it'd be good to hit you over the head with that

[covert arrow]

||the path always leads you back to the start eventually||
fwiw this wasn't true for the inputs (it was a slight variation that doesn't change the outcome), but arguably finding the patterns in the input that lets you generalize to it is part of the puzzle too

[rapid wolf]

I disagree - I mostly want to write a solution should work for all inputs, not just the specific one I am given. There are cases where you can get lucky and not have some edge cases in the input that others do, like I did in d5p2.

[magic oxide]

after seeing part 2, I agree

tho I actually haven't done the puzzle yet

so my opinion may change

[magic oxide]

or like day 6, where some people had ||inputs where you could reach exactly d for two times, and some people like me didn't, so there was an entire edge case that we just skipped past||

[rapid wolf]

fun

probably my case as well tbh

[magic oxide]

I had a similar gripe 2021 day 24, but it was less bad because it was pretty obvious that brute forcing wasn't the method and you had to analyze the input

but i still didn't like it

[rapid wolf]

2021d24 I probably did via a constraint solver, so I wouldn't have noticed.

[polar fractal]

i just discovered advent of code yesterday and i got a question
on the first day i compute that number with my code and then what?
do i just put that number in that textfield online or do i upload my code somewhere and it automatically checks?
and is it the same for any day?

[rapid wolf]

just put the number in

[long jacinth]

you get your input from the website, process it yourself, and then upload your result to the text field on the website and hit submit, then part 2 will appear

[rapid wolf]

nothing checks automatically, you can solve some problems with a calculator

[simple raven]

you put the number in, your input is unique so it does the check for you (but your code is not checked)

[polar fractal]

ah ok thx

and its the same for every day?

[rapid wolf]

yes

[polar fractal]

👍

[simple raven]

yeah you just get one number as the output, the challenge is meant to be solvable with any language so it's flexible like that

[long jacinth]

any brainf*ck enjoyers?

[earnest forge]

No

[rapid wolf]

I dont think AOC is solvable in any existing BF interpreter

[earnest forge]

I mean there are jit/aot compilers

[rapid wolf]

I dont believe they are good enough

[earnest forge]

I remember that one of them literally just compiles it to a state machine in C lol

[rapid wolf]

even if you get their fancy jumps are O(1) rather than O(n), I would be surprised if they could get the same time complexity as a more conventional language.

but maybe I am out of date on optimised brainfuck

[polar fractal]

are we allowed to use any external packages like numpy or just builtins?

[simple raven]

anything and everything

you are just asked not to feed the problem into chatgpt

[polar fractal]

got it thx 🙂

https://tenor.com/view/charlie-woooo-yeah-baby-gif-24480925

got the first one 😄

[sleek cave]

don't tell me the input is literally ||an adjacency list||

[hidden musk]

@cyan sleet on the bright side, no language to install today 😩😩

[zenith elbow]

TIL, bash supports C style for loops (i = initial; i < cond; i += step), and maps!

but why is bash so slow even with the ||lcm|| solution

[fervent helm]

i agree that for this problem, non special input would be too hard, and giving the trick too easy

but ideally neither would be the case – it would just be a slightly different problem that's the right difficulty without simplification

[polar fractal]

how do i get the advent of code badge/role on this discord like some of you have?

[hazy ermine]

[Day 8]
will it reach the end node only after full repetitions of the direction string or it can reach in, say 2 and a half repetitions?

[cyan sleet]

what's your idea of slow?

my solution takes like 2 seconds total

[meager ice]

it can reach in half

[hazy ermine]

damn

[hazy ermine]

turns out it reaches only in full repetitions, and also another detail

[rocky spire]

Mine also only had full repeats, so probably not luck but intended

[meager ice]

No, mine reached about one and a half

Maybe you just got lucky?

[rocky spire]

Hmmm... My solution 100% would've failed if an endpoint appeared in the middle of the instruction

[meager ice]

I can send you my input of you wanna test

[rocky spire]

sure?

[hazy ermine]

||others also have told that it only takes full reps, and this is a very important detail (excluding another one)||

[rocky spire]

Today's inputs have so many weird assumptions you can make

[meager ice]

i'll check once i get home. I had such a stupid bug and was debugging it for an hour so maybe my brain mixed my memories lol

oh yeah nvm. I checked my answer and remember my input instruction string had 263 characters, so dividng answer with 263 gave me whole number. So it seems like it made full repeats

But I coded my solution so it works even if it's not a full repeat

[hazy ermine]

wait you solved part 2 as well?

[meager ice]

not yet, had to go to university

But I read instructions and it doesn't seem that hard

Boy will I probably mess up lmao

[rocky spire]

If you're thinking about simulating then, uh, I'm sorry to inform you that (at least in my input) the answer was 14 digits long

[meager ice]

oh 😑

then i gotta figure out something else

[cyan sleet]

so yeah, get comfy with associative arrays (and arrays in general)

it's my first time using associative arrays actually

[hidden musk]

I'm very thankful for eval

[hazy ermine]

wait so i cant use 32bit int?

[hidden musk]

the input was luckily very similar to the syntax for inserting things

[rocky spire]

well... yeah, ints only go to like 2 * 10**9
You can just use a 64bit int though, which goes to like 9 * 10**18... unless the language you're working with doesn't have that?

[hazy ermine]

ah thanks, Go does have 64bit ints

[hidden musk]

I have done unspeakable things to achieve 2d "arrays"

[sleek remnant]

In what range is the answer of part 1? Cause I get like more than ||10k|| and it’s still running and it really feels like I just created an infinite loop

[rocky spire]

Mine was ||22k|| so don't worry

[sleek remnant]

Guess I just gotta wait till I get home, my school Chromebook and online ide is way to slow for that haha

[halcyon pulsar]

Can someone help explain how to do part 2 in a realistic time?

[long jacinth]

||16k|| here

[lime fulcrum]

What language are you using?

[lime fulcrum]

||cycles||

[rocky spire]

Hint: ||The solutions of the individual starts (like what you did in part 1) relates to the whole solution that considers all starts||

[long jacinth]

wdym, P2 is same as part 1, it just takes longer 👀

[ripe meteor]

Pt2 runs instantly for me

139ms for both parts

[halcyon pulsar]

How

[sleek remnant]

Python

[lime fulcrum]

Using a dict?

[rocky spire]

155ms both parts for me

[sleek remnant]

I got an actual pc now but it’s killing the the process after ||150k||, I feel like now something is going wrong

[ripe meteor]

my pt1 is <20k

pt2 is 14 chars long

[lime fulcrum]

150k is probably too high. I had nothing higher than 25k

[halcyon pulsar]

Is it ||the lowest common multiple of each of the cycles it takes to complete each one?||

[ripe meteor]

||👀||

[halcyon pulsar]

Damn that works

Much easier than I thought

[rocky spire]

If you inspect the inputs then you can conclude that it works, which is probably the intended solution
If you only look at what the problem says then not so much, and you would've needed a much more complex solution

[mellow mesa]

What node are you initiating with?

[sleek remnant]

Wdym?

[mellow mesa]

When starting off the simulation what 'AAA' - 'ZZZ' are you starting with

[sleek remnant]

I used the first to last line, in my input it’s ‘VTM’ - ‘JDH’

[mellow mesa]

Thats the reason, you wanna start with 'AAA'

[sleek remnant]

Oh I go from AAA to ZZZ in the input as well?

Oh yeah that’s it

I feel stupid now 😂

[mellow mesa]

If you look at the sample input the first row is 'AAA', if you look at your input it's not. Which means you wanna start at the row where 'AAA' is

[sleek remnant]

Ty!

[mellow mesa]

I would predict that is the reason your code keeps running without getting to 'ZZZ', I did the same mistake

[sleek remnant]

Yes it was, I got the right answer now

Thanks for your help

[long jacinth]

i wonder how many ||loops || there are in the input

ive had ||6 starts|| so theres at least that many

[jovial imp]

what is the question

[magic oxide]

after doing p2, i agree even more

||i'm fine with some assumptions since otherwise the problem is not really tractable||

||but there were far too many assumptions today for my liking||

[hazy ermine]

yea ikr

[ripe meteor]

No idea what you all mean, just follow the directions 💀

[rapid wolf]

Did you actually ||solve the full problem in the general case? LCM/product of each path is incorrect as per the problem statement||

[mellow mesa]

Can you explain this a bit further, saw this somewhere else earlier, ||the part where LCM is incorrect per the problem statement||

[rapid wolf]

Consider a ghost whose path goes ||a->b->z->c->z->c->z->z->z->z||, then your ||lcm solution|| will be wrong. ||no such ghost just happens to exist||

[hidden musk]

||it's not correct to assume there's a cycle||

[mellow mesa]

Ah alright

[rapid wolf]

well, there has to ||be a cycle, since there is a finite number of states - but the cycle does not have to be simple||

[hidden musk]

right, yeah

[mellow mesa]

But the same way, there also isn't a ghost ||that takes such a route so couldn't you reverse the argument to this too||

Sorry for excessive spoilering, rather do too much than too little

[hidden musk]

right. but it just so happens to be that way. it's not guaranteed

[mellow mesa]

Interesting day to say the least

[hidden musk]

i don't really like this problem for that reason, actually. i think that means i agree with stickie

[magic oxide]

damn pub agrees with me?

[rapid wolf]

I think the problem statement should be the complete description of the problem, and the input should use the ||full extent of the problem statement to test that you implemented it correctly, it should not expand on the problem statement, making it easier.||

[hidden musk]

as a side note, bash functions are 🥴

[minor cave]

"what did you do for AoC this year?"
"create unspeakable horrors in bash"

[hidden musk]

it's actually not too bad

the input was nicely almost valid syntax

so i just sedded it then evaled it

[hidden musk]

a technically incorrect assumption being the only feasible way to approach the problem is meh

[magic oxide]

||my main gripe with it is that one of the assumptions doesn't even make it that less feasible
making the offset a multiple of step length but different from the cycle length would still be feasible with CRT, and i wouldn't be as annoyed||

[hidden musk]

the bus problem was good

a few years ago

[low condor]

that was nice but that was a later puzzle

for a day 8 puzzle I think the difficult is ok

[magic oxide]

my opinion is that you shouldn't take a hard problem and adjust the difficulty to fit an easier day by putting in assumptions about the input

i would've enjoyed doing this day without the assumptions on a later date

[scarlet wasp]

You don't need to "assume" anything though. Analyze, yes, but that's not a surprising part of problem solving imo

[hidden musk]

sure you're exploiting a pattern you observed, but it's technically incorrect

[scarlet wasp]

Idk, I don't agree that a solution necessarily needs to be able to handle all kinds of input. Story-wise, there's just one input - not multiple - so you'd analyze it and solve the situation accordingly 🙂 But yeah, it's more satisfying to write general solutions

[magic oxide]

by assume I mean anything that isn't there on the problem statement and has to be gleaned from the input
I am somewhat ok with two of the assumptions today, but the third is just

[woven sable]

||i feeel like the fact that u had to derive that it cycles is kinda dumb||

nowhere did it say ||it cycles||

[rapid wolf]

careful, this is not the spoilers channel

[woven sable]

oh oops sorry

[scarlet wasp]

Doesn't this clarify that?
"Of course, you might not find ZZZ right away. If you run out of left/right instructions, repeat the whole sequence of instructions as necessary: RL really means RLRLRLRLRLRLRLRL... and so on."

[magic oxide]

||you wouldn't, and shouldn't, expect the starting offset to be the same as the cycle length, and it doesn't even make the puzzle that much harder, and it awarded people who incorrectly thought "oh i just need to lcm the starting offset/cycle length" and didn't think about CRT, and checking if starting offset == cycle length is such a weird, unusual thing to do that I don't think it should've been an assumption today||

[woven sable]

i feel like that's more of behavior w the instructions themselves

[magic oxide]

||you had to derive that Z cycles, and only one Z per cycle||

[rapid wolf]

I find scanning through the input and trying to find some janky way to "scam" a solution out of the problem remarkably unfun.

[magic oxide]

||cycles are basically guaranteed by assuming a large enough answer and pigeonhole principle||

[scarlet wasp]

Well you can analyze it with code though

[rapid wolf]

of course I can, but that doesn't make it interesting. I should not have to double check whether each possible case actually happens in the input.

[scarlet wasp]

shrugs Different perspectives I guess 🙂

[woven sable]

i feel like the most fun problems in past years have been path finding problems like flood fill and all or like

snailfish was really fun

and day 24 2021

the arithmetic one

[simple raven]

@mossy basin your visualisations are amazing, good job! 🙂

[magic oxide]

from the perspective of solving the problem blind, you have no idea which assumptions you should be checking imo

[mossy basin]

thanks

[magic oxide]

for stuff like day 24 of 2021, the assumption is kinda ok, because there isn't much there to do other than analyze the input

for today, I would not have batted an eye if we had to solve the general case

especially given the difficulty spikes this year

[rapid wolf]

2021d24 also literally tells you to decompile it.

[simple raven]

I feel like ||6 and 7 went down in difficulty quite a bit||

[magic oxide]

yeah

I did have another gripe with 6 because of the lack of edge case testing for some inputs

but that's a separate matter

[simple raven]

do you mean day7 with the cards? or what edge cases are you thinking of?

[untold briar]

Or more like day 3 and 5 were way too hard haha

You can see from statistics how many ppl dropped the challenge in those days

[low condor]

day 1 3 and 5 were the hardest so far

uncharacteristically difficult for those days

day 1 is hardest day 1 I've seen for years

[simple raven]

to me day4 was hard as well because I was determined to ||write a recursive solution but I have never practiced recursion||

[untold briar]

I think recursive should be avoided in general

Loops are safer, imho

[simple raven]

tbh I couldn't think of how to write the loop solution, ||recursion|| just made sense to me for that problem

[untold briar]

I normally change recursion into while loops with a couple of flags

[compact wasp]

loops are antiquated and should be deprecated, tail recursion is the way 🥱

source: i use functional languages

[untold briar]

Idk, I just find recursion very hard to debug

The order of what's happening in recursive loops is hard to track for me

[lime fulcrum]

I like while loops

[low condor]

It takes me a while to come up with a recursive solution

I'm kinda surprised people went for it for day 4

I have an iterative solution that looks fairly reasonable

[ocean grove]

Do you get any prizes for doing well in Advent of Code>

[minor cave]

If you get all 50 stars, you get a cute Advent of Code Completionist Role for the month of January

[ocean grove]

alright!

[covert arrow]

I like to assume that the function is already perfect so I only have to think about getting from one case to the case immediately before it

[wet yoke]

Just gotta think of the function as running backwards 🙂

[hidden musk]

it's nice to think inductively. how can i create the n=1 case from the n=2 case, etc

[jovial jay]

i never had to use recursion. though I use G'MIC as main, and in the math parser, recursion isn't possible most of the time, so I always have to figure out a iterative solution. by then, when I think of solving a problem, I always end up solving it anyway with no recursion.

[rapid wolf]

I use recursion all the time, I find it more convenient then rewriting an already correct function into a loop.

my ||range splitting solution|| used recursion, for example.

[jovial jay]

i'm thinking of solving day 5 part 2 by using a while loop to modify dynamic array or list of values, and if the last value of first index reaches the minimal, I then find the minimal values from what point. kind of like modifying all the dynamic array until the last value first index reaches 0.

[jovial jay]

hmm, i need to figure out how to split a range given ranges here. Hmmm, I"ll think about that

[dreamy arch]

One message removed from a suspended account.

[low condor]

is it to do with Advent of Code?

I never use recursion unless the solution screams for it, personally

Python feels like an iterative-first lang

Sometimes recursion is obviously called for though (e.g. parsing recursive structures) and then I will use it

[dreamy arch]

One message removed from a suspended account.

[low condor]

the answer to my question is clearly no then

this is the wrong channel for you, please look at #❓｜how-to-get-help in order to open a help channel for your question

[jovial jay]

there's languages better for recursion? why would some one use them in that case?

[low condor]

there are people better than me (and almost certainly in this channel) to answer this, but I'll have a go

Functions that are recursive-first (typically functional languages) have features geared to recursion. The key one that comes to mind is "tail call optimisation"

[minor cave]

Python doesn't have tail recursion, which feels like an easy answer to why it doesn't necessarily handle recursion well. Without using cache I feel like it's easy to run into a RecursionDepth error for any moderately sized problem/input

[low condor]

^

[minor cave]

Then if you want to use cache, you need to make sure you're using inputs that can be hashed. That could result in having to finetune your data structures a bit

[low condor]

I'm probably going to butcher this explanation, but "tail recursion" is where, if the last line of a recursive function is the recursive call, you don't add another stack frame

I think

[minor cave]

side eyes dataclasses

[low condor]

I just slap frozen on it and it's never failed for me so far (though I'm sure one day it will)

[low condor]

Also it's worth mentioning that the creator of Python typically is not too much into functional programming and recursive style

map, filter and reduce were almost not going to be in Python 3, for example

(in the end reduce was downgraded out of the std lib built ins to functools)

[rapid wolf]

A valuable aspect that makes recursion significantly easier to think about is functional purity - a functions result only depends on its arguments, and it only returns a result, doesn't modify any global variables etc. At which point, recursion get less "leaky" as an abstraction, and you can mostly just call the same function as you would any other helper function. functional languages tend to make pure functions the default, or the only kind of function.

[jovial jay]

if every 5 days is hard. i bet day 10 is going to be about writing a OS

[sleek cave]

today looks super cool. looked at part 1 and thought it was trivial, given the input is ||an adjacency list|| and all xD
thought maybe part 2 would be ||disregard instructions, find shortest path|| but i was like, "||isn't that a lil too textbook :P||"
and wow does it hit me with a twist! i'm currently thinking ||cycle detection and least common multiple||, but i complete forgot ||how to do the former ... need to brush up on graph theory|| xD

[low condor]

can't wait to see your solution!

[untold briar]

Tip: ||you can reuse first part||

[sleek cave]

oh man this is harder than i thought
i think i can go for a clunky solution under the ||very jank, yet (at least for me, i have tested) correct assumption that for every starting node, using the instructions to traverse the network will certainly encounter a cycle, during which a single, unique ending node is encountered||

[low condor]

Well if true then it won’t be as hard as you might have thought :)

[hidden musk]

@minor cave what does this mean

[sleek cave]

but i don't like it 😩
this assumption is so specific

it feels like cheating, honestly :P

what if, in a more robust version of the problem with multiple test cases, ||all ending nodes are reached before any cycle commences? some starting nodes don't fall into a cycle at all? two or more ending nodes are encountered in a cycle?||

of course, ||all the above are not true for my puzzle input|| :P

but still

[low condor]

It is often the case that the input is crafted specifically with assumptions in mind

For better or for worse

[sleek cave]

that's a great point

so at the end of the day, we go back to rule #1 of programming

if it works, it works

[low condor]

If the day has assumptions then I’ll often write assertions to prove the assumptions hold

[sleek cave]

oh great idea

[low condor]

And document that the solution only works for those assumptions

And write unit tests to prove that the exceptions are raised when the assumption breaks

[sleek cave]

kk, i think i'll do just that

time to take a break and watch movie today :D

will resume with my janky solution tomorrow lol

[low condor]

No go code it now!!!

Jkjk

[jovial jay]

i use visual debugging and parameter checking and a bit of hypothesis checking to see how is my code is going, and then if i see it working well, i continue.

[low condor]

That’s perfectly fine too :) I’m just going OTT for my solutions

Honestly for P2 I did that first before I wrote my test

It’s one of the areas where TDD didn’t work so well for me, because you don’t want to write tests just to find that the assumption is trash

[wet yoke]

I did just start incorporating tests early in the day myself. Not so bad when you've got the samples and their answers.

[brazen bridge]

does anyone know some good graph libraries for the usual type of AoC graph problems that will appear in the near future? I'm looking for something that handles DFS BFS and potentially shortest path type stuff?

[rocky spire]

If you mean visualization, then networkx is a popular choice
If you mean one that contains actual implementations of these algos then idk

[solid rover]

I opened up a thread in #1035199133436354600 if anyone already did day 3 and would be willing to guide me in where my current logic has gone wrong

[rancid creek]

You can take a look at my solution: #1180738420272799814 message
If you're around I can look at your code after I finish tonigh's puzzle

[solid rover]

I got help, thank you though! I was trying to avoid just looking at solutions and seeing where I might have missed something in my logic, turns out when constructing numbers my program only checked surrounding indices once it hit a non-numeric character, which doesn't work when the number reaches all the way to the end of the line and resets it.

[rancid creek]

Ah cool. congrats.

After you're done though defintly look at other solutions for new ideas.

[earnest forge]

GLHF everyone!!

[minor cave]

Language Roulette: Day 9
The language is ... Scala

[peak dock]

the prophecy

[shut trellis]

<@&518565788744024082> Good morning! Day 9 is ready to be attempted. View it online now at https://adventofcode.com/2023/day/9. Good luck!

[magic oxide]

oh wtf it's time?

[flat sequoia]

3/1

lmao

nice

[magic oxide]

ooh finite differences

[pine tiger]

today's was underwhelming

i mean like that was it?

[covert arrow]

my heat is in 5 minutes 😔

no leaderboard again ig

[lusty rover]

_I still don't even understand the problem and you're done already _

[pine tiger]

holy crap still in the middle of the second michael buble song

[honest nacelle]

I can't even understand what the question is asking...

[hexed axle]

That leaderboard was nice and fast

[earnest forge]

Spoiler channel?

268/133 ._.

Made a silly mistake that slowed me way down for p1

[minor cave]

god damn, can't even look away for 10 minutes

[pine tiger]

yeah like

you refresh once and the lb is full 💀

[mossy basin]

rank 262 is about the best i can get

i'm too slow

[hexed axle]

Right? I was modeling the completion time and went, "That means it should be done by 5 minutes past the hour. What time is it? Oh, it's 6 past the hour let's check the lead--it's full."

[minor cave]

#1182911467717144616

[pine tiger]

doesn't this mean the leaderboard didn't meet your expectations?

[hexed axle]

No, far from it. It means my copy-paste foo is too slow to keep up

[pine tiger]

i mean sorry if i'm not reading that message right but it appears you thought it would fill in 5 mins but it filled in 6

[modern terrace]

that was relatively easy

[hexed axle]

No by the time I finished running the model it was already well past when the leaderboard filled

[pine tiger]

oh lmao

[minor cave]

I was genuinely settling in for a 20-30 minute leaderboard fill time

[hexed axle]

right? Awesome that it filled so fast but I've been looking forward to predicting the leaderboard filling up, and it's just been too quick most of the time

[minor cave]

The times to fill each day have been so variable compared to prior years. Difficulty is all over the place

[hexed axle]

It's a weird balance of parsing and math (granted I've only got one star this year so I've been mostly watching comments on it) but it's nifty to see how this year feels different

[zinc kindle]

Yes. Top 1000 for part2.
When doing part1, assumed I'd need to get 100000'th value for each row or something, which would require something more efficient. As such, ||I decided not to bother with the supplied algorithm, switched to python, and used numpy.polynomial to derive the formula, which would give me an easy time in part2||
result: ||I was wrong about what we'd have to do. Still, managed part2 in 30 seconds (would have been less but reran it on sample input after gave suspiciously small answer before submitting||

[hexed axle]

The difference between getting the first star and the second star each day is really wild

[flat sequoia]

Today's is like, the same though

And it's interesting that the person who got global 1st on Part 1 didn't even place for Part 2

[flat sequoia]

Yeah that should have been part 2 ☹️

[peak dock]

thats sort of what i was expecting too

[honest nacelle]

... huh

this means tomorrow is gonna be awful, huh?

[untold briar]

Or it could be easy again, who knows

[zinc kindle]

Weekends (utc-5 time) tend to be harder than weekdays. For people many time zones ahead like myself, that tends to be Sunday and monday.

[stable shale]

Spent 40 minutes debugging the problem only to figure out I didn't handle ||negatives|| in input parsing

[covert arrow]

I did abs(b-a) instead of b-a while debugging and forgot to remove it

[jovial jay]

yeah. i'm surprised too, but then again Day 5 Part 2 is a real PITA

[zenith elbow]

today was probably the most uninteresting part 2

[jade atlas]

I almost feel disappointed. I thought today would be a hard day, I blocked 2-3 hours for today's problem, and this is the problem that came up

[royal gull]

I don't know how they even comprehended the problem statement that fast.

[untold briar]

Dont worry, you wont be dissapointed anymore one of these days

😏

[zenith elbow]

how about use the rest of the free time to solve past year AoC - unless you've done all

[jade atlas]

Yeah, doing that right now 😅

[jade atlas]

I know.. but I blocked some time today for aoc you know..

[hexed axle]

Same. But I'm too slow of a reader in general, I know I'll never read fast enough to keep up with the global leader board so I've made a made of modeling the leaderboard instead

[untold briar]

Yeah you can use the rest of your time to solve past puzzles or optimize your current ones

[untold briar]

I didn't even finish reading and comprehensing the puzzle and the leaderboard was already filled

[royal gull]

Maybe we need a leaderboard for mortals.

[rancid creek]

There are various leaderboards. There's 2 for reddit, 1 for discord I'm on, And another one I think is for streamers or something I'm not sure why I'm on it.

I only recognize my name and jp

[rancid creek]

Have you finished past years already, I can give you a past year probelm that was harder

beacons or snail math were both a lot of fun

[simple knot]

does anyone here know any java?

i cant do day 7

well, im just getting the wrong answer

but i really think it should be working

[untold briar]

Why are you asking about java in a python server?

[simple knot]

i mean, a lot of people can do two languages, i thought it might be worth asking lol

[real quartz]

Well, dang. Never thought I'd have this much difficultly on the day 1 of the advent puzzle, lol and I'm trying something new I guess to learn, outside of doing my main project that I'm kinda stuck on.

Dang, I really should have gone for trying 15 minute projects and slowly build up, rather than just go for major ones but advent of code seems fun

[covert arrow]

I only know the basics, but the people at https://discord.gg/WEnkdVMQ will know much more

[simple knot]

oh bet

[simple knot]

what are you stuck on? whats your approach

[covert arrow]

usually the puzzles get harder as the advent goes on but this year the difficulty is a bit all over the place

[real quartz]

Right now, I'm looking at puzzle input, and I've converted entire thing into a rows of string to be put inside a variable, and I'm working out the best way to pull the numbers from the list of strings and then using sum to get the final number that I can input to get into 2nd day.

If my approach is wrong, then sigh, I'll figure out a different way. I'm googling all over the place without using ChatGPT

[covert arrow]

Feel free to post your code in #aoc-solution-hints if you want help

[real quartz]

Will do after much headscratching and if I can't find a solution.

[jade atlas]

which year and day is it? let me know

[covert arrow]

Googling "advent of code [thing]" will get you the right puzzle most of the time, beacons and snailfish numbers are 2021 day 19 and 2021 day 18 respectively

[jade atlas]

thanks

[rancid creek]

https://adventofcode.com/2021/day/18 snails!
https://adventofcode.com/2021/day/19 beacons!

[low condor]

Just renewed my AoC++ for this year

Feels good

[magic oxide]

nice

[low condor]

At some point I need to transfer my account to another one (next year perhaps)

my username looks a little silly next to everyone else on my work leaderboard

I'm almost certain no one at work authenticated through Reddit lol

I'll probably switch to Github

[hazy ermine]

you have a work leaderboard?

[low condor]

yup

majority of people have like a proper name, and then there's me with my Reddit account lol

[hazy ermine]

reddit/email usernames are easily in the top regrets of anyone

[magic oxide]

I'm still 🥴 about using my proper name for stuff online

but it'd probably be good for finding work

[low condor]

just show everyone how high up you are on Advent of Code leaderboards

instant hire

[hazy ermine]

wait really?

[magic oxide]

if only 450 stars on aoc was enough to get a job

[hazy ermine]

didnt know AoC has employability value

[low condor]

nah lol - but it's something you can use to show off programming skills

[hazy ermine]

right

[low condor]

and it helps get practice for actual interviews too

you might be asked these kinds of questions

[hazy ermine]

true but everyones grinding away on platforms like leetcode

its kinda sad that leetcode type questions have become the standard for interviews, but understandably theres probably no other easy / short / objective method to evalutate candidates

[covert arrow]

I wish I was more interested in leetcode/dsa in general, but tbh I just hate it

[hazy ermine]

dsa is actually interesting, its the grind part that i dont like

[covert arrow]

Yeah I agree, I'm fine with dsa but I'm not really into competitive programming

[hazy ermine]

same

[magic oxide]

I'm somewhat into competitive programming, but that might just be my interest for maths leaking into it

I suck at cp anyways so

[hazy ermine]

although i see leetcode like some sort of puzzles to solve just like AoC

[low condor]

I'm not really into competitive programming but I do enjoy AoC

[simple raven]

But aoc is more fun 🙂

[hazy ermine]

agreed

[low condor]

difficulty is nicely scaled and I enjoy the community around it during Xmas

[covert arrow]

yeah I like aoc/codewars style puzzles more

[magic oxide]

I like well made codewars ones

[zenith elbow]

so relatable

[low condor]

I get to write my own answer, and then I enjoy code review and debugging other solutions too

[hazy ermine]

havent heard of codewars

is it still a thing?

[covert arrow]

Yeah, it's like leetcode but more casual

[hazy ermine]

wow

[magic oxide]

it's a shame I'm always on holiday during december

[low condor]

I avoid jetting abroad during Dec

[magic oxide]

my family is overseas so I kinda have to
tho this trip is different

[zenith elbow]

we even have a .challenge command on sir-lancebot for it

[hazy ermine]

is someone tyring to send code lol

[covert arrow]

Seems like it

[magic oxide]

graduated aus hs, and it's a tradition over here to go on a trip with ur school friends after that

[hazy ermine]

sure, i'll take a look into it

[covert arrow]

Is this related to Advent of Code?

[hollow wharf]

soru

sory

new here

[low condor]

#❓｜how-to-get-help to get help

[hollow wharf]

k

[covert arrow]

You could ask in #1035199133436354600 , but don't expect people to write code for you

[low condor]

!rule 8

[marsh current]

Rules

8. Do not help with ongoing exams. When helping with homework, help people learn how to do the assignment without doing it for them.

[simple raven]

I've just learned about the rule not to share your puzzle input on github, I had to do a quick purge 😰

[low condor]

ideally you should remove it from git history too

[simple raven]

how do I do that

[low condor]

I forget lol

[long jacinth]

what order are the answers for D7P2?

or well, how many digits

[low condor]

apparently according to SO you should use the third-party tool git-filter-repo

then when you install it you do you git filter-repo --invert-paths --path <path to the file or directory>

...or if you don't care about history just transfer to a new repo and start again

[modern terrace]

let me look at mine wait

251mil d7p2

it's pretty close to p1

[simple raven]

oh day7 sorry

[modern terrace]

perhaps you mean d8 p2 which is a more reasonable day to ask such a question

[long jacinth]

what, my P1 was in tens of thousands

no i mean D7

wait

todays the eight

yes, i do mean D8, im sorry

[modern terrace]

(yesterday's the eight)

[long jacinth]

ninth*

yesterdays, please, lol

[simple raven]

okay so I was right 😄 part 1 tens of thousands, part 2 tens of trillions

[modern terrace]

aanyway yes d8 is in the tens of trillions for part 2 and tens of thousands for part 1

[long jacinth]

okay, thanks

[hazy ermine]

today was so easy people be discussing about previous days

[modern terrace]

d8 and d5 are probably the hardest of the first 10 days

d5 mostly

[rapid wolf]

I still don't know a satisfying solution to d5p2.

[modern terrace]

where's that fancy dot graph

ah

[hazy ermine]

are there any people from global leaderboard here?

they might be onto something... right?

[modern terrace]

range splitting

[magic oxide]

calculating intervals is pretty satisfying to me

[hazy ermine]

is that the only way?

[modern terrace]

bruteforcing is possible

but that isn't "satisfying" per se, is it?

[rapid wolf]

I meant day 3, oops

[hazy ermine]

apart from bruteforce ofc

[magic oxide]

I think some people also said something about dfs from the back?

[rapid wolf]

Honestly d5p2 felt really easy to me

[magic oxide]

ah

[rapid wolf]

If I knew how to write Multiline functions in APL, I would be done fairly quickly

[magic oxide]

I think the name numbers and combine them -> iterate over gears method was pretty satisfying

[modern terrace]

oh..

yeah

[hazy ermine]

from what i've seen, part 2 solution is always a not-so-major change to your part 1 solution if you had chosen the right method from the beginning

[rapid wolf]

That's what I did in Haskell, but it didn't really feel great. I would want a solution that finds them without duplicates in the first place, not one that just removes them.

[modern terrace]

i honestly don't understand what i did there 😬

[hazy ermine]

i anchored around the gears, which made things easy

[rapid wolf]

It wasn't difficult per se, though my code was pretty horrid. Just felt inelegant

[modern terrace]

for each gear, detect all numbers within a certain span??

[long jacinth]

i feel like im doing AoC wrong

i made a grid object which held references to every cell, and part that cell belonged to

[hazy ermine]

did you anchor around numbers?

[rapid wolf]

No

[hazy ermine]

ah

[rapid wolf]

I kind of want to try doing it with a proper graph kinda thing

[hazy ermine]

I used a 2D array and the entire file for day3 was around 100 lines of Go

[rapid wolf]

oh yeah, I didn't have a lot of code either.

[magic oxide]

I think a neat solution would be systematically 'melting' away numbers and gears as you iterate over gears

tho that wouldn't quite work for p2

[rapid wolf]

my idea was to have almost an undirected grid graph, but merge digit nodes into one big number node

then both p1 and p2 get trivial

[magic oxide]

I want to try my cursed convolution method for p2 sometime in the future

[hazy ermine]

ahh yes, that does sound satsifying already

never thought about convolution

[rapid wolf]

my d5 is almost entirely convolution (though it is like fancy APL convolution, not the basic multiply and sum)

[magic oxide]

there's the neat method with convolution where you preprocess and pack numbers together and then convolve

and then there's my method which operates on the array with minimal preprocessing (turn the symbols into numbers, and have a shift of +1 and set empty spaces to 0)

but has 12 convolutions and a whole bunch of masks afterwards on the convolved arrays

[long jacinth]

oh, the solution to yesterdays was so much easier

than i thought it would be

[long jacinth]

okay, day 9 was super easy, i need to return to day 5 today and possibly finish it

[low condor]

best of luck! let us know if you need any assistance

[long jacinth]

i just need to finish the || range interpolation || algorithm

shouldnt be too bad

[untold briar]

Well that algorithm choked 2 hours out of me

[low condor]

lucky for you it was only 2 hours

took me something like 3-4

[untold briar]

The majority of the time was spent looking for when things went wrong

[low condor]

ditto

[untold briar]

Cause the ranges got splintered more and more after each mapping

[magic oxide]

I had a really fun time doing that day actually

it took me like 1-2 hours as well but I really enjoyed it

[untold briar]

Yeah I really enjoyed the first hour too

But past that and it started to get stale for me

[low condor]

I found it quite fiddle-y which gets annoying after a while

[simple raven]

I figured out the range stuff resonably quickly, but then I ended up adding some of the splits into the wrong list and that sent me on a 6hour-long journey of rewriting everything only to realise I had to change one variable name and I would have been done -.-

[magic oxide]

surely fiddle-y stuff is your forte, right?

[low condor]

apparently not!

[flat sequoia]

Man I still can't believe yesterday was just LCM

that was legitimately so dumb

[magic oxide]

agreed

[flat sequoia]

I was spending forever trying to remember how to do CRT before I saw that everyone was already done

and then was like oh, what

[magic oxide]

luckily I had spoiled myself by looking at code golf chat earlier that day

saw them using lcm

and was kinda confused how they did it

[eager lake]

I didn't know that math had an lcm function, so wrote my own broken version -- luckily it worked 😃

[flat sequoia]

Lol

that seems quite nontrivial

I guess it's not

You can do two numbers at a time I guess

[simple raven]

you write it for 2 numbers and then you map it onto (previous result, next number)

[flat sequoia]

Yeah I thought it was nontrivial for a sec because lcm(a, b, c) != abc/gcd(a, b, c)

[eager lake]

well, i factorized the numbers (poorly) and found the common prime factors

[flat sequoia]

but if you just fold then it's trivial so

ah yeah that works

well that's just gcd

[flat sequoia]

I wonder what the relation actually is

[jade atlas]

what does this mean? I can't get 50th star because I didn't solve previous days? Is it usually like this?

[flat sequoia]

Yeah it's always like that

Last year I lost a point on Day 25 because I choked and someone clicked the button faster than me

[flat sequoia]

oh I guess it's lcm(a, b, c) = abc/gcd(ab, bc, ac)

[jade atlas]

I just realized you are one of the top coders in the leaderboard. Could you please let us know how do you guys solve a problem before us humans can't even finish reading the problem?

[untold briar]

Took me 2 hours to realize

Then a couple more mins to solve it since I recycled part 1

🤪

[low condor]

People going for the leaderboard have done a lot of prep writing a bunch of functions for stuff they expect to see come up

There’ll also be code to programmatically retrieve the input and post the answer

[hidden musk]

I mean, these don't really help unless you're actually good at problem solving

[low condor]

Well yes they have that as well :)

[simple raven]

the guy that solved today's questions as 3rd is copy-pasting the input manually 😄

still, solving these in 2 minutes is incomprehensible to me

[low condor]

Maybe they could have gotten second if they used aoc-lube :P

[modern terrace]

hey i used to do that

[simple raven]

I still do it...

I'm hopping between two laptops while working on the solutions and I want all of my random envs to be able to run the solutions, so I never installed aoc-lube

I could just set my stuff up better but 🤷‍♂️

[low condor]

I don’t use it either tbh

I never go for lb anyway since puzzles open at 5am for me

[simple raven]

western europe gang (I guess some of Africa would be the same timezone?)

[low condor]

I prefer to go for most verbose solution instead

[zenith elbow]

for me it's 1pm, you might say it's convenient but actually not really since I might have existing schedules at the time

[low condor]

Yeah, Morocco is UTC+1 for example

[low condor]

I don’t think that’s convenient for the majority of people lol

But at least you can get started on it quicker after your existing schedules

West Europe is one of the worst cos if it’s a weekday then I have to wait all day to complete it, if I don’t want to get up at 5am

[zenith elbow]

kind of reminds me of the last guy from here

[low condor]

I am 100% the guy at the top lol

[slate berry]

Lol today's part 2 was just ||reverse input||

[simple raven]

I'm above the guy on the top 😦

that awkward moment when the 200k/year in an eastern european country is in a currency that is 1:30 weaker than euro

[leaden tartan]

Ive completed the first 4 days, but got busy with work so havent done the rest. How brutal are the next 5

[simple raven]

I had a good laugh reading these 😄

[simple raven]

don't let 5 prevent you from trying the rest 🙂

[modern terrace]

i am 50% the top guy

the only changes are that i use notepad and print() instead of a full IDE and a debugger

[low condor]

How comes you use print instead of a debugger ?

[modern terrace]

works in notepad

and terminal

[leaden tartan]

Yeah I worked a 12 hour day, opened day 5 and decided my time was better spend asleep

[low condor]

Even in terminal you can use a debugger - such as pdb

But it might be good to use VS Code or PyCharm (or another editor of your choice) and use a debugger for that

[modern terrace]

i don't know how to use pdb

print() usually works well enough

[low condor]

You do you, but debuggers are amazing

Don’t know how I lived without them

It’s basically print without actually having to type print all the time

[modern terrace]

i know

but i had a problem with speed in the IDE i used

in another different language

doing a different thing

[simple raven]

what's your go-to debugger for vscode?

[royal gull]

Here, have a free ‘skip a day’ pass 💳

[low condor]

I believe you just install the Python extension and then you’ll able to use the debugger inside the app

Personally I use PyCharm which has that all set up though

[shut yarrow]

well i have some issue with PyCharm debugger's stuff

[simple raven]

ah I see, I have to set it up somehow...

my friend likes pycharm, I should probably switch but I already got used to vscode

[shut yarrow]

May i know which edition you use for PyCharm

[low condor]

Just community.

[low condor]

Curious - what issues do you have?

[shut yarrow]

the debugger just wont work

[low condor]

Both are excellent choices - use the one you prefer

[low condor]

Hmm - could be a few issues why that is. Hit me up in #editors-ides if you want to go over it more

[simple raven]

well I'm off to work for an hour in the middle of a storm... it's not like I needed a weekend to rest or anything

[low condor]

On AOC or actual work?

[simple raven]

actual work 🥲

[hazy ermine]

waiiiit WHAT

i totally forgot

it didn't cross my mind at all

[hidden musk]

banned

[hazy ermine]

nooooooooooo

[austere basin]

yo guys, rlly quick question - how do i increment a global scope total variable from inside a function?

[hazy ermine]

you declare total as global

[austere basin]

OMG YEAH TY

[hazy ermine]

like so:

total = 0
def some_func():
  global total
  total = 1

[modern terrace]

there are probably other better alternatives but
global total at the top of the function

[simple raven]

Global total

[austere basin]

YES THANKS GUYS

[low condor]

Ah my eyes

[hazy ermine]

is it that bad 💀

[fleet fox]

what--- why--

[low condor]

It’s a gratuitous global so yes !

[fleet fox]

what happens if i share it

💀

[low condor]

Eric, the author of AOC, has requested people not to do it

[fleet fox]

oh damn 💀

time to purge then...

[low condor]

Multiple inputs is one way he prevents people ripping off his content

[fleet fox]

where did u find it

[hazy ermine]

wow that makes sense

[low condor]

Can I copy/redistribute part of Advent of Code? Please don't. Advent of Code is free to use, not free to copy. If you're posting a code repository somewhere, please don't include parts of Advent of Code like the puzzle text or your inputs. If you're making a website, please don't make it look like Advent of Code or name it something similar.

https://adventofcode.com/2023/about

[fleet fox]

thanks