jrke: hey how can we command dead pacs
njaber: you can't?
jrke: so can we see pellets that our dead pellets can see
njaber: you can't
jrke: so what can dead pacs do
eulerscheZahl: they are just there so you know you killed them
njaber: nothing, they send the information to you for decdution
jrke: means no use just to show how many of our pacs died
Neabfi: yeah and to know how many enemy pacs are dead too
eulerscheZahl: and where they died
jrke: K
jrke: means they can't see anything
eulerscheZahl: correct
njaber: indeed
jrke: thanks :)
eulerscheZahl: this info is only relevant when you are competing forthe top
mohammadhei05: shut
Zenoscave: shut what
mohammadhei05: http://chat.codingame.com/pastebin/b836d157-e1b9-4670-9657-94441dc09bb0
mohammadhei05: give me code
Zenoscave: no
Zenoscave: do your own homework
cegprakash: anyone doing depth 30?
cegprakash: :D
mohammadhei05: ok
mohammadhei05: :joy:
njaber: cegprakash I'm curious, when you're talking about depth, do you mean depth in number of turns, or in number of moves?
cegprakash: turns
cegprakash: ofc
Zenoscave: Does speed count as 1 or two turns?
cegprakash: 1
njaber: so top gold are doing 20-30 turns...
njaber: I need to find the answer to my problem quickly so I can implement it
Kellthazar: I want to hit gold to tried on Monte Carlo Tree Search again...
Kellthazar: to try*
Zenoscave: Kellthazar why not try it now
Csipcsirip: MCTS works pretty good if you combine it with a strong rollout heuristics
BenWo: I cannot get up the motivation to code tonight :P
Neabfi: yeah MCTS should guarantee you at least top 200 gold
Kellthazar: Csiripcsirip, but how to use a heuristic to make the random moves? This part is not clear to me.
Neabfi: I never implemented a MCTS on CG, I feel like it is not as simple to make it so that there are enough playout in 50ms
Csipcsirip: you only count moves in your random generator that makes sense. for example you dont go left right left right left right
cegprakash: I win against someone almost everytime on IDE but I lost 0-5 to him on arena
thethiny: randomness ftw
SurajMishra2209: I am still stuck at bronze...type F in chat
JohnCM: ya i'm a strong believer of monte carlo tree search too
JohnCM: just not for codingame, the timeout is killer
JohnCM: if it's 1 pac vs 1 pac still got chance
JohnCM: but the monte carlo complexity is exponential in the number of agents
JohnCM: as you need to multiply num actions per agent * num agent per time step
JohnCM: and it quickly explodes to 4 moves per agent per 10 agents (including opponent) = 4^10 calculations per time step
JohnCM: that is just only assuming up down left right
Csipcsirip: you can have a separate tree for every pacman
JohnCM: if you include speed and transform, there'll be more actions
eulerscheZahl: https://www.codingame.com/playgrounds/36476/smitsimax
JohnCM: well that won't be very accurate if we assume that the other agents are part of the environment
AntiSquid: did you see the player count euler? 5605
eulerscheZahl: wait, what?
AntiSquid: 4605 *
Csipcsirip: you run every pacman in the same env
eulerscheZahl: was 4.5k
eulerscheZahl: ah
AntiSquid: but it's bigger than accountant
eulerscheZahl: still: all-time highest :tada:
eulerscheZahl: might even reach the 5k mark
JohnCM: hmm i guess smitsimax might make a good approx. if you know your enemy posit
Csipcsirip: I don't count enemy at all yet. except I reset pellets, that are 100% eaten by him
Csipcsirip: but when the scores are 0 simmetricity is a good approximation
Kellthazar: Dont you consider enemy moves Csripcsirip?
Csipcsirip: not yet I just use the MCTS to make some collaboration between my pacs
Csipcsirip: but I guess I'll add the enemy with limited search depth / score, to avaoid chasing pellets, that will be eaten in few turns
Kellthazar: Really nice article...
eulerscheZahl: damn you random Csipcsirip
eulerscheZahl: now i have to debug offline :/
Csipcsirip: you can expect me submitting in every 10minits from now :p
Csipcsirip: but that win was no coincidence :D
JohnCM: haha that's good
JohnCM: i'm hoping to implement some advanced kill techniques like euler. hopefully that will give me a boost
eulerscheZahl: i clearly have a bug there. i was aware of it before but that match is a good way to reproduce it
eulerscheZahl: as advanced as in frame 85? :D
https://www.codingame.com/replay/464158575
Csipcsirip: I think I submitted that version again btw
eulerscheZahl: that's the bug i'll address now
JohnCM: ah yup that kind
JohnCM: kill in a dead end
eulerscheZahl: and i miss it
eulerscheZahl: i exactly know why
JohnCM: that's good:)
eulerscheZahl: but needs a bit of cleanup
JohnCM: woohoo my recent submit is doing well
eulerscheZahl: you are close to 1st python already
Csipcsirip: I'm not sure btw to waste 5-10 rounds to kill an opponent pacman. especially at the end of the game
JohnCM: i optimized a bit of my heuristics to prioritize exploration more than exploit
Csipcsirip: *if its worth it
JohnCM: oh it's worth it at the start
JohnCM: either way, a sure kill helps remove one moving piece
JohnCM: quite worth it
JohnCM: romka does it quite often also
JohnCM: so funny how changing one of my heuristic by 0.1 can jump me 50 places
JohnCM: i feel like i'm the one tuning a "neural network" of heuristics
JohnCM: maybe it could be automated
vbplaya: I got on day one and did some coding but ran into a problem where on bigger maps the visible pellet count was not correct so it was failing because it was trying to take 1 too many "input()".
Did anyone else have that problem?
Csipcsirip: I'm almost 100% sure that the problem is not in the inputs given by cg :P
JohnCM: great.. happy with this recent submit. hopefully it'll be better after i implement the trap mechanism
JohnCM: :)
vbplaya: Ok, I understand that but I tried the starter code and it had the same issue. I had to change the line: http://chat.codingame.com/pastebin/de571b75-fd80-4bef-838f-c90f776a4ed6
JohnCM: could it be the problem is not in the visible pellet input
JohnCM: but in the lines before?
JohnCM: like you read a line two times?
JohnCM: why not try printing out the inputs given and see if they make sense, maybe this would help you find out where the bug is
NapTown: never heard anybody else with that issue, and it would be a big issue
vbplaya: I did do the printing out .... I can't imagine nobody else had it but maybe it was something else (it was pretty late). I was thinking that they might have fixed it. Who knows ... but it was odd
JohnCM: maybe you could give us the chunk of code where you did the input parsing (only the input)
vbplaya: If nobody else saw it ... then it must have been an issue with my code (or a python thing)
JohnCM: i could help you debug
NapTown: i don't think he still has the issue
JohnCM: oh that's good news then
JohnCM: don't need to worry about it
vbplaya: Ok ... I'll try it with the original again and see if it pops up again. Thanks
vbplaya: BTW, JohnCM ... thanks for offering to help me debug ... very cool of you.
Csipcsirip: https://www.codingame.com/replay/464167359
Csipcsirip: you shall not pass
JohnCM: vbplaya, no worries. i help when i can
JohnCM: cause i use python also, so it is easier for me
JohnCM: @csipcsirip, actually staying still in a collision seems pretty impressive
JohnCM: it is like.. i don't care if you want to go through, i'm not moving
JohnCM: wow csipcsirip you sure rose very high up
Csipcsirip: thx. my search started to be working...wasted 2 days for not working algos
JohnCM: i don't really do monte carlo tree search, but i have an essence of that in my heuristics
JohnCM: works much faster for python
JohnCM: i think this competition is not settled yet
JohnCM: everytime someone becomes 1st, there's a counter-strategy to their strategy
JohnCM: it's very hard to be all-rounded
Zandy156: It's almost like its a game of paper rock scissors.
JohnCM: or rather a game of hide and seek
eulerscheZahl: sounds like vbplaya is not flushing the error stream
vbplaya: eulerscheZahl ... I'll bite ... what do you mean?
eulerscheZahl: looking at your code snippet: you assume that you don't get the last line of input?
eulerscheZahl: and that would be wrong. you just don't see it in the error output
eulerscheZahl: print(..., flush=True)
vbplaya: Yes, in order to not get a timeout waiting for the last input (which never came)
eulerscheZahl: https://www.codingame.com/forum/t/python3-buffering-causes-confusion/181353/
eulerscheZahl: read this
vbplaya: reading it now
vbplaya: very interesting ... I am a little confused on how to implement a "fix" on my end though.
eulerscheZahl: you flush the output yourself
vbplaya: are you saying when I send the move command?
eulerscheZahl: print("Debug messages...", file=sys.stderr, flush=True)
eulerscheZahl: just add the flush=True part to your printing
vbplaya: ok ... so put in a flush in the loop which reads in the points.
eulerscheZahl: that prints the points
vbplaya: Will do ... thanks euler ... always learning things ...
Bernard_Lowe: is anyone else frustrated that silver introduces showing dead pac-men now?
Bernard_Lowe: Time to fix up all my code to not avoid dead pac-man.
BenWo: not really
BenWo: I mean you already had to account for dead ones, now you just have another way to identify them
Bernard_Lowe: I didn't account for dead pacmen before. I was thinking about writing code for that.
Bernard_Lowe: I wonder what changes will show up in gold league. They said bronze would have all rules, but they kinda lied.
thibpat: hello :wave: I'm going live to continue my Spring Challenge bot https://twitch.tv/thibpat
BenWo: but you always had to not send commands to dead pacs
Bernard_Lowe: correct.
Bernard_Lowe: But, now dead pacs stay in my lists, so now I need to factor them out.
Bernard_Lowe: With no changes in my code, bronze --> I lost some optimization in my code.
CanisLupus: thats not truw, you can ignore dead status and command them. Game is protected: "- Pac 2 is dead! It cannot be commanded anymore!"
BenWo: hm, ok, I'm sure we structured things differently
Bernard_Lowe: for friendly pac-man yes.
Quidome: or drop the pac when getting the input, minimal change
Bernard_Lowe: for enemy dead pac-man, no.
MSmits: if(type == "DEAD") continue
MSmits: problem solved
Quidome: \o/
Bernard_Lowe: yeah, but I have many sections of enemy pac-men checks. so I gotta findem all.
Bernard_Lowe: Are there any changes in gold?
Quidome: no
Bernard_Lowe: Start with 20 pac-man? lol
BenWo: lol ouch
MSmits: if you do it at input, it will be the same for your code
MSmits: your 20 checks will be the same
Bernard_Lowe: good point MSmits, I'll do that, thanks.
MSmits: the enemy pacs wont show up in any lists
MSmits: kk
perseverent: hello
MSmits: btw, i also found it slightly annoying, but it was mostly my own fault. I had already written 1500 lines and I didn't remember what I did about dead pacs
perseverent: still in silver after implementing pathfinding
perseverent: hardest silver ever
perseverent: what should I do to get to gold?
eulerscheZahl: kill all the opponents
MSmits: perseverent, my theory is that it's hard because silver was best 300 out of 4000 when it opened
eulerscheZahl: and harvest on your way there
MSmits: when often it is best 300 out of 1500
perseverent: I see
MSmits: also more people may have spent more time coding because of stay at home
Zandy156: It definitely changes things
perseverent: I mostly ignore enemy pac if I can
eulerscheZahl: many new players who are probably weaker than veterans on average make it a little less drastic as it sounds. but still harder going by numbers
MSmits: there was also little time between ooc and this
mlomb: I still don't attack other pacs
MSmits: that means people may have gone in more prepared and again, spent more time coding
mlomb: I think it's time
perseverent: what are you doing in gold if in silver is this level
perseverent: ?
MSmits: my code hasnt changed much since silver
MSmits: still 22 :P
perseverent: wow
perseverent: great for you
MSmits: yeah, but i want to get legend, so to be safe, i do need some real improvement today
MSmits: though likely i will be close enough to gold boss that i will get in randomly over the weekend
perseverent: real improvement? levitation?
BenWo: I'll be pumped if I somehow make Legend
BenWo: feel like I can do gold if I get some motivation here
MSmits: i still need to hardcode superpellets properly
MSmits: get some dead end kills when i detect them
MSmits: avoid some dead end losses
MSmits: that should be doable today i hope
perseverent: do you make some kind of simulations?
MSmits: yes, monte carlo
MSmits: do random moves, pick the best combination
perseverent: is this level necessary to beat the silver boss
perseverent: does not seem so hard
MSmits: no idea
MSmits: gonna get to work. Good luck perseverent
perseverent: good luck tou you too MSmits
eulerscheZahl: i'm pretty sure that the silver boss itself doesn't do anything fancy
eulerscheZahl: CG likes simple bosses with short code and low CPU footprint
Rickardo: hi
NapTown: hello
Rickardo: does anyone know shadow of knoght
NapTown: first one?
Rickardo: yea
NapTown: you need help with it?
NapTown: https://www.codingame.com/forum/t/shadows-of-the-knight-episode-1-puzzle-discussion/264
Rickardo: yea
AntiSquid: someone does know i am sure
Rickardo: yea maybe
AntiSquid: ah Rickardo bubble sort
AntiSquid: + get an avatar
Rickardo: nah not important
eulerscheZahl: #1 :o
that's a resubmit of my #16 bot
eulerscheZahl: and it's gone
kovi: so random
ScryptAn: Could someone give advice about "Ghost in the Cell" in bronze?
eulerscheZahl: ScryptAn did you check the forum? lots of players describing their strategies
AntiSquid: random, but some are at specific ranks more often
Rickardo: pls y can't someone learn html
351062: The toad going strong
AntiSquid: what do you mean?
AntiSquid: i know html @_@
eulerscheZahl: still the recalc will be fun again
NapTown: someone can
Rickardo: like when i want to solve a game html isn't part of the options
AntiSquid: troll ?
Rickardo: wat u mean
AntiSquid: use javascript / typescript @_@
Rickardo: type script
AntiSquid: do you want to learn html ?
AntiSquid: i can give you some useful links
Rickardo: but i know html well but y can't i use it
eulerscheZahl: for bot programming?
eulerscheZahl: you must be trolling
AntiSquid: then you would know what html is Rickardo
AntiSquid: you clearly don't
Rickardo: i know wa5t html is
eulerscheZahl: hyper text MARKUP language
Ramzel: In the night my rank raised by 100 xd
eulerscheZahl: not a programming language
AntiSquid: also the html joke is shit and old
Rickardo: oh no wonder
TinfoilSaint: to be fair you can solve fizz buzz with CSS
eulerscheZahl: html+css is turing complete
Rickardo: well can u show me some useful java and javascript libks
eulerscheZahl: because java and javascript are almost the same?
AntiSquid: i can show you the way out
TinfoilSaint: If you haven't checked out freecodecamp.org Rickardo then you should
eulerscheZahl: calm down AntiSquid
TinfoilSaint: some of the best free beginners coding tuition out there
Rickardo: yea but do u hav any links
TinfoilSaint: ...
AntiSquid: yes
eulerscheZahl: google.com
TinfoilSaint: I just gave you the link
Rickardo: funny
TinfoilSaint: :joy:
AntiSquid: was about to suggest freecodecamp too
sgalasso: Hi everyone, does anyone have an idea of how we can get the current turn of the game ?
Washier: haha.
AntiSquid: sololearn if you need something easier Rickardo
Rickardo: ok i will try out those web and apps
AntiSquid: keep track of turns sgalasso
AntiSquid: ok bye
Rickardo: thnks everyone
AntiSquid: np
AntiSquid: you just count the turns sgalasso
AntiSquid: first turn of inputs = turn 1, increment number afterwards
AntiSquid: no point pming me sgalasso
AntiSquid: also that part is what everyone struggles with so idk
Rickardo: gg
Rickardo: g
AntiSquid: are you a prince btw?
Adnmaster: do i need to know pathfinding or something like that to get into silver?
Doju: you don't have to but it'll surely help
Adnmaster: so just with coding knowledge without any other tricks i can get into silver?
Doju: uhh
Adnmaster: ?
Doju: if you give your pacs sensible commands and have considered other things as well as pellets
Doju: then sure
Adnmaster: ok
Doju: but it'll be hard to make them collaborate for example
Doju: if you don't have pathfinding
BenWo: heya Doju
Doju: heya BenWo!
BenWo: it's our day for Gold league, right?
AntiSquid: Adnmaster probably more effort to write the if statements for that than doing pathfinding
Adnmaster: ok
Doju: BenWo yup
Doju: AntiSquid i agree
Peudoki: how many people do you think will make it to legend?
Denebil: I haven't used pathfinding before silver
BenWo: Legend, no idea... 200?
AntiSquid: 15 max
AntiSquid: imo
AntiSquid: and from monday it's a multiplayer anyway so i don't see why they would go for bigger legend
Doju: Do any of you have a recursive "conflict solver" thing?
Doju: where pacs want the same thing or are gonna crash
Doju: I think that'd be an easy and best way to go about it but it'll sure be slow
AntiSquid: not recursive
AntiSquid: but yes
Doju: Okay
Doju: How can it be not recursive?
NapTown: i just check if i'm in stuck in the same spot and whether a friendly is the reason or not
Doju: you don't do the same check for the next points?
Doju: NapTown i do that too but it wastes so many moves
NapTown: doesn't have to
NapTown: depends how you handle it
Doju: at least 1
Doju: or actually 3 in total
AntiSquid: people dropped so much, some 150 ranks
qjzx1314: so amazing
AntiSquid: hi qjzx1314 here's a fun challenge for you https://www.codingame.com/contests/spring-challenge-2020
WINWINWIN: Top 3 is varying a lot
BenWo: doing a big rewrite, this is going to have so many bugs by the time I can run it again :(
NapTown: that was me 2 days ago
Doju: Oh huh there are a lot of people here from a local company :D
Doju: that's cool
WINWINWIN: Contest is getting really tough, I`ll be surprised if I can end up in top 1000
qjzx1314: yeah
qjzx1314: Bless to you
thethiny: My Pathfinding went from 50ms to 3ms
thethiny: WOOHOO
NapTown: how
thethiny: I've never optimized code as much in my life
eulerscheZahl: try ultimate tic tac toe if you like speed optimizations
thethiny: @NapTown I replaced my "visited" nodes with only 1 node that contains the parents as a reference
thethiny: saves me some recursions
thethiny: by some, that's 47ms
Doju: wow
thethiny: NonOpt Total Time 0.04488039016723633
Opt Total Time 0.0029914379119873047
qjzx1314: so cool
dbf: :thumbsup:
AntiSquid: recursions are slow
thethiny: yes
NapTown: nice
thethiny: let's hope that I can do something now with all of this
Rprades: I am little bit lost on how to find a dead end for an enemy so I can kill him for sure
Rprades: any ideas are welcome!
Rprades: should I just simulate his moves to see if he can escape?
Doju: Rprades i've been thinking about that too
NapTown: you can find the locations that are dead ends
WINWINWIN: doju have you implemented a pathfinder?
Doju: WINWINWIN yes i have BFS
jrke: i am trying BFS
Csipcsirip: you start a bfs from enemy loc you go until a cell that more than 2 neighbours. if you count only 1 of them he is in a dead end
Doju: hmm
Doju: sure
Doju: that makes sens
Doju: sense
NapTown: i have my graph nodes marked as deadend, but that's only for scoring pellets, haven't trapped anybody yet
Adnmaster: is it possible to see the ability cooldown of the opponent?
[CG]Thibaud: print it and you'll see
Adnmaster: ok
Adnmaster: is there any place where i can learn pathfinding?
Doju: A lot of people linked this page https://www.redblobgames.com/
dbf: 'pathfinding algorithm' -> google
Doju: but personally i just read this https://en.wikipedia.org/wiki/Breadth-first_search
Adnmaster: ok thanks
[CG]Thibaud: https://www.codingame.com/playgrounds/39380/finding-shortest-path-in-the-plane-with-obstacles
[CG]Thibaud: (perhaps not the best)
[CG]Thibaud: (best in the sense "adapted to your need")
kovi: no leaderboard?
kovi: or just sloooow
ashelkov: empty)
kovi: huh, chat is also slow
ashelkov: oh no . seen it/ you 1st
ashelkov: congratz!
AntiSquid: he and 5 other people or so :D
AntiSquid: maybe there should be a counter somewhere showing how long everyone held 1st place for ?
Csipcsirip: pb4 made a diagram like that for one of the contest for top10 places
Ramzel: I m almost always winning with silver boss in IDE, but i cant reach him while submiting ;c
alinde2: 50 in silver, so close! :fearful:
BenWo: hey nice
BenWo: so you're one of those people who pushed me back over 200 :-/
alinde2: I'll put you back under 200 by going to gold :stuck_out_tongue:
Ramzel: Domiko congrats gold
BenWo: haha great thanks ;)
jrke: domiko going in gold
LelouchVC2: I don't even wanna check if the bug in the contest is fixed
eulerscheZahl: which bug?
LelouchVC2: Giving me a dead pac as viable
struct: impossible
wlesavo: lol
eulerscheZahl: dead pacmen have the type DEAD
eulerscheZahl: and that's a feature, not a bug
wlesavo: and its not fixed, rather introduced
struct: I found the origin of the bug
struct: The DEAD pac is for silver+ only
LelouchVC2: mentally retarded either way
LelouchVC2: It wasn't doing it one day
LelouchVC2: The next day it was, I outputted the given pacs after I looked through everything else
LelouchVC2: And it definitely gave me a dead pac
struct: it gives you dead pacs
eulerscheZahl: on purpose
eulerscheZahl: the statement got an update too
eulerscheZahl: are you in silver? there's a green section
eulerscheZahl: highlighting the changes
struct: he is in bronze
struct: Thats why no dead on typeId
LelouchVC2: I didn't touch it since last week
LelouchVC2: silver hadn't even open yet
struct: if your pac dies
struct: you dont get him in your input
struct: at least I think
struct: at least you didnt before
LelouchVC2: That's what it should have been
LelouchVC2: But there it was
eulerscheZahl: if type == "DEAD": continue
fixed
DJDiscoWorm_ce7a: hi
Ramzel: 24'th in silver 50'th battel
struct: no euler
LelouchVC2: Contest pissed me off :/
struct: it wont show dead on bronze
DJDiscoWorm_ce7a: is there a way to read the answer for the quiz\
Ramzel: Is still a chance to go up yo gold?
struct: LelouchVC2 I just checked
NapTown: he doesn't have type=DEAD yet
struct: if your pac dies
LelouchVC2: Pacs were kicking ass and then couldn't win cause of it
struct: it doesnt show on input anymore
wlesavo: LelouchVC2 thats really sounds like your own bug btw
LelouchVC2: Not gonna keep doing it while behind
eulerscheZahl: then i will just ignore the chat for the next couple of minutes. you seem to understand the issue better than i do
LelouchVC2: I know wlesavo
struct: I meant no disrespect euler :/
LelouchVC2: But it really isn't though
DJDiscoWorm_ce7a: is there anyone can help me solve the problem
eulerscheZahl: i didn't read it that way struct
NapTown: LelouchVC2 are you checking the mine flag?
eulerscheZahl: just bored of the discussion, your job now ;)
struct: LelouchVC2 look
struct: Standard Error Stream: http://chat.codingame.com/pastebin/3f0a9c32-8f37-4053-8ba1-42473e6d56f6
LelouchVC2: yep
struct: after pac dies it doesnt show anymore
wlesavo: LelouchVC2 if that would be the case all the bots would crash instantly, and they dont
LelouchVC2: I outputted the pacs that i was given right after the pac died
LelouchVC2: it was clearly there
struct: maybe you checked wrong frame
wlesavo: error stream may look to be delayed sometimes
wlesavo: you need to check coors to be sure you looking at the same frame
LelouchVC2: Outputted to standard stream
LelouchVC2: I never use the error stream
LelouchVC2: Set it to output the pacs given right after a pac died, and it gave me all my pacs
LelouchVC2: Maybe fixed now, just saying it pissed me off
WINWINWIN: #432 Silver now, its too tough :(
AntiSquid: giving up too soon
Ramzel: #9 in Silver after so many tries
WINWINWIN: :( I have no idea how to write a BFS so I dont think I can get any further
Ramzel: What language?
WINWINWIN: python3
Ramzel: Cant help u
WINWINWIN: :) which lang?
Ramzel: As i heard graph tasks in pyhon are horrible
struct: WINWINWIN there are lots of resources for BFS online
dbf: WINWINWIN, what issue do you have with BFS?
Ramzel: In c++ it is easily
WINWINWIN: Not learnt it and have no idea how to modify it for this
eulerscheZahl: https://www.geeksforgeeks.org/shortest-distance-two-cells-matrix-grid/
eulerscheZahl: i bet you can understand the code and apply the logic in python
WINWINWIN: Even if I can write one, what is thedifference between my BFS and the in-built pathfinding?
eulerscheZahl: even if i'm not completely happy about the coding style in the article i linked. lots of code duplication
dbf: WINWINWIN, for example, you can calculate actual distance to pellets
struct: How do you calculate distance WINWINWIN?
eulerscheZahl: the in-built path finding gives you less flexibility
eulerscheZahl: and finding the closest pellet of course. good point dbf
WINWINWIN: I copied the distance function in the referee
Csipcsirip: yeah codes in articles like that have usually bad quality
offroff: how deep are you guys making those searches?
eulerscheZahl: only 4 turns :/
Csipcsirip: (2 or 1) * 15
eulerscheZahl: but my opponent tracker improved
eulerscheZahl: https://www.codingame.com/replay/464232411
that kill... :)
AntiSquid: built in pathfinding does more damage than good
offroff: nice one
struct: current rank 1 plays coc
struct: coc is good
dbf: @eulerscheZahl check your tracker for timeouts: https://www.codingame.com/replay/464248625 :)
offroff: I guess 4 turns is enough for some nice tricks
BenWo: that was impressive euler
Rickardo: i just entered level 4
Rickardo: and i am a rookie
eulerscheZahl: 16ms to track, 90ms to print an answer
eulerscheZahl: i don't see the problem dbj :joy:
eulerscheZahl: and i can't type your name. and not tab-complete without pinging dbdr
dbf: haha
dbdr: thanks :D
Ramzel: Clyde has not provided 1 lines in time
Ramzel: when bot gets bugged
eulerscheZahl: i should add a header before legend opens.
"This bot has a kill switch. Do you feel confident to remove it or do you want to choose another boss?"
NapTown: poor clyde
eulerscheZahl: i fear that the boss will be well studied, giving a disadvantage to the player
Ramzel: Lets resubmit almost same code
struct: you can break your bot and then submit again
struct: But you can also not beat boss :p
AntiSquid: tiny change can mean 100 rank difference so
Doju: How is it possible that my last bot which wasted a turn every time there was no pellet directly next to the target ranks 200 places higher than the fixed one :/
LelouchVC2: Because Physics allows for E=MC^2
LelouchVC2: Which invalidates your logic
Ramzel: The funny fact is that i almost always beat boss in ide but i m losing to often with pp close under boss
BenWo: no idea Doju
zarif98sjs: This randomness sucks
BenWo: maybe need some values tweaked
NapTown: Doju when you find out why you can fix it again:grinning:
BenWo: or just needs time
WINWINWIN: The game stops automatically after 200 turns.
WINWINWIN: is this true?
NapTown: yes
BenWo: yes
WINWINWIN: https://www.codingame.com/share-replay/464277907
BenWo: but it's turns, not frames
RoboStac: viewer shows more than 200 frames due to speed though
wlesavo: euler how is it usually done, dont boss have to be resubmitted right?
WINWINWIN: 299 turns
Doju: NapTown yeah prolly that
wlesavo: otherwise cut off wont be so perfect for sure
BenWo: 200 turns, plus 99 SPEED frames
eulerscheZahl: in the past they did submit the boss again. but now they just take the player and copy the rank
wlesavo: i see
eulerscheZahl: they even had a test account
eulerscheZahl: https://www.codingame.com/profile/fc95dc0400be658ec8266b29286952285877531
wlesavo: that make sense in this one for sure
Earthborne: Does activating speed boost take up one round?
BenWo: yep
eulerscheZahl: so we were all chatting and watching default AI climb on the leaderboard which submit slowly finishing
Earthborne: Well that explain some of the mess in my rules
Doju: Maybe i'm timing out
WINWINWIN: Another submit, this time I will start using SPEED :)
Doju: gotta watch some losing games
HKTran: when I try to counter some trapped-SWITCH strategy, I lose to some opponent who not even try any strategy
HKTran: :joy:
NapTown: sounds about right
zarif98sjs: how actually is the standings calculated ?
AntiSquid: someone manually inserts a rank, if you don't believe me just watch top 10 over a long period of time zarif98sjs
Doju: :D
eulerscheZahl: with the Trueskill algorithm
Scarfield: I thought it said in the faq, but this is actually all the faq says:
Scarfield: How are scores calculated?
Your score is awarded when you submit your solution: your code is tested with sets of test cases which cover the game’s specifications. The more test cases your code can pass, the higher your score will be.
Ramzel: Removed defence mechanism and winning almost everything
AntiSquid: for more detailed read on the ELO algorithm zarif98sjs : https://www.random.org/
AntiSquid: (well it feel that way at least)
eulerscheZahl: Trueskill and ELO are not the same
BenWo: lol
NapTown: Gold finally :grinning:
NickStahl: Good job NapTown
BenWo: nice
NapTown: thanks now i can go to bed
Ramzel: Lets go GOLD(submit)
zarif98sjs: NapTown any suggestion for gold ? been stuck around 250-300 for the past 2 days
eulerscheZahl: share a replay of your bot
zarif98sjs: https://www.codingame.com/share-replay/464297339
eulerscheZahl: that's some bad coordination at the bottom right corner
NapTown: zarif98sjs the biggest thing was enemy tracking, mine is very simple but it shot me up pretty quick
Ramzel: Enemy tracking? Isnt is slow much?
NapTown: maybe if it was complicated, but mine isn't
Ramzel: Would u say sth more?
Ramzel: :D
NapTown: :stuck_out_tongue_winking_eye:
zarif98sjs: The enemy tracking seems to be tricky since I can't always see them :|
NapTown: if you could see them you wouldn't need tracking :)
zarif98sjs: :cry:
eulerscheZahl: but understanding your opponent helps you to know which pellets are gone
Doju: yuck, i'm gonna be calling two recursive functions on each move
Doju: that might not be a smart idea
BenWo: wow, what for?
Doju: checking for conflicting targets and paths
Doju: im gonna have some functions which grab new targets until none of them are the same
Doju: i guess i could just do this with while loops, but i like the idea of recursion
Doju: and it cant be that much slower, can it?
BenWo: probably not
Ramzel: eulerscheZahl are u using some of euler tricks?
Doju: actually i think this might lead to an infinite loop :d
BenWo: I'd be surprised if it didn't at least in a few cases
Doju: having a pen and paper close by is reeeally handy while coding stuff like this
SparshMc07: ooooooooooooooooooooooooooo
SparshMc07: kpihlk/k
Doju: wait, that's supposed to be called "pen and paper", not "a pen and paper", right?
Doju: trying to improve my english as well as my coding :p
NapTown: i don't see a difference really
Doju: Oh you're from the US
Doju: i've been wondering why there are very few american people here
NapTown: couldn't say
Doju: hm
Earthborne: Maybe most of them are asleep when we are awake?
Doju: maybe i'm usually just sleeping while you're here
Doju: yeah
[Csongor]: finally silver
NapTown: yes probably
NapTown: 6:30 am where i am
llllllllll: grat! :)
Doju: can you teach me your ways NapTown
Doju: is the secret napping?
NapTown: napping is the secret to everything
alinde2: Consistency winning against the bot in ide, but can't reach it in the leaderboard :upside_down:
zarif98sjs: what's your position now
Doju: Is there a way to do something like this in python?
Doju: for pac in pacs if pac.alive:
Doju: it's messy to write this every time
Doju: for pac in pacs:
if pac.alive:
Uljahn: if not pac.alive: continue
ntroPi: you could just fill an array of living pacs once per loop.
Doju: hm
Doju: yeah i thought of splitting my existing dict of pacs["my"] and pacs["en"] into pacs["my"]["alive"]
Doju: for example
Doju: not sure if it's worth it to start modifying everything though
NickStahl: Doju, why do you keep a ref to your own dead pacs? Why not remove them at start of the loop?
Doju: Hmm, that would make sense as well
Earthborne: You can nest in a list comprehension:
for pac in [pac for pac in pacs if pac.isalive]
Earthborne: Doju
ntroPi: Is it known, whether the pac game will stay active after the contest? Will there be a pause period before the eternal season starts?
Earthborne: Still messy though :)
NapTown: fond memories of pacs lost
Doju: Earthborne ooh yes. Is that fast?
Doju: i like list comprehensions
Doju: NickStahl i think my logic for not deleting them was something like keeping it consistent
Doju: if enemy pacs have ["dead"] and ["alive"] then my pacs should too
Earthborne: Dunno, I wouldn't think so since you're iterating and creating a new list before iterating over that one
Doju: actually i did do have a delete dead pacs method but it doesn't do anything anymore since it relied on vision
ArmoredAssassin: hey guys, am I correct that in step when: Only pacs with the SPEED ability enabled can move, I basically dont get any input and game just moves pacmans if they got cmds to move more than 1 step?
Earthborne: Fastest would probably be `if not pac.isalive: continue`
NickStahl: Doju, mybe you just pointed out a bug in my code :D
Doju: Hahah NickStahl
NickStahl: I think I'm still commanding my dead pacs around D:
Doju: somebody told me that i was commanding my dead pacs
Doju: and then i found that bug
Earthborne: But with a length of max 5, I don't think that's where you'll timeout Doju
Earthborne: Oh, I did that for a while too!
NickStahl: lol Doju, same bug here :D
Doju: Earthborne fair point
Doju: i'm gonna be doing a lot of recursion through that list so i dont want it to be super slow though
Doju: but i guess anything is fast enough with a length of 5
Doju: NickStahl you're welcome :P
Doju: Earthborne i think your solution is good if you change the brackets to braces instead to make a generator
eulerscheZahl: codingame IDE just randomly decided to run a syntax check on my code. now everything is red for no apparent reason
RoboStac: you'll have to delete it all and start again
eulerscheZahl: :D
Doju: yeah your code is just one big syntax error
Doju: sorry euler
eulerscheZahl: https://prnt.sc/shada4
so much red
NapTown: mine has been doing that too
llllllllll: CODE SHARING, ALERT, ALERT
eulerscheZahl: luckily i don't use that web IDE
dbf: code is shared -> ban
llllllllll: ^^
dbf: :expressionless:
eulerscheZahl: that's only the input parsing from the default code
alinde2: That's the default input code @dbf...
eulerscheZahl: and an encoding error at using System; because of my bundler
dbf: no, you shared your secret dx/dy arrays
RoboStac: default c# code has an enemy tracker? :O
weegee: hello
RoboStac: I knew I should have used c#
eulerscheZahl: i beginner's mistake not to take C#
Csipcsirip: isnt't it because of that special character before the second using ?
eulerscheZahl: actually using System; is my first
eulerscheZahl: but the bundler (python) uses a set for unique usings
eulerscheZahl: so order changes everytime i compile my bot
Csipcsirip: one big file ftw
eulerscheZahl: 7 big files
BenWo: ok, got a good part of new pathing done
BenWo: now I have to recode my pacs to understand the output of it :confounded:
llllllllll: MOVE 0 21886 2046702944
ok, something is wrong here
BenWo: lol wow
eulerscheZahl: :D
Doju: nah the target is just a bit further away
llllllllll: my C++ porting didn't go so well
SemChumboPT: those pacs are going to the moon
eulerscheZahl: your pacman will reach that cell by the end of the contest
NapTown: round to 2 digits :D
llllllllll: LOL
MahirSez: While being in speed mode, Is it possible for a pac to go 2 steps where the 2nd step would return to its original position?
MarekM: yes
MarekM: oh so, no probalby
BenWo: no
BenWo: code is coming up on 900 lines :/
BenWo: though there is a fair bit of redundancy at the moment
Rprades: was legend going to be open in 16th always? I thought it was on 15th
Rprades: have they changed the date or did I get confused with the dates?
eulerscheZahl: 6pm CEST
eulerscheZahl: that is in a little less than 5h from now
Rprades: check now, it says 16th May
eulerscheZahl: as 15th for me
Clagus: do I need to beat the boss 100% to advance to gold?
eulerscheZahl: i don't know why, but CG has some problems with dates, showing the wrong day to some users
Rprades: wow, I refreshed the paged and changed back to 15th
Adnmaster: you need to beat more people than the boss Clagus
Doju: No, you need to beat him in the rankings Clagus
eulerscheZahl: you have to rank above the boss on the leaderboard
ntroPi: No you just need to be above him in ranking which is possible even if you beat him <50% of the time
Rprades: thanks :)
Clagus: ok, thanks
BenWo: actually, where are they announcing the date?
BenWo: I haven't even looked for it
eulerscheZahl: click the league at the top
eulerscheZahl: in the IDE
Rprades: it is my first serious contest and i am a little bit excited
Rprades: hopefully I can make it Legend!
BenWo: oh ok
BenWo: thought there was a blog or something
RoboStac: refresh, it'll go back to 15th
eulerscheZahl: about top10-15 players will be in legend when it opens
Clagus: for me it's:
"League opening: 5/17/20" :D
eulerscheZahl: sorry to tell you, but you have to improve your bot first
Rprades: rank 32 atm, hope I can beat the boss later on
BenWo: says 5/15 for me
eulerscheZahl: you'll probably have to improve a bit for legend
Rprades: will try, but Im running out of ideas, python is very slow!
eulerscheZahl: on the other hand: that's by far the largest contest for a long time. maybe they show mercy
Clagus: ok, so I need to be better in ranking than boss, but how can I tell which ranking boss have?
eulerscheZahl: it's even the biggest contest ever
eulerscheZahl: the boss will appear on the leaderboard
Rprades: let's see then!
eulerscheZahl: no way to miss it when it's tehre
Clagus: but for silver it should be there already right?
kovi: too bad, it is so random
BenWo: boss is generally in rank 1
eulerscheZahl: only visible if you are in silver
BenWo: click on "last battles" at left of IDE
<<==========
eulerscheZahl: the randomness is frustrating
eulerscheZahl: there are some deaths that you just can't avoid
BenWo: well gotta run, ttyl
Rprades: yes, speed is much more important than switch
Rprades: most of the games are decided in the frst 15 turns
Adnmaster: well by me, i win by killing my opponent using switch
Adnmaster: so theyre equal
Rprades: unless he is on a dead end or you have speed activated (thus no switch available), your opponent will always be able to escape
Rprades: and if you have speed and he does not, he will be ablw to switch before than you
Adnmaster: i dont use speed :sweat_smile:
Rprades: I think you should :)
eulerscheZahl: +1 for speed
Adnmaster: i made a system that when the enemy isnt near it uses speed if it is near it uses switch
Adnmaster: but for some reason the speed part doesnt work
Adnmaster: it calculates the distance wrong
eulerscheZahl: what means "doesn't work"?
Adnmaster: it calculates wrong
Adnmaster: calculation works for switch, but by me it doesnt work for speed
Adnmaster: even if the enemy is near it uses speed
MahirSez: How many people will advance to legend any idea?
MahirSez: *on 1st go
eulerscheZahl: thibaud discouraging us from reaching legend?
eulerscheZahl: about 10-15 players legend opens
eulerscheZahl: at least that's the usual size. not sure if it will be a little more given the higher player count
Csipcsirip: hm I learned it again that it's not a good idea to make functions like this
Csipcsirip: PacmanType counter(PacmanType pt) {
return (PacmanType)((pt + 1) % 3);
}
dbdr: eulerscheZahl discouraging?
Ramzel: Where is legend league?
llllllllll: opens at 6
weegee: ```js
console.log("hi)
```
Doju: Legend opens today?
Doju: oh dear
Doju: i gotta get at least to the bottom of gold before that
eulerscheZahl: read your emails dbdr
eulerscheZahl: i know you stopped doing that
eulerscheZahl: The Legend league will open in a few hours, but for many players (including me), it's really difficult to reach it.
eulerscheZahl: Improve your own score
It's easier to try to beat your own record than to aim for the moon.
Chkoupinator: speed doesn't limit the ability to turn right?
Adnmaster: no
dbdr: you just cannot turn 180
eulerscheZahl: which is sad
Chkoupinator: yeah makes sense since you can only input one move statement per turn
weegee: how long have you guys been playing?
eulerscheZahl: 8 days
muy31: 8 days
Doju: ^
Illedan: eulerscheZahl, did he tell how big the Legend league will be?
eulerscheZahl: no
dbdr: always a rurprise
eulerscheZahl: Hello, http://chat.codingame.com/pastebin/2c00f879-4943-4a8c-bb21-12cb6b2d4354
Illedan: Hope at 22.63 points :D
eulerscheZahl: no way
dbdr: OOC was a bad surprise for me, was sure to get in at opening, missed it by an epsilon
Illedan: Nah, I hope not
eulerscheZahl: more like 25-26
Illedan: Yeah
Illedan: Then I have to get out of hiding :(
dbdr: 30
dbdr: put a boss above everyone :D
eulerscheZahl: for OOC i was expecting a larger legend because we already had 3 weeks to code ecent bots
Illedan: Take kov i and fix his code? :D
eulerscheZahl: decent*
Doju: Why even is there a league system?
dbdr: take a player bot, improve it, release it as boss
dbdr: yeah Illedan
Doju: or okay, leagues make sense as "markers" for how far you've gone
Illedan: That would be impressive though
eulerscheZahl: to get less noise Doju
Illedan: W00t, only C++ in the top
Doju: Okay, thx eulerscheZahl
Illedan: is it getting that hard u there?
eulerscheZahl: it's a pain to submit with 3k players and climb up all the way in the ranking
eulerscheZahl: 50% of your submit done and you are around rank 100
Illedan: *up
EchoBravo: pls what are constraints
eulerscheZahl: there are some fun observations for tracking Illedan
eulerscheZahl: i shared a replay earlier, let's see if i can find it again
eulerscheZahl: https://www.codingame.com/replay/464232411
eulerscheZahl: the kill was not a coincidene
eulerscheZahl: conincidence
eulerscheZahl: requires you to know where the opponent is and what he's about to do next
dbdr: you can't be sure what they will do
eulerscheZahl: of course not
eulerscheZahl: educated guessing
Illedan: Nice
dbdr: (TM)
eulerscheZahl: analyzed my last battles. tried 5 times, 4 kills
Illedan: Still only 15th with that tracking
eulerscheZahl: was 5th before
eulerscheZahl: unlucky submit
Illedan: Oh, k. Spammers?
Illedan: k
eulerscheZahl: just me losing the first battles
dbdr: damn, that dbdr arena timesout in the IDE
eulerscheZahl: 4 losses in first 10, was at rank 20 when submit finished
dbdr: I wanted to play against him :(
Csipcsirip: yeah this first10 games confuses me this time too. usually a top20 bot wins 9/10. now (4-8)/10
Rprades: is it normal to have a score between 22 to 25 depending on the submit?
Rprades: is it due to randomness?
eulerscheZahl: yes
Rprades: my code is deterministic and I have not changed it
Rprades: ok :)
kovi: robo is on the roll
eulerscheZahl: just in time for legend opening
eulerscheZahl: tomorrow afternoon he won't have time for us, it's codejam
MadKnight: hey eulerscheZahl do u code in c# ?
MadKnight: this contest i mean
eulerscheZahl: no, i use plain text
eulerscheZahl: "if the pacman is chasing you, please SWITCH and kill, thanks"
AntiSquid: are you doing codejam euler ?
eulerscheZahl: yes
eulerscheZahl: that's the tshirt round
eulerscheZahl: finals will be online this year btw
dbdr: max_value: std::f32::BEG_INFINITY
dbdr: what am I tryiing to say there? :D
AntiSquid: have you been to the finals? what are they like?
AntiSquid: not sure what the finals being online means, do you need to turn on your webcam?
eulerscheZahl: either BIG or NEG INFINITY, not sure
AlluringChair: is it normal that i have 600 lines of code and i'm not even close to finishing? i think i'm doing something wrong
eulerscheZahl: i don't know either. codejam just sent an email announcing that finals will not be in person
eulerscheZahl: due to corona
Doju: AlluringChair that's perfectly fine
AntiSquid: you're begging for a win dbdr
dbdr: infinity, I beg you...
eulerscheZahl: i can top that AlluringChair, 1400 lines
Doju: AlluringChair i have 750 and i have maybe 2/3 of my lines written
AlluringChair: ok, thanks for the answers :D
AntiSquid: managed to somehow climb a tiny bit
Doju: Still have to do a lot of messing around with numpy and probabilities
AntiSquid: but need to try harder
Doju: atm pellets either exist or they don't, there's no in between
Doju: but i want some schrödinger pellets on my map
AntiSquid: ooc felt easier, i just was lazy to fix my mine tracker btw
eulerscheZahl: otherwise you would have won the contest?
AntiSquid: yes sure
AntiSquid: otherwise i would have gotten a better rank
AntiSquid: who would you bet wins this contest euler ?
eulerscheZahl: no idea
eulerscheZahl: some strong players are trying
Scarfield: 1:12000 odds on me ;)
jrke: what a lucky match i had https://www.codingame.com/replay/464384338
jrke: :)
eulerscheZahl: i won't put money on you Scarfield
Scarfield: wouldnt be smart xD
AntiSquid: 1 penny, return rate is high
Scarfield: wonder who would pay :thinking:
eulerscheZahl: you. you would have a monitor to sell
eulerscheZahl: who needs curved anyways?
Scarfield: lol thats true
Illedan: Me!
eulerscheZahl: wtf Illedan
Scarfield: ille will buy, bets on :D
Illedan: I need a 3rd monitor for random graphs looking cool
Illedan: Now I sit on a coffetable with only laptopscreen and no external mouse :sob:
Scarfield: why not sit on a chair?
Illedan: haha
cegprakash: http://cgstats.magusgeek.com/app/multi-codebusters/salim.aboubacar why is this guy always getting matched against me? He rekt my rank badly
AlluringChair: make sure it's alluring
Scarfield: xD
cegprakash: He played 108 games and all against me?
eulerscheZahl: lol
cegprakash: oh wait why am I seeing codebusters
cegprakash: my bad
Illedan: Because you spammed..
AntiSquid: no more python RoboStac ?
RoboStac: python bot was only meant to unlock bronze rules, it just worked rather well. Need to go faster to get to the top though
eulerscheZahl: yeah, ceg submitted last
Rprades: it seems I had a timeout sometimes due to many prints!
RoboStac: though I did use it for testing some smaller tweaks while writing a better version
Amrhossam: why do you love coding
Doju: Rprades i feel you hahah
Doju: debugged a working function because it was timing out
Doju: and it was due to prints
Scarfield: lol doju :D 2 hours well spend
Doju: indeed
Doju: lesson learned: prints take time
Scarfield: i think everyone here has done that :)
Doju: maybe
muy31: so ive implemented a basic enemy tracking, now im removing switching so i can progress
Doju: nice
eulerscheZahl: thibaud didn't promise too much. the leaderboard is indeed much faster now
muy31: my code no longer traps opponents tho :(
eulerscheZahl: that's the only thing that my bot does decently well
muy31: i beg to differ
llllllllll: I was also thinking about creating a 'hunter' bot, which just tries to kill others
llllllllll: but im not sure if it worth the effort
muy31: dont do it waste of a pac
llllllllll: but if you kill all the enemies, then you win
muy31: cause winning is about pellets. not killing
llllllllll: but it doesn't sounds feasible
muy31: but most of the time pellets can escape
llllllllll: pacs*
llllllllll: but they cant, if you target one with multiple others
muy31: maybe make a hunter only if there are more than 3 pacs that you have
llllllllll: my plan was that all pacs are hunters
muy31: not good...
llllllllll: yeah, probably
AlluringChair: my pacs are gonna be pacifists, they'll only run, never fight
AlluringChair: unless cornered maybe
MarekM: :)
kovi: most of has pac-ifists
eulerscheZahl: mine fight where the opponent asks for a kill
kovi: 3 hungarians in top, where are you dbdr
Doju: I bet someone is hiding a killer bot that will be strong in this pacifist meta
vtenneke: haha
eulerscheZahl: that killing isn't that simple/reliable
dbdr: kovi I started working on my real bot
Csipcsirip: hm time to add switch command to my repertoire
eulerscheZahl: sometimes there's a good chance but harvest is priority
lifetimeLearner007: anybody calculating distance from every cell to every other cell?
eulerscheZahl: yes
llllllllll: yep
lifetimeLearner007: without timeout
eulerscheZahl: yes
llllllllll: yep, in the first cycle
lifetimeLearner007: i tried with diijkstra but timing out
eulerscheZahl: takes me 20ms on large maps
llllllllll: I do the same, and it doesn't
eulerscheZahl: BFS
Ramzel: Third in silver
Ramzel: 82%
llllllllll: just keep in mind that dist(a,b) = dist(b,a)
llllllllll: and it should not time out
Doju: i don't even do that
Doju: and i'm doing it in 100ish ms
Ramzel: Second in silver
eulerscheZahl: in python
lifetimeLearner007: how does that matter if u've to visit the same cell more than once just bcz source is different
Doju: yes
Doju: in PYTHON
llllllllll: track visited cells
llllllllll: if you visited it, don't update it again
vtenneke: Would more seasoned people recommend to use C++ instead of other languages for this?
lifetimeLearner007: visited cells set is specific to considered source right?
Ramzel: YESSSS
llllllllll: @vtenneke im not that seasoned, but yes.
eulerscheZahl: just check the top of the leaderboard vtenneke
Ramzel: NOOOO
Ramzel: Second place in silver
eulerscheZahl: top10: 1x C#, 9x C++
llllllllll: im currently porting my python code to c++
Ramzel: just below boss
lifetimeLearner007: so close Ramzel
vtenneke: yeah xD I saw but was wondering if people using c++ are just really smart or that the language also actually works better for stuff like this haha
vtenneke: or a combination of course
llllllllll: probably both
eulerscheZahl: i think both
Ramzel: Lets go with next submit
lifetimeLearner007: visited cells set is specific to considered source right? so definitely same cell has to be visited more than once as source changes... else it wouldn't give shortest distance
amriswil-it-1: hola
llllllllll: no
lifetimeLearner007: I'm kinda stuck at this distance calculation. (facepalm)
llllllllll: do bfs and you will get the shortest path for every cell
lifetimeLearner007: bfs doesn't guarantee shortest distance right....
eulerscheZahl: do floyd warshall and you will timeout
Doju: it does
eulerscheZahl: A* doesn't guarantee a shortest path
lifetimeLearner007: then, I guess there is some bug in my code...
muy31: are you snaking?
eulerscheZahl: do you use a stack instead of a queue?
lifetimeLearner007: what is snaking, muy31?
muy31: this sucks! a resubmit and now at 300s
alinde2: A* always gives the shortest path! But your heuristic can be wrong (can't simply use euclidean distance here).
muy31: lifetimeLearner007 snaking is like moving back and forth
muy31: across a map
eulerscheZahl: and with an incorrect heuristic it won't give a shortest path
eulerscheZahl: hard to tell if your heuristic is 100* correct for all cases though
lifetimeLearner007: I'm using queue for bfs, euler
alinde2: With an incorrect heuristic it isn't an A* algorithm.
muy31: https://www.google.com/url?sa=i&url=https%3A%2F%2Fwww.shutterstock.com%2Fsearch%2Fsnaking%2Broad&psig=AOvVaw1kuFlKpXbFkN8HacypnRTh&ust=1589634087029000&source=images&cd=vfe&ved=0CAIQjRxqFwoTCMDzrNL2tekCFQAAAAAdAAAAABAD
muy31: like that, except your BFS may look like that w/ incorrect heuristic
eulerscheZahl: A* just says that you need a heuristic function to estimate the distance from a node to the target. it can't define that it has to be 100% accurate. otherwise that would mean you know the shortest path already
eulerscheZahl: Csipcsirip calm down
eulerscheZahl: you scare me
Csipcsirip: my last submit crashed when all pellets were eaten during simuation..
eulerscheZahl: yet you were on par with me already
eulerscheZahl: damn
cegprakash: my 1 day old code is still at #41 I'm surprised
cegprakash: my latest code is at #120+
muy31: what the hell do i need to get out of silver!!
eulerscheZahl: beat the boss
muy31: shut up
WINWINWIN: lol muy31
Braykin08: lol
eulerscheZahl: :mouth_zipper:
cegprakash: get rekt eulerscheZahl
cegprakash: muy31 go to closest pellet
cegprakash: that's what I did to get to gold
muy31: for real
cegprakash: yes
WINWINWIN: :( he means with a BFS muy31
Doju: Just did a major modification
Doju: let the buggening begin
WINWINWIN: gold worthy Doju?
Doju: for sure WINWINWIN
cegprakash: but u have to do pellet tracking + little team cooperation muy31
Doju: WINWINWIN i'm just hoping for bugs and not a timeout
muy31: so cegprakash, ive been using BFS this whole time
muy31: i implemented a score map to track pellets and enemies
muy31: and to direct movement
muy31: ive independently been choosing my next location
muy31: and now you are telling me that i dont need all this
cegprakash: yes muy31
Clagus: ;_;
WINWINWIN: Anyway to get to gold without a BFS?
Rprades: who wins if both have same number of pellets? I thought it was the one who took the pellets before, but now I lost and he was the one picking up the last pellet, and I have more pacs standing
RavenKZP: how do i know if my pac is dead?
eulerscheZahl: it's a tie
dbf: RavenKZP, it has status 'DEAD' starting from silver league
AlluringChair: you don't get input from it
cegprakash: all u need is go to closest pellet + little team cooperation
Rprades: ah both are 1st
AlluringChair: oh
AlluringChair: really?
Rprades: I got confused because I was in second position
cegprakash: these two are enough to get gold
WINWINWIN: @cegprakash any way without a BFS?
muy31: team coop as in...?
eulerscheZahl: there has to be some order. writing both on top of each other is unreadable
muy31: i already have that
eulerscheZahl: hail our new leader
cegprakash: without BFS u deserve in Wood1 WINWINWIN
Illedan: Damn Csipcsirip :o
Scarfield: coop as in not going for the same pellets, not trapping own unit and the like
kovi: nice csipcsirip
Illedan: The Hungarian top is going to be hard
Csipcsirip: tyty
Illedan: That should be barely enough for Legend
Clagus: damn... and they say that hungarian notation sux...
cegprakash: if I find a middeway between yesterday's bot and today's bot I'm sure I can improve much higher
RavenKZP: @dbf thanks somehow i miss it
cegprakash: I probably have added bugs as I added features
eulerscheZahl: you are far away from legend Illedan
eulerscheZahl: fix it
WINWINWIN: @cegprakash :) Really no way without BFS?
Illedan: I will submit later, just chillin
cegprakash: no WINWINWIN BFS is 4 lines of code
eulerscheZahl: i know, you code at night
cegprakash: learn it WINWINWIN
jrke: hey is sim better than BFS?
Illedan: Yeah, prime time after 10 PM
MarekM: some using recursion in silver or higher?
Illedan: Sure MarekM
Doju: Umh... does anybody spot the issue here?
Doju: print(pac, file=sys.stderr)
pac.command()
Doju: it prints a pac object
cegprakash: BFS can find pellet at 30 distance jrke
Doju: but says nonetype object is not callable
WINWINWIN: https://www.tutorialspoint.com/breadth-first-search-or-bfs-for-a-graph
cegprakash: if u ca write a sim at depth 30 u donn need BFS
WINWINWIN: Is this link good for BFS?
jrke: K
cegprakash: sorry BFS can find npellet even at 100 distance
eulerscheZahl: or you combine sim with BFS heuristic
jrke: ya it can infinite if no time limit is there
cegprakash: u can't go depth 100 using sim
Doju: MarekM i just implemented a recursive function into my code but still got some bugs
cegprakash: not in 50 ms
MarekM: i switch to non recursive algo, in python 3 is way too slow with recursion, now i am under 5 ms per turn, with whole field checked
WINWINWIN: cegprakash is tutorialspoint a good place to learn BFS?
Doju: Okay
cegprakash: just write ur own BFS code without looking tutorial. Learn the concept from youtube WINWINWIN
cegprakash: all u need is a queue
jrke: and push
cegprakash: and u push and pop things out
cegprakash: it's nothing magical
AlluringChair: i have been using python since 4 months and i'm still typing null instaed of none...
cegprakash: damn u are a javascript coder AlluringChair
Rprades: and multiple at the same time, I am using a modified BFS to return distance of every pac from same pellet
WINWINWIN: Hmm, for this I will need a 2D array
WINWINWIN: Uptil now I have been using a 1D array of point objects
AlluringChair: no, before python i only coded in c c++ and c#
cegprakash: use both WINWINWIN
AlluringChair: and i usually just type null and press tab
cegprakash: or use 2D alone
AlluringChair: and it gets me the right null name
AlluringChair: but in python i have to type none xd
cegprakash: it's not none. It's "None"
cegprakash: and null is javascript
AlluringChair: yeah but i type none and press tab for N :)
cegprakash: NULL is c++
Scarfield: you could add neighbors for the point object i suppose WINWINWIN, but 2d arrays are easier to use for stuff thats on a plane imo anyway
AlluringChair: same with NULL nullptr and other crap
MarekM: how many people are in GOLD now?
WINWINWIN: K, will make a full rewrite :(
Gafagarion: 401
MarekM: thx
MarekM: now i am 26th in silver, may be waiting is enough :)
Ramzel: Nope its not
Beerus: did codin ide always gave warnings?
Beerus: nice job anyway
muy31: no Beerus
eulerscheZahl: we have the Monaco editor now instead of ace
jrke: hey python vs c++ which has less reponseime t
struct: ofc you can go depth 100
jrke: *time
struct: is the same as depth 30
struct: it wont do well
MarekM: anybody uses graph theory? with junctions as vertices?
Doju: somebody, yes.
Doju: me? no
Doju: jrke c++ is much faster
jrke: K
cegprakash: jrke there are 4^100 paths at depth 100
jrke: ya
eulerscheZahl: there are pruning possibilities
cegprakash: sorry struct*
eulerscheZahl: of course you can reach depth 100
llllllllll: if you died -> skip
llllllllll: if path is bad ->skip
eulerscheZahl: just without much exploration
cegprakash: someone was asking using sim without a BFS
struct: so 4^30 is possible?
njaber: can we do parallel computing in our bots? Or is it disabled?
alinde2: What are you guys running to need c++? Y'all gold league?
With javascript and bfs in bfs I'm high silver..
cegprakash: can't prune without a BFS
struct: good luck
eulerscheZahl: i can still prune without BFS
eulerscheZahl: just won't do well
njaber: Frick there is no info on the performance of bit arrays on google
njaber: I'm hesitating between taking sha256 ints or bit arrays
eulerscheZahl: sha256?
eulerscheZahl: computing hashes?
njaber: no just for the size of the int
muy31: ok if i win the next three matches against silver bot im submitting
njaber: I want to do bitwise operations as fast as possible on large number of bits
Uljahn: you mean with AVX2?
njaber: ?
Csipcsirip: I'd recommend std::bitset if you use c++
njaber: I'm in c# saddly
njaber: An I don't have time to convert everything to c++
Csipcsirip: https://stackoverflow.com/questions/8144349/what-is-the-c-sharp-equivalent-of-stdbitset-of-c
njaber: Ok doesn't seem like sha256 is really what I'm looking for, I'll see if the parallel implementsations of bitwise data structures are available
njaber: Yeah I saw those but 32 is way too small, and if I can get a hardware accelerated structure I'll go for that in priority
njaber: else I'll just resort to BitArray
MarekM: spatial hashing? :)
njaber: not exactly, it's an experiment more then anything
njaber: Heavy spoilers if it works tho
njaber: So I'm not gonna go into details
jrke: hey is 10 turns enough for simming
jrke: or more
njaber: Hardware acceleration is on, nice
struct: all depends
struct: on the kind of search you use
Doju: njaber what on earth are you doing hahah
Doju: hardware acceleration
njaber: I'm doing things that need as much optimization as possible
Doju: that was a rhetorical question, no need to respond
Doju: mmkay
Doju: that's a solid answer
eulerscheZahl: that's how we lost struct's last account
eulerscheZahl: he got mad coding and forgot to submit
eulerscheZahl: then rage-deleted his account
jrke: is predicting 10 turns are enough for simulation
jrke: ?
njaber: lol, I don't rage, I just like the challenge, but if I fail oh well, I still had fun
muy31: im raging
eulerscheZahl: yes jrke. i'm doing 4 turns
jrke: what
jrke: means then 6 turns are ok
eulerscheZahl: wasted
Hjax: it depends on how sophisticated your scoring is @jrke
muy31: less turns for less sophisticated
muy31: so you dont screw yourself over
Hjax: ? no
muy31: Hjax yes
struct: more depth doesnt mean you will get better results
Hjax: less turns for more sophisticated, because its slower
muy31: thats what i said
muy31: for struct
muy31: but Hjjax no
muy31: i mean yes, but logically no
Hjax: why would you have a simple scoring function with little lookahead? wheres your strength
Hjax: either you have a fast simple scoring function, and can do a lot of lookahead
Hjax: or you have a complex scoring function, but cant search as deep
muy31: no the simple one will screw you...
muy31: if you look too far
cegprakash: bye all I should close this tab and go do some coding for what I'm being paid..
cegprakash: this thing is addictive
struct: Hjax is right
njaber: yes it is
njaber: cegprakash
Doju: yup but this is not a bad thing to be addicted to imo
Hjax: i took today off from work
Hjax: so i could work on this
njaber: indeed
eulerscheZahl: same Hjax
Doju: unless you're already a coding virtuoso
muy31: for example... pac thinks there is # pellets 5 moves away, but there is one right next to you. An unsophisticated algo might choose the pellets 5 moves away, but unfortunately they get eaten before you get there
muy31: but if you gave it less depth you would be more "content" with what's in front of you
muy31: and probably perform better
muy31: do I was talking about logic not runtime
Doju: and this is why i'm not doing a sim
muy31: *So
Doju: also python xD
Uljahn: you could use a discounting factor for future rewards :/
muy31: you could, but that makes it more sophisticated does it not?
muy31: and then you could do a higher depth
Hjax: when i say sophisticated i dont mean little tweaks like discounting based on depth
Hjax: that really wont slow you down much at all
Uljahn: ^
muy31: well whatevs i guess, all of y'all are higher ranked than me anyways
muy31: im done for the day i think, not getting out of silver
Uljahn: not really xd
Doju: uh, what
Doju: in my function definition i have
Doju: "existing_targets=set()"
Doju: oh
Doju: wait... one sec
Doju: found it, nevermind
Doju: aha, just what i expected
Doju: "RECURSION
RECURS...
REC...
Doju: "
times out
Doju: :D
MarekM: :)
AntiSquid: i have a feeling most people in gold aren't really improving but just optimizing vs the others in the arena, change my mind
Hjax: whats the difference
Hjax: a win is a win
IAmNoob: a clean win is not a win
IAmNoob: :)
muy31: resubmitting my old code, seeing what happens...
muy31: AntiSquid improving your code is counteractive to your ranking, it's been proven
AntiSquid: it's ok i have some ideas
AntiSquid: euler's bot looks like a juicy boss target
AntiSquid: wonder if cg staff noticed
Ramzel: Can i chose anyone to fight against?
AntiSquid: yes
Ramzel: How to do it?
AntiSquid: delete opponent in the ide
AntiSquid: then select new one
eulerscheZahl: you can only choose top1000, starting from your own league
Ramzel: Thanks
eulerscheZahl: so you can't fight gold while in silver. or wood because there are too many players in the contest
AntiSquid: which sucks imo
juanotres33: what tutorial made you guys really understand dynamic programming?
muy31: just practiced it
muy31: i guess
muy31: yo just learn the basic constructs at first
Doju: wait, what?
muy31: and then you think about how they apply to your situation
AntiSquid: google, occasionally duckduckgo
Doju: can you modify a default parameter from inside the function
Doju: in python
AntiSquid: yes
Doju: ohhh okay
Doju: so i have an empty set as a default value
AntiSquid: or did you mean smtg else?
juanotres33: my problem with dp is in constructing constructing the (mathematical) recursion. Recognizing patterns I guess... any tips on how to get better on that?
Doju: if i call it with no value and then add something to the set
Doju: (add inside the function)
Doju: and then call it again with no specified values
Doju: then the items i added the last time are there?
AntiSquid: you need an object or variable
Doju: i do have a variable
Doju: but i guess it just points to the default value?
AntiSquid: can make function part of the object
AntiSquid: use global keyword? not sure what you mean
Doju: erm
Doju: def foo(a=set()):
Doju: b=a
Doju: b.add("something")
Doju: then i call that twice
Doju: and on the second time a contains "something"
Doju: is that correct?
RoboStac: https://florimond.dev/blog/articles/2018/08/python-mutable-defaults-are-the-source-of-all-evil/
RoboStac: yep
Doju: oh huh
Doju: Yeah i agree with the title of that article
Doju: thanks RoboStac and AntiSquid
Doju: i think i have these bugs everywhere :|
muy31: im so depressed w/ this right now...
muy31: i just submitted my old code, and am in 258th place, 100th higher than my most recent code
muy31: i remember a day when i was waiting to promote from 9th place with this same code
AntiSquid: interesting URL name
muy31: how is it fair that Clyde-bot gets to keep his rank? but i cant keep mine?
muy31: silver bot suck about as much as i do
muy31: i have a question: What determines the score? like 20.XX?
muy31: how does it work? does resubmitting set you back to 0 and start again, or does it just continue from where you are?
RoboStac: it uses the trueskill algorithm
RoboStac: resubmit starts from scratch (uses first 10 games to roughly place you in the division then you play games vs people close to you in rank)
Hjax: not sure, i dont think it sets you to zero though
Zenoscave: Hjax it does
Hjax: really?
eulerscheZahl: and the boss doesn't keep the rank. whoever gets above, promotes to gold. league size is increasing
Hjax: why not just make the variance bigger?
muy31: eulerscheZahl it was supposed to be a joke..
muy31: and wimper
eulerscheZahl: oh damn, didn't realize to whom i'm replaying
muy31: sp RoboStac it's about first 10?
muy31: cause i always lose like 2 in the first ten
muy31: 2 - 4 variable
RoboStac: they've got a big effect at the start, but you can still move ranks a lot afterwards
muy31: does it become harder to promote as more people join as well?
Hjax: no, it stays the same difficulty, because the boss doesnt get better
Illedan: Might be harder, as some bots are stronger vs others
AntiSquid: .-.
| OO|
| |
'^^^'
muy31: but there's no way someone w/ lazy/simple heuristics is getting better than other players even if we are all better than boss
AntiSquid: where are the pacman ghosts
Illedan: Add ghosts and remove fog of war :P
AntiSquid: ghosts trains !
Over9000: pde-bakk has some nice messages :thinking: https://www.codingame.com/replay/464488929
AntiSquid: https://youtu.be/0bRTVk5ckYo?t=563
muy31: last submit of the day?, 4 losses in first 15 battles
SkyDreamer: can someone help me
llllllllll: with?
SkyDreamer: can i write a function in java that swaps the value of 2 int variables everytime i call it?
llllllllll: I guess so
SkyDreamer: a function i know returns a value, not the swap
isdemus: what do you mean by return a swap?
muy31: make the int variables static or something,
muy31: or global
muy31: you cant swap primitives as parameters
muy31: but you can if they are global variables
muy31: example: functionName(int a, int b){
muy31: a = a +b;
muy31: }
muy31: but a stays the same after method is called
muy31: but if they are not parameters but rather global variables you can do it
muy31: so functionName(int a, int b){
a = a +b;
}
muy31: so functionName(int a, int b){
a = a +b;
}
muy31: *oops
muy31: messed up
muy31: int a = 0;
SkyDreamer: but at the end of that funcion i must return something
muy31: unctionName( int b){
a = a +b;
}
muy31: make it void
muy31: ooh yes, in the 100s again! (silver)
SkyDreamer: http://chat.codingame.com/pastebin/1d7bdfc0-b99b-4672-8339-eb27a37c4828
SkyDreamer: but it doesnt swap
muy31: it wont work
muy31: yeah b/c primitive data types as parameters
k4ng0u: is the initial number of ppl in legend fixed or does it depend on the number of participants?
muy31: it will only work with reference types and objects if you want ou use a and b as a parameters
VicHofs: yo superstars
VicHofs: any ideas how to improve in bronze from this gameplay? https://www.codingame.com/replay/464502929
muy31: otherwise SkyDreamer you need to make a and b global variables
VicHofs: I keep getting close calls like this and have run out of ideas
SkyDreamer: thanks
muy31: VicHofs you are wasting your speed
muy31: you need to choose a location at least two cells away
muy31: for you to use it
VicHofs: huh
VicHofs: ok thanks
VicHofs: I don't know how to use it properly
muy31: frames 22 - 26 or so
VicHofs: I was using if off cooldown
muy31: use it off sppedTurnsLeft
VicHofs: *it
muy31: speedTurnsLeft > 0
muy31: using speed
Scarfield: you get to move 2 cells per turn when speed is active, but if you only target a cell one cell away, you will only move one cell
muy31: also VicHofs you don't always choose the best pellet, either, you might want to keep track of where you know for sure there are no pellets
muy31: many times though you ass up visible pellets
muy31: *pass
VicHofs: I do that already tho
VicHofs: keeping track of no pellets that is
VicHofs: might not be working properly tho
muy31: perhaps
muy31: but dont pass up perfectly good pellets
VicHofs: ty anyway
VicHofs: I think i have an idea what might be contributing
muy31: good luck :)
MSmits: errm, with some luck i could be gold boss, but more likely i will just get stuck under it :P
eulerscheZahl: i hope for a PHP boss
MSmits: that would be smart
MSmits: my bot is not a good choice
eulerscheZahl: you are too low anyways i think
MSmits: right
MSmits: 17 now
eulerscheZahl: i fear that i have a good rank
eulerscheZahl: but i don't want to be the boss :fearful:
MSmits: they dont pick rank 7 for gold boss do they?
MSmits: i thought it was more like 15?
eulerscheZahl: for OOC it was 8
MSmits: oh really
MSmits: explains why it was so hard to beat i guess
eulerscheZahl: they considered dbdr first. but IDE timeouts for a boss? nope
dbdr: :D
eulerscheZahl: in 30min we will know
dbdr: you say they considerd because Apo commented on my indenting?
MSmits: thats probably bad yeah
MSmits: is something wrong with your indenting?
dbdr: destroyed by my byndler
eulerscheZahl: because they commented on you timing out and that that's the reason you aren't the boss i think
dbdr: my source is beautiful of course ;)
eulerscheZahl: on which day was legend opened?
MSmits: but you are still *a* boss though dbdr
dbdr: using tabs, as God intended ;)
dbdr: lol MSmits
MSmits: ah, my tabs are atheist
dbdr: but you are stilling a coding God, MSmits ;)
MSmits: thanks :)
dbdr: *still
RockyMullet: if your a god a believe in yourself, it makes you a theist
dbdr: MSmits you are in the coin flip zone for legend it seems
RockyMullet: self esteem is important
MSmits: I wont get legend immediately I think
MSmits: but often something happens to cause a flood
dbdr: better to prepare for that
MSmits: a submitter that has RPS vs gold voss
MSmits: boss
dbdr: I would be very happy if I do
eulerscheZahl: [CG]SaiksyApo: dbdr il a tellement de caractères qu'il doit supprimer l'indentation apparement è_é
dbdr: yeah, top gold should be fine. though sometimes the boss starts climbing as well
MSmits: yeah
MSmits: we'll see
dbdr: yeah. not sure if I had to, or my budler just did it anyway. it also included the unit tests in the submission :D
eulerscheZahl: and his next sentence: Then you will stay in Gold !
llllllllll: ah I LOVE IT SO MUCH when I timeout and my debug messages are not printed
llllllllll: even if I put it in the first line of the game loop
eulerscheZahl: timeout in game: https://www.codingame.com/replay/464473429
games parsed: 136 timeouts found: 1
dbdr: and I did, bad surprise, I was so sure I did not pay attention
eulerscheZahl: that's acceptable i guess
MSmits: I have them sometimes too, but I dont know why
MSmits: pretty rare though
llllllllll: yeah it is
llllllllll: which makes it specially annoying
dbdr: lol, now you are above me MSmits
dbdr: 0.01 between 4 bots
MSmits: damn
Uljahn: llllllllll: put them after the first inputs in the loop also add flush=True if it's py3
MSmits: my bot is weird though. I did a submit that ended at 170 as well
MSmits: it does relatively well against top bots
MSmits: but I still get trap-killed a lot
llllllllll: @Uljahn now its C++ & I have a print after each function call in the game loop. still nothing is printed
MSmits: I'm not touching my submit until after legend opens
eulerscheZahl: same
VicHofs: ty muy31
YvanTCHOUDIE: Hello everyone
AntiSquid: looks like everyone is getting their best bot out of their cupboard
alinde2: I'm climbing a rank every couple of minutes, just gotta wait it out eh? 42@silver
AntiSquid: python at #27 , is that the boss ?
eulerscheZahl: no, too low
AntiSquid: maybe they gonna take it and improve it
MSmits: if they pick the php, things will get really painful
eulerscheZahl: PHP at 4 is too high
some C#/Java around 10-15
MSmits: C#/java, might not be better than just c++
MSmits: many will use the full calc time
eulerscheZahl: or JS. but that's low again
MSmits: bah I'm dropping
eulerscheZahl: you shall not pass!
MSmits: I shall!
eulerscheZahl: i want to see you suffer :popcorn:
MSmits: :cry:
eulerscheZahl: for shooting at my submarine
MSmits: oh, right that one
MSmits: this is a pretty random ladder, just like ooc
dbdr: some bots suffer in gold, but then zip through great positions in legend...
Illedan: 17-18 is a god spot to pick the boss
dbdr: why?
eulerscheZahl: i bet it will be around 10
Illedan: 24-25 score is very close
AntiSquid: fingers crossed for top 5 legend
Illedan: Need the score around 25.5
eulerscheZahl: 25.5 promotes like half of gold
MSmits: back in a bit, someone push me while I am gone
AntiSquid: push down
VicHofs: petition to make it a win if a player gets 69 pellets
icecream17: no, do it when a player has 0 pellets but 100 turns have passed
VicHofs: that's fine too but dude
VicHofs: 69
VicHofs: lmao
icecream17: 69 turns
icecream17: oh well submit
icecream17: wow it's updating fast
Uljahn: time to make some mass promotions in csb :smirk:
eulerscheZahl: where is my boss? :fork: :knife:
eulerscheZahl: no fork?
eulerscheZahl: :spoon:
Scarfield: there is no spoon
icecream17: the fork is lost in the cake
dbdr: late by several seconds already! ourageous!
SurajMishra2209: I have made all the improvements I could and I still cannot get out of bronze
SurajMishra2209: I am giving up
Doju: Don't give up SurajMishra2209
VicHofs: ooh don't give up dude
VicHofs: I suck too
Uljahn: ye, just gid gud ezpz
dbf: SurajMishra2209, don't give up, silver doesn't require complex code, just make sure that there are now visible errors
jrke: does legend open yet
Doju: ^ words of wisdom right there
Doju: (dbf)
Doju: xD
icecream17: This is aargh. I lose 5x times more often if I lost last time in the submit battles
VicHofs: not sure if I agree
eulerscheZahl: yes jrke. any minute now
Doju: VicHofs is your code timing out?
VicHofs: nope Doju
VicHofs: I just suck
Doju: are your pacs collaborating?
VicHofs: lmao
VicHofs: what does that mean
Doju: at least not crashing into eachother
Doju: for extended periods of time
Doju: and not following eachother
VicHofs: yeah they don't crash
VicHofs: I can't stop them from following yet tho...
Doju: that not following thing was key for me to getting to silver
VicHofs: any tips?
Doju: do you have a pathfinding thing working?
Doju: (bfs perhaps)
daanolav: So does anyone know how late legend is opening?
Hjax: soon™
Illedan: Any time now
BenjaminUrquhart: make kovi the boss /s
eulerscheZahl: i want a PHP boss
BenjaminUrquhart: but the php guy is 4th
Csipcsirip: c# would be much better
BenjaminUrquhart: ey
eulerscheZahl: boss
BenjaminUrquhart: java boss
wlesavo: yay
BenjaminUrquhart: Npi2Loup
wlesavo: big legend
eulerscheZahl: yay https://prnt.sc/shgjyw
Illedan: 25.5 :tada:
Illedan: Told ya ;)
BenjaminUrquhart: MSmits you just missed it
eulerscheZahl: will blow up quickly
Illedan: :clap: eulerscheZahl
Rprades: 15 ppl to legend i tseems
eulerscheZahl: 16
wlesavo: so Npi2Loup is the boss?
BenjaminUrquhart: I believe so
jrke: 16
BenjaminUrquhart: or it's another java bot that so happened to have the same score as him
eulerscheZahl: https://www.codingame.com/share-replay/464552384
https://www.codingame.com/replay/464551779
Illedan: Random boss?
Hjax: looks deterministic
eulerscheZahl: i thought my replays show the opposite?
dbdr: now we see the real ranking free of spam? :)
eulerscheZahl: no, someone will spam before it finishes
Illedan: :D
MSmits: 0.32 rating below gold
Illedan: haha
eulerscheZahl: and late promotions
MSmits: thats like 3 games
dbdr: ValGrowth back to #1
kovi: finally we will have a real eval
kovi: with best versions and quite a few games
MSmits: 0.17 below gold boss
MSmits: karliso promoted
eulerscheZahl: and 2 players already resubmitted
dbdr: 3
eulerscheZahl: you too dbdr?
AlluringChair: can i somehow get the board in text # #... so i can test it locally?
dbdr: no
BenjaminUrquhart: print it
dbdr: XL* karli* Jir*tech
eulerscheZahl: your second one is a promotion actually
dbdr: and w*la
eulerscheZahl: not a resubmit
dbdr: ah possible
dbdr: sry
solaimanope: will one get promoted if it's better than boss when BATTLES IN PROGRESS
eulerscheZahl: no
RockyMullet: good legend is out, time to bump my face a couple of times on that gate
dbdr: 3 HU in top 5 :D happy kovi? :)
eulerscheZahl: you don't like the boss?
Illedan: I'll join RockyMullet
dbdr: 3 in top 3
dbdr: ok, chnging too fast
RockyMullet: good good
MSmits: You are better than the boss! You will be promoted to Legend League
RockyMullet: ok first game in IDE I win, I might be lucky by spamming the submits :P
dbdr: yay, gg MSmits!
wlesavo: nice MSmits
RockyMullet: gg MSmits
MSmits: no resubmit needed :)
MSmits: thanks
RockyMullet: resubmits are for losers
RockyMullet: like me...
**RockyMullet plays violon
dbdr: I have submitted since yesterday morning :D
MSmits: first time i get to legend this way. All previous time I was either above boss, or i got it like 2 days later
RockyMullet: ok trying my luck
RockyMullet: THEN I'll code
AlluringChair: i'm probably dumb but how can i print the board, the console only shows 1 line and i can't print another..
MSmits: luck before code?
icecream17: save each line in an array
RockyMullet: AlluringChair you can print in the errors
Doju: yeah you need to print in the errors
RockyMullet: MSmits yes, cause luck is easy
MSmits: if you're lucky :)
alinde2: 30 silver, now winnings against ide bot. Gotta re-submit :grin:
AlluringChair: that stil wil be 1 line, \n doesn't work :/
icecream17: http://chat.codingame.com/pastebin/85b886e4-d0f5-4166-997f-b331bfcbb796
RockyMullet: :four_leaf_clover:
RockyMullet: gl alinde2
AlluringChair: i guess i will separate it later...
icecream17: Alluringchair see the pastebin
Doju: hmm
RockyMullet: oh this luck thing is not looking good lol
kovi: we will have good baseline cgstat
RockyMullet: oh wait, the luck is coming back
Waffle3z: how small is legend?
MSmits: 20
MSmits: 21
RockyMullet: oh, nice
RockyMullet: not one of those "10 people in legend"
RockyMullet: so you're telling me there's a chance !?!?!?
eulerscheZahl: of course
MSmits: this game is pretty random. Get in range of gold boss so you get some battles against him. Overfit vs the boss, then chain submit
RockyMullet: trolling aside, I might be able to get legend
eulerscheZahl: even for 10 in legend, there is a chance - as these 10 prove
RockyMullet: yeah depends if I'm already in the downward spiral of "every I do make it worse" which is, for now, not the case :P
MSmits: hope I am not there either... havent really improved much since monday
MSmits: even though i did some pretty cool stuff in my bot. It just didnt help :P
RockyMullet: yah, I fixed so many edge case that happens once every 100 games, its nice, but doesnt affect the rank much
solaimanope: "even though i did some pretty cool stuff in my bot. It just didnt help" sums up my last two days :3
MSmits: right
RockyMullet: rn most of my game I lose because when tehre is not much pellet left, it takes too much time for me to find them, so I work on my deduction of where enemies have most likely already taken the pellets
RockyMullet: should help
MSmits: yes, thats on my list as well
MSmits: and finally try to kill and avoid dying :P
RockyMullet: knowledge is power :sunglasses:
MSmits: ye
Rprades: same for me RokcyMullet
Rprades: that is my bottleneck right now
RockyMullet: yah, Rprades, I improved my early gathering A LOT but then I slowly lose if it goes late game
RockyMullet: which is usually the case against good bot
RockyMullet: so unless I kill pacs (which my bot is ok at) I lose the long games
MSmits: you check for dead-end kills RockyMullet?
RockyMullet: yes
kovi: tiber has nice lead, 2-4 is pretty equal
MSmits: did you test how much stronger your bot is with the kill thing RockyMullet?
RockyMullet: hum ahrd to tell, I did that a couple days ago
Peudoki: time to join the legends for me hopefully
RockyMullet: but it doenst happen everygame
MSmits: I am going to check the last battles and see how many times it happens to me.
wlesavo: 24 already
MSmits: do an accurate count
wlesavo: looks promising
RockyMullet: I dont activly check for a kill, its just if I see an enemy in a dead end that I can reach before they can exit, I go for it
Peudoki: half the games done, 0.67 ahead of boss
Peudoki: 0.49 now
MSmits: RockyMullet thats exactly what i mean
RockyMullet: I seen a game once where it wasnt the right choice tho, I was losing by like 3 pellets and chased for the kill while I could have taken pellets and win
RockyMullet: killing an enemy pac is not as valuable late game
k4ng0u: never chase in end game :P
RockyMullet: yeah, I could probably do some dumb check like "if im that close from the end, dont go for kill"
k4ng0u: I read somewhere that you could divide the pellets left by the number of alive pacs to get an approximation of the value of killing a pac
MSmits: makes total sense
RockyMullet: yeah, not a bad idea
Doju: isn't it double that?
Doju: since they can't collect pellets and you can if their pac is dead
MSmits: doesnt matter, there will be a constant anyway
Doju: sure
MSmits: if the value is doubled, the constant will also and you will fit the constant
jrke: now tommorow we can make out who will be the winner
Kellthazar: if myScore < enemyScore, dont ignore the enemypac
Kellthazar: ops, I mean the oposite
Doju: It took me 8 hours to code a simple collaborative targeting thing ...
MSmits: most of my losses are corner kills where I dont know my opponent is around the corner =/
Kellthazar: yeah
WINWINWIN: SIlver boss occasionally has timeouts, is that ok?
eulerscheZahl: yes, it's fine
Doju: luckily i can quickly adapt it to work for paths too
eulerscheZahl: the boss is just another player like you and me
AlluringChair: i have just wasted nearly 3 hours of my life because deque in python with pop() works like a stack
Doju: you need to do popleft()
AlluringChair: THANKS
AlluringChair: I KNOW
Doju: sorry xd
AlluringChair: why name it DEQUE if it might as well be a stack..
Doju: double ended queue?
AlluringChair: e maybe i;m just tilted beyond reasoning now
kovi: very random. top6 is in a bunch now
Peudoki: I'm 0.00 behind the boss, nooo!
Peudoki: can't get to legend
MSmits: so.. .you have floating point issues?
Peudoki: we're both at 25.23 when I finished my games
ZarthaxX: lol
MSmits: Peudoki, the first one of you to win a game will be 1st
MSmits: so just win your next game
MSmits: and hope he loses his
eulerscheZahl: your score has more decimal digits than shown
Peudoki: I have no games left though
eulerscheZahl: others submit
MSmits: players will play against you..
Doju: yeah Peudoki just hype your bot up
NinjaDoggy: :O he did it
MSmits: .and against the boss
NinjaDoggy: the boss lost
Peudoki: it happened! I'm legend!
Doju: gratz
MSmits: grats :)
NickStahl: Peudoki congrats
Peudoki: well
Peudoki: in 30 sec
Peudoki: I have to wait it out
Peudoki: and hold
MSmits: no
ZarthaxX: nah you cant
MSmits: if it gives a popup, you're in
ZarthaxX: once u get that msg you promote
ZarthaxX: even if you resubmit
Peudoki: it's official!
eulerscheZahl: it was before
Ramzel: YEY GOLD !
WINWINWIN: well donw Peudoki, Ramzel
Peudoki: thank you
NickStahl: Ramzel congratz
Peudoki: I'm dead last in legend, but I'm here ^^
Peudoki: the smallest legend known to humanity
eulerscheZahl: nah
MSmits: it will grow
MSmits: it's always small at first
eulerscheZahl: 44 at the end of the contest for code of ice and fire, 49 for mean max
MSmits: this one will be bigger
eulerscheZahl: this one will exceed 100 i think
Peudoki: 100 legends? :O
eulerscheZahl: just wait till monday
alinde2: Edge of my seat https://www.codingame.com/replay/464600335
kovi: its hard to climb with 55% winrates
Doju: only 2 days left... i guess it's time for bad coding
Doju: ctrl + c and ctrl + v to the rescue
wlesavo: does legend feels less random?
kovi: not at all
WINWINWIN: I dont understand what you mean by random wlesavo
WINWINWIN: Are all the bots deterministic in legend?
Xlos: Alright I'm finally beating inky's butt
Xlos: silver here we come
wlesavo: WINWINWIN random pac deaths are almost unavoidable, so it can result in a loss with nothing you can do about it
Rprades: 26% battles and beating the boss!
Rprades: hope it lasts until 100%
WINWINWIN: Thanks wlesavo, understand now
wlesavo: Rprades quite huge advantage actually, 1 point above the boss
Rprades: just lost 7 in a row :(
Ramzel: Is there a way to fall a rank down?
wlesavo: i think you will make it anyway
Ramzel: league down*
Rprades: thanks for the trust wlesavo
wlesavo: Ramzel no, only in test mode
Quidome: Legend is already big, congrats guys
Quidome: Well played :)
mpt: still got 800ms to spent on precalc. What should I use it on, I wonder.
kovi: increase depth of 1st turn action
jacek: make opening book
mpt: thanks for the input
NinjaDoggy: mtp
NinjaDoggy: calculate cuts/trapping :P
MSmits: cuts>
MSmits: ?
MSmits: you mean dead-ending a pac?
NinjaDoggy: yep
MSmits: kk
MSmits: stil havent coded that
MSmits: seems fairly easy, but it will add more mess to my mess
NinjaDoggy: how's the enemy tracking going now?
MSmits: I dunno.. I have something that works, but it doesnt help me win more
eulerscheZahl: trapping is fun
MSmits: i mainly try not to die around corners, but it makes my pacs afraid to go around corners :P
eulerscheZahl: i try to kill around corners :imp:
MSmits: eulerscheZahl have you reduced your trapping? I saw a replay from you where you had 2 pacs travel quite far to kill another pac... thats pretty costly
eulerscheZahl: i still do it
MSmits: do you take into account the cost though? For example near end games?
NinjaDoggy: evil euler :imp:
eulerscheZahl: no, i just like kills
MSmits: lol ok
NinjaDoggy: LOl
MSmits: it seems pretty complicated to code with multiple pacs involved
MSmits: dead ends are pretty easy
eulerscheZahl: like this? https://www.codingame.com/replay/464470922
MSmits: yes that
some_person_that_dan_doesnt_know: helo
some_person_that_dan_doesnt_know: 14
eulerscheZahl: it's not that hard to code if you are happy catching 95% of the cases
MSmits: you also need to take into account your other pacs blocking your trap
MSmits: stuff like that
eulerscheZahl: and that's where i screw up :D
MSmits: hehe ok
NinjaDoggy: hm...
MSmits: thats my problem. I see a problem I dont know how to fix and then I dont code an less effective version that would still work 95%
RockyMullet: hu, I could probably improve my "kill in dead ends" algo to make traps like that
some_person_that_dan_doesnt_know: where did the button where u can join a random 15min clash go
Nerchio: do we know who is the gold boss?
NinjaDoggy: wait you didn't detect trap on frame 33 but did on fram 37?
NinjaDoggy: you have to have a pac next to them? lame -_-
eulerscheZahl: npi2 is the boss
NinjaDoggy: that's not 95% :P
CodingCarter: 2 hours b4 AP Computer Science A!
MSmits: NinjaDoggy
MSmits: it's cooldowns
MSmits: the opponent can switch
MSmits: then you cant trap him
NinjaDoggy: oh nvm i'm dumb :(
NinjaDoggy: but you can also switch!
NinjaDoggy: ha! lame -_-
MSmits: yes, but that doesnt work
MSmits: the one who switches first loses
MSmits: you get a standoff
MSmits: takes too much time, tying up two pacs vs his one
MSmits: not good
NinjaDoggy: true
NinjaDoggy: but he doens't care, he just wants kills
MSmits: thats true. Out of character euler, you should park two pacs and wait
eulerscheZahl: hey, i won the game
NinjaDoggy: :P
eulerscheZahl: i also saw my bot doing this in turn 2 with bad start locations
MSmits: yeah... seen plenty of those
MSmits: all games I lose
DiegoVega: can anyone tell me how the starting positions are chosen? is it truely random or there is a fairness involved so that your pac is not in an impossible starting position?
dbdr: it's fair since it's symmetrical
Peudoki: my AI just went on a winning stream of 13
Peudoki: crazy rampage
Peudoki: streak*
MSmits: I turned that mode off... felt like cheating
Doju: :D
eulerscheZahl: https://prnt.sc/shibdp :D
MSmits: see, that's sporting. Giving others a chance
MSmits: try that Peudoki
dbdr: are you on stike euler?
some_person_that_dan_doesnt_know: h
some_person_that_dan_doesnt_know: h
some_person_that_dan_doesnt_know: h
some_person_that_dan_doesnt_know: h
h
h
h
some_person_that_dan_doesnt_know: http://chat.codingame.com/pastebin/693a0aea-2cc3-4115-87c3-de3cc25e9b31
some_person_that_dan_doesnt_know: http://chat.codingame.com/pastebin/94de7a21-549a-440b-b2b7-b5b58812951d
some_person_that_dan_doesnt_know: http://chat.codingame.com/pastebin/c1a425c2-6df8-4463-beb9-d967d756e786
some_person_that_dan_doesnt_know: h
h
h
some_person_that_dan_doesnt_know: h
h
h
llllllllll: stop, maybe?
njaber: smells like imminent mute/ban
NinjaDoggy: where the mods at
some_person_that_dan_doesnt_know: h
eulerscheZahl: chose your mod
NinjaDoggy: euler aren't you a mod?
eulerscheZahl: i'll ping antisquid
eulerscheZahl: but tab completion doesn't trigger
NinjaDoggy: why aren't you a mod :(
struct: I ping MadKnight
eulerscheZahl: i'm only on discord :(
NinjaDoggy: weak
cegprakash: eulerscheZahl "i also saw my bot doing this in turn 2 with bad start locations" can u show me
eulerscheZahl: didn't save it
struct: Need top 100 in CSB for mod
eulerscheZahl: sqiddy got that?
NinjaDoggy: i don't believe it
NinjaDoggy: not 1 bit
struct: yes
cegprakash: what is it about eulerscheZahl
struct: rank 91
NinjaDoggy: :O
some_person_that_dan_doesnt_know: h
daanolav: i don't even know what i can do to improve my bot anymore, that's how bad i designed it in the first place
Jesus_inn_party: sup every one
Doju: sup jesus
cegprakash: out off topic: have anyone worked on "sign in with google" on backend? I need help
jrke: hey i wanna ask one thing in c++ suppose their is nested loop and if command break; then it will exit from all loop or one
eulerscheZahl: and you can't google for help because you aren't signed in?
eulerscheZahl: jrke for my prof that's the only acceptable use of goto
cegprakash: eulerscheZahl too many documents
cegprakash: don't know which one is right
jrke: K euler thanks
struct: it only breaks the current loop
struct: not nested
EvModder: :O they added a new arena
EvModder: time to submit again and hope i make the cut
eulerscheZahl: which arena?
eulerscheZahl: i see no new multiplayer game
EvModder: legend
eulerscheZahl: oh :D
jrke: hey when i add goto to switch it time out else works why is this in c++
struct: How does your switch looks?
EvModder: idk, should work. i dont use gotos much personall
jrke: https://pastebin.com/81jcYTqL
jrke: my switch
struct: you need break; on switchs
jrke: same timeout then also
jrke: what can i do
cegprakash: goto guy spotted
jrke: means
cegprakash: if something works ship it
jrke: Ceg i didn't get you
[Csongor]: The boss just timed out...is that normal?
cegprakash: you've to type "goto cegprakash;" to get to me
eulerscheZahl: yes, the boss can timeout too
cegprakash: probably eulerscheZahl is the boss
EvModder: noob boss lol
EvModder: what causes it to timeout?
jrke: hey anybody to my ques plz
cegprakash: EvModder thinking
cegprakash: thinnking too much
EvModder: i dont beleive in thinking
[Csongor]: the boss timed out again
struct: jrke what is s4?
EvModder: link to game?
cegprakash: [Csongor] definitely it's eulerscheZahl
cegprakash: if it timeouts often
cegprakash: let me guess there are 4 or 5 pacs
jrke: its just an int s4
struct: What is this switch supposed to do ?
struct: What do you want to do?
jrke: i am using it for sim
struct: Yes but what is it meant to do?
jrke: its one of my 6 switch to read all possibilities for next 6 turns
jrke: how can i solve this probleme
jrke: *problem
struct: are you using goto to exit switch?
jrke: break and goto both timeout
Counterbalance: jrke you forgot the break;
jrke: nope its also giving timeout
eulerscheZahl: the goto is conditional
eulerscheZahl: you still need an unconditional break
jrke: means
eulerscheZahl: oh wait... just poorly formatted
RoboStac: no, it's outside the {} so it's the same as break
Counterbalance: right
dbf: eulerscheZahl, congrats with legend, boss is in java, so you are happy? :)
Rprades: come on
eulerscheZahl: no, i wanted PHP
Rprades: first after the gold boss
Rprades: haha
jrke: whats wrong in this code
Rprades: story of my life
RoboStac: theres nothing wrong (other than the formatting) with that code jrke
jrke: so what i am suppose to do?
Counterbalance: might be because there are no bounds checks here so maybe you're accessing somearray[-1] ?
dbf: Rprades, just leave your bot overnight and check tomorrow morning :)
Rprades: do you think that without submiitting there is a chance to get promoted? if he loses more than me?
Rprades: thanks dbf, that is what I was asking
Rprades: I wont resubmit then
eulerscheZahl: odds are good to get pushed
eulerscheZahl: only submit if you have a real improvement
dbf: the gap is not small, but there is allways a chance for #1
dbf: yes, don't resubmit old code
eulerscheZahl: how big is the gap?
Rprades: thanks :)
Rprades: 0.9
dbf: 0/85
eulerscheZahl: still in range of random
Rprades: crossing my fingers
dbf: close the leaderboard, start working on large improvement and check tomorrow :)
dbf: (pro-tip from a person who was never in legend, lol)
Rprades: haha
Rprades: thanks
cegprakash: dbdr it doesn't work lol.. small iffs are the ones making diff atleast to me
Illedan: Open the webpage in Inspect. Remove chat, ranking and leaderboard. Continue to work.
eulerscheZahl: add ?disableChat to the URL, way easier
dbf: wow
Illedan: Wont remove leaderboard and ranking
struct: Just dont submit
dbf: so I need +4 scores, looks like a major improvement is required :(
eulerscheZahl: hm, i like the idea of removing that distraction too
HiIamFriend: https://www.codingame.com/replay/464686757 Aham, what was that
Illedan: I removed those on that 4 hour sprint :P
cegprakash: Illedan I'd love an extension for that
eulerscheZahl: looking at the rank too often despite knowing it's meaningless
Illedan: ^
HiIamFriend: how do u beat that :DDD
HiIamFriend: explain pls
MSmits: it's a special move. Output SLURP_PELLETS
Illedan: I wish this game had a 5 BFS ranged vision instead of along the lines -.-
HiIamFriend: hmm
MSmits: HiIamFriend i am kidding... you timed out
dbf: Illedan, so you can use more code from unleash the geek? :)
HiIamFriend: ow :D
MSmits: meaning you didnt provide a proper command
Illedan: Or OoC :P
HiIamFriend: thanks :D
MSmits: np :)
jrke: https://pastebin.com/81jcYTqL
hey anybody just solve this problem plz
i am getting timeout on applying break or goto statement
and works without break and goto why?
MarekM: jrke: which language?
jrke: c++
MarekM: Use of goto statement is highly discouraged because it makes difficult to trace the control flow of a program, making the program hard to understand and hard to modify. Any program that uses a goto can be rewritten so that it doesn't need the goto.
struct: jrke I think someone already mentioned the possible problem
jrke: hey but break is also not working
MichaelVonWolfen: Try and change that switch to if statements
struct: if px - 1 == -1
struct: you gonna check array[-1]
jrke: K
Doju: oh great, my pacs see ghosts
HiIamFriend: so if i am rank1 what does that mean O_o "last battles"
NapTown: then you're promoted
HiIamFriend: so i wrote a single line code and i won ? thats weird
Illedan: what league?
HiIamFriend: nvmd, wood league
HiIamFriend: i am not first now
Illedan: :)
Illedan: Wood is introduction
HiIamFriend: wood 1 :D
HiIamFriend: this is a fun website tbh
HiIamFriend: i was searching to code for fun
Nerchio: welcome !
HiIamFriend: ty
AntiSquid: hey was away, how many did get promoted initially ?
RoboStac: 16
cegprakash: AntiSquid I think eulerscheZahl's timeouting bot is th eboss
AntiSquid: java
cegprakash: oh then Npi2Loup may be
HiIamFriend: does enemy is represented by "#" or no ?
cegprakash: no HiIamFriend
cegprakash: # is wall
[Csongor]: the walls
HiIamFriend: k
wlesavo: ceg teaching about the rules :smirk:
cegprakash: enemy is represented by mine=false
cegprakash: I get helped from CG community, I give it back
cegprakash: so never troll CEG
wlesavo: CeG
ashikujjaman: which algo is best for silver league?
Peudoki: anything people use in gold and legend
wlesavo: btw, cegprakash, what does your nickname mean? is it an actual word?
cegprakash: I studied at CEG
cegprakash: but everyone in my college call me CEG coz of my handle
cegprakash: so I started using it everywhere
cegprakash: and when I came to CG everyone started calling me CEG as well
wlesavo: is prakash your lastname then?
cegprakash: firstnname
wlesavo: oh i see
cegprakash: I don't have a lastname
struct: if you have a firstname you have a last name
Doju: hm
Doju: hah
Peudoki: you can have only one name
Peudoki: that isn't the last
Doju: wdym
Doju: i have 3 names that aren't the last
AntiSquid: but what about 1 first name and 20 middle names?
[Csongor]: 3 nicknames
wlesavo: struct sounds like a programmer would say
wlesavo: like smth
wlesavo: nice Illedan
Rprades: yes! congrats!
Rprades: that give me hope, now I am 0.5 away!
dbf: :thumbsup:
ameler: gg Illedan
eulerscheZahl: legend is too crowded already, can we get a new league?
dbf: yes, if you take gold boss with you
Hjax: In 2 days 12 hours the last league opens, its called champion :P
kovi: why wait so long, do it now
wlesavo: 40 ppl already wow
Jesus_inn_party: how do you know this much stuff?
Nerchio: i wish submits were faster so i can spam submit and hope for a lucky legend run
Astrobytes: cegprakash "I don't have a last name" - yes you do and I know what it is :P
Illedan: w00t
Illedan: Still my code from last saturday with isdead tweak :D
Just changed some numbers lol
Colios13: Tourist is coming :astonished:
Illedan: Nice
Illedan: Rank 3k
Illedan: :D
fatrujilloa: Guys I keep getting timed out but I am not sure where the problem comes from... any hints in how to find the problem?
Doju: print everything
Doju: except if you print too much make sure it doesn't time out because of that
Waffle3z: made a change and submitted. 22nd to 175th. resubmitted. 1st
MarekM: measure time on code segments
fatrujilloa: I am doing it... it gets to the end of the dycle actually :S
fatrujilloa: How do you measure time?
Astrobytes: Psyho, tourist... who else will join
Illedan: psycho didn't do much right?
fatrujilloa: I am using Python
Astrobytes: In gold last time I looked
MarekM: start = time.time()
MarekM: then same at finish
fatrujilloa: Thx!
MarekM: substract and print
Doju: it's hilarious that if you google "pacman mouth angle"
Doju: it actually gives you the answer
Astrobytes: Still have a small bet with myself that dapo will show up at the end :D
Doju: in one of those little handy quick answers
kovi: quite a few topcoder marathonists play
kovi: i envy how fast romka descended
Illedan: Astrobytes :D
kovi: at this point my vote is on karliso
Illedan: I vote for T1024
NapTown: Betrayus switch counter:worried:
Illedan: Or myself if I every get this code working xD
eulerscheZahl: i bet karlis o too
eulerscheZahl: no Illedan. you just aren't a contest killer
Illedan: I know, my motivation was done last saturday :P
Illedan: I still have 3 beers and a redbull. Will see what the night brings
Astrobytes: that's some submit by Romk a
MattSOrme: I thought id made loads of improvements. Played against the boss and won a couple games without paying much attention to anything but the final score.
Submitted and list 600 places...
eulerscheZahl: headache for the next morning
Illedan: Used to that
MattSOrme: All of the cool stuff that i had working this morning, was broken by the stuff i did this afternoon :(
cegprakash: I think the rate at which I stare at leaderboard has been increased since the leaderboard fast updating from CG
Astrobytes: lol, it is a significant improvement
struct: Romk a with a new bot?
Astrobytes: It's impressive
Doju: what on earth
Doju: my pac updating method is so broken. How has it worked till now
Doju: Or maybe it hasnt :/
Astrobytes: lol just noticed it's a top 10 Hungarian takeover :D
kovi: psyho playing btw
MarekM: gold league, after all, in the middle of debugging :)
EvModder: did all of the other top players get a lot better since yesterday or did i get a lot worse lol
EvModder: probably both xD
MarekM: or out of luck
EvModder: ya i agree. i will blame luck
Adnmaster: so you work in google
Adnmaster: nice
AntiSquid: he is part of the search engine itself
Icebox: he *is* the search engine
EvModder: kind of ironically, i dont think my pellet search algorithm is much good
DevLugo: Hello!, about the batman binary search...How i know the current position of batman?
Icebox: you track it
Icebox: yourself
Opeyevil: Am i able to move Coders strike back into visual sutdios to play and debug properly? i still dont understand why next checkpoint x and y change with every loop.
DevLugo: ok i think i have an idea. Thanks!
cegprakash: yes u can Opeyevil
cegprakash: that's what most of us do Opeyevil
tobk: Huh, what's the story behind the Gold boss? Never been much into the Lore of Pacman...
Opeyevil: how do i move the cins from the website to my ide? only been coding for 1 year so noob
eulerscheZahl: what's the name?
tobk: jup
Opeyevil: is there an API i can access?
Hjax: Lord Betrayus @eulerscheZahl
eulerscheZahl: betray us, interesting :D
eulerscheZahl: Opeyevil https://www.codingame.com/playgrounds/53705/contest-tools-and-workflow/debugging
cegprakash: Opeyevil use cerr
tobk: haven't seen it play yet, maybe a reference to the tactics. just came back here
eulerscheZahl: for CSB it is probably easier to import a single turn
eulerscheZahl: they usually decide about the names before choosing a bot
cegprakash: Opeyevil just print to cerr and copy paste it to ur input file
Doju: erm...
Doju: anybody else having issues with pac info?
Doju: i don't think i changed a thing and my bot is broken
Doju: could still very well just be me though
cegprakash: #3 in gold time to stare the leaderboard more
cegprakash: 97%
Chkoupinator: can pacmen see in diagonal?
[Csongor]: Does it matter witch language you choose to write your code in? I chose js for this pac-man competition, but I see that all the top ones use C++.
Astrobytes: You can do it cegprakash!
cegprakash: I couldn't I'll park my bot for sometime
cegprakash: hoping ppl will push me
cegprakash: I didn't get enough games against boss
Opeyevil: how would you import a class or .h file into this website?
eulerscheZahl: you write everything in a single file
Astrobytes: You can code everything offline in multiple files then use a script to merge into one file for CG of course
EvModder: watching games where players don't activate speed until the 2nd turn, and I didn't understand why....
until i played one and they ate my packed because their speed lasted 1 turn longer
EvModder: ate my pac*
EvModder: it's some next-level thinking to be sure
cegprakash: it's a bug EvModder
cegprakash: it's same as using speed on first turn
EvModder: oh?
cegprakash: if u see someone doing that help them by sending them the replay that they have a bug
MSmits: ??
EvModder: ah ok
eulerscheZahl: i disagree ceg
Icebox: lol
MSmits: what is this bug you're talking about ceg?
eulerscheZahl: i don't speed in turn 1, if i spawn next to an opponent that could eat me
MSmits: are you smoking your pellets?
Icebox: nice try to hide a strat ceg
cegprakash: well that's a rare case
cegprakash: 1 in 1000 times it can happen
EvModder: i agree eulerscheZahl; always save ability if a scary opponent is next to you
eulerscheZahl: also saw another player optimize for super pellets: no need to speed, when the first is 1 cell away
MSmits: a little bit more often than that cegprakash
EvModder: unless they have speed
GauravGosain: hello guys im new here
EvModder: welcome :)
GauravGosain: thanks a lot :D
Adnmaster: how many lines of code did you write to get out of bronze?
eulerscheZahl: hello new i'm here for longer already
EvModder: i have programmed strats for a lot lot lot of edge cases / rare maps
MSmits: I'm newer than here for longer already
eulerscheZahl: let me check...
silver opened wednesday evening
EvModder: but my general pellet collection is kind of mediocre
MSmits: many doubles I take it?
eulerscheZahl: my bot which promoted to silver had 1250 lines
GauravGosain: just got a problem with a solution of x * (x+y) and my fellow clashers are struggling
Adnmaster: omg
eulerscheZahl: oh wait, that was gold
looked at the wrong day
Adnmaster: whew
EvModder: lots of lines man
eulerscheZahl: 1014
Adnmaster: nooooooooooooooooo
cegprakash: good job kord lets bring down this boss's score
eulerscheZahl: then again: i promoted by being top10 when the league opened
GauravGosain: lvl52 @eulerscheZahl wow
EvModder: mine is like 1000 but it has a lot of stuff in commented out blocks
Adnmaster: what strategy did you use? (dont tell if its forbidden to tell) eulerscheZahl
EvModder: (cool features i tried to implement but made my rating drop when i tried them out lol)
eulerscheZahl: i don't want to go too much into detail. but i'm planning a few turns ahead and tracking my opponent location
Adnmaster: so you are going to chase him?
MSmits: only if he cant escape
EvModder: i chase enemies, if i know i can catch them :)
eulerscheZahl: GauravGosain you saw my rank by hovering over my avatar i suppose. there's also a text "don't invite me to clash of code"
GauravGosain: oh im sorry
King-Of-Code: My gameboard doesnt show up?
Icebox: xd
King-Of-Code: new here
Astrobytes: Is this the first competition where you haven't "gone into too much detail" eulerscheZahl?
eulerscheZahl: yes i chase: https://www.codingame.com/replay/464470922
Icebox: eternal struggle euler :D
MSmits: eulerscheZahl he just saw clash of code. If you look at the text, both sides of that sentence are a in a different color background
Icebox: why isn't there an option to disable invites :D
eulerscheZahl: yes, i'm usually too talky Astrobytes :/
Astrobytes: :)
MSmits: he said plenty
MSmits: my bot is Euler's idea
eulerscheZahl: stop trolling
njaber: Anybody knows if we can get perf improvement from parallel computing, or do they prevent threading entierly?
Astrobytes: lol
MSmits: :)
Adnmaster: lol :D
AntiSquid: lol
AntiSquid: (didn't read chat yet)
EvModder: eulerschezahl that replace is quite cool
EvModder: replay*
EvModder: i still haven't got my pacs to collaborate with eachother to trap enemies
MSmits: me neither
MSmits: you can get legend without such things
MSmits: but it's cool
EvModder: ya
EvModder: it is cool to watch
Hjax: its so hard to understand why my bot wins and loses
EvModder: tell me about it
MSmits: I can usually tell... it's just hard to find a generic fix fo rit
eulerscheZahl: yeah, you don't see what your bot is seeing
perseverent: I still can't believe the silver strength
Astrobytes: Hjax - it wins when you beat the opponent, it loses when the inverse is true! (sorry)
perseverent: what the hack
njaber: I think there's a better way to get the answer to my question, do people in Legend usually use parallel computing?
MSmits: noone does
perseverent: what do one have to do against that boss
MSmits: one core
Hjax: like, i write a feature, it makes my bot worse, dont really know why
Hjax: guess i try to write a different feature
EvModder: threads are nasty anyways
njaber: Ok then I guess I need to find another way to optimize my thing
EvModder: who wants that
Doju: Hjax oh yep
MSmits: happens to me as well Hjax. It's discouraging
eulerscheZahl: been there too Hjax
MSmits: havent gotten a solid improvement since monday
eulerscheZahl: but right now i know what i broke
Doju: i just spent the day writing things that make my pacs not crash
Doju: rank dropped by 200
Hjax: but your monday bot is better than my friday bot MSmits
Hjax: :D
MSmits: I know... that's the silver lining
eulerscheZahl: https://imgur.com/a/6I1Hfvl
some pellets have a probability > 1 to exist :/
MSmits: then the gold lining
MSmits: and now legend :) :)
MSmits: but good to see you got your search working at least Hjax
Hjax: yeah it works, i think its just optimizing for terrible things
Doju: eulerscheZahl great then you only need to send 0,76 pacs there
EvModder: hmm, that pellet looks very likely to exist
cegprakash: makes sense Doju
EvModder: also, nice image
wlesavo: more than likely
MSmits: thats annoying Hjax, your type of search is hard to debug also
EvModder: i use cerr in the IDE to debug my grid lol
njaber: Ok frick I think I need to switch to c++, I'll have to rewrite everything :(
EvModder: why?
njaber: Not enough perf for what I want to do
Astrobytes: You need to use the D
EvModder: ah, that's unfortunate
njaber: I need some precalcs in in turn 1 bue they take 4 times too long with c#
**AntiSquid watches the pendulum swing rank up and down
EvModder: i wrote in c++ because of familiarity, but my code should be fine in python or anything really
njaber: I'm not even sure c++ will give me enough perf
EvModder: best of luck antisquid, don't get too wrapped up watching it lol
pmor: write in machine code
njaber: Actually it's probably more the 4 times too long since I haven't implemented all of the logic yet
AntiSquid: it's hypnotizing EvModder
zarif98sjs: @eulerscheZahl how are you generating this image
Astrobytes: He takes all his output data and manually draws each fram in GIMP
Astrobytes: *frame
eulerscheZahl: directly in my bot code, with pragma not to do it on submit
eulerscheZahl: that's just 30 lines of code to generate them
cegprakash: I am waiting for eulerscheZahl's gif after the contest already
eulerscheZahl: not that interesting. you see some tracking and path clearing
[Csongor]: What is the new thing in gold?
cegprakash: thinking about entering legend without having an opponent tracking I must be too greedy
eulerscheZahl: silver rules are final
[Csongor]: really
eulerscheZahl: yes
eulerscheZahl: was planned that bronze rules are final. but the missing dead players were a "fix" during the contest
cegprakash: even if I'm #2 and there is a big gap from the boss in score I'll get very less matches against the boss?
eulerscheZahl: i think only codingame can answer that
EvModder: i dont know, i had to resubmit my bot for it to be matched against the boss
cegprakash: omg Zylo what did u add
cegprakash: u just promoted few ppl
Zylo: yee... I'm stronger than boss and weaker than a few others :-/
Kukiss: I see you are almost 1 point ahead now :)
Doju: I seriously need to add some kind of combat logic but i have no idea where to put it or how to even do it :/
Doju: currently my pacs only focus on pellets and not screwing the pathing up
llllllllll: thats perfectly fine imo
llllllllll: combat wastes time
Doju: No but they like run straight into enemies
llllllllll: ah ok, thats bad
Doju: if there's a pellet in between them and the enemy
Doju: and don't even consider killing them even if it's possible in one move
zarif98sjs: I still have no idea how to track the enemy
zarif98sjs: looks like without adding that my bot is not improving
HiIamFriend: did anyone tried to trap enemy packmans if u have more score ?
MarekM: yes, collision resolving is essential, but i have still problem when attacked by the corner, no chance to see him
HiIamFriend: to prevent enemy to gain more score
romanshevchuk: can someone please explain how this rule works?
"Only pacs with the SPEED ability enabled can move:".
I mean is it random?
VicHofs: No
VicHofs: SPEED grants an extra move round for those who have it on
VicHofs: all other pacs who don't can't move
romanshevchuk: I though it's an ability that I can enable from time to time. I don't see anything in the rules about stopping other pacs from moving
VicHofs: It is though
VicHofs: But that's how it works
VicHofs: if any of the pacs has speed on, there will be intermittent rounds when only pacs with speed enabled can move
Valiant_Wolf: to rephrase: each round has two movement phases; only pacs with the speed ability enabled can move in the second phase
Kellthazar: Wow... I think that Ill be promoted
VicHofs: sick @Kellthazar
Kellthazar: Thanks... And the @ is not necessary to mention an user VicHofs ;)
VicHofs: I am well aware Kellthazar ;) ;)
HiIamFriend: how to know if you will get blocked or no
HiIamFriend: you always has to check around pacs for that ?
VicHofs: wym
HiIamFriend: have*
romanshevchuk: @VicHofs, @Valiant_Wolf, thanks a lot for explanation, I can see now how it maps to what I see in game summary
VicHofs: what league are you in? @HiIamFriend
EvModder: you dont always know you will be blocked until it happens
HiIamFriend: wood 1
VicHofs: I would say not to worry about that just yet
HiIamFriend: coliding into each over
HiIamFriend: and get stuck
EvModder: you can keep track of your position last turn, and if you are in the same spot next turn you can deduce you were blocked
HiIamFriend: ow
EvModder: and try doing a different move next turn if you want
VicHofs: try to improve your pellet algo first HilamFriend
Kellthazar: HilamFriend, I check if the bot is stucked by comparing the current and last position
HiIamFriend: i am stupid wow
HiIamFriend: ... :D
VicHofs: nah famm
HiIamFriend: ty
VicHofs: focusing on my pellet algo is what got me into bronze tho
HiIamFriend: from zero to hero
HiIamFriend: lets go
HiIamFriend: :D
VicHofs: I didn't even have a collision avoidance system just yet
VicHofs: np
EvModder: is it possible to go down a league if i do badly enough? :)
VicHofs: no
HiIamFriend: rank 39/669 but randomly i get stuck by no solving coliding
HiIamFriend: so i cant get promoted
HiIamFriend: :D
Doju: Ok i give myself an hour to make some super basic combat logic and then imma go to bed
VicHofs: huh
VicHofs: @HilamFriend well if you feel like your pellet algo's good enough already then go right ahead
Kellthazar: If is the case, I try to find another floor to go which not include the "stuck Point" on the path.
HiIamFriend: will try
Kellthazar: Gold League :)
HiIamFriend: cool
VicHofs: damn bro how lmao
muy31: come on... almost there...
kovi: nice robo
RoboStac: and now I fall back down again :)
Akarachudra: but dat was impressive
kovi: hungarian pattern
kovi: not any more
kovi: 5 hungarian now
RoboStac: ah, do I need to move to hungary to be able to stay at the top?
Astrobytes: lol, told ya it was a Hungarian takeover! Gratulálunk!
llllllllll: elég király :^)
kovi: van meg hely :)
VicHofs: yes
cegprakash: mataranga leading boss by 2 points
cegprakash: u just rekt him like a piece of cake
Astrobytes: Other than egészségedre that's about the extent of my Hungarian language skills. Though I used to know the words to Hang és fény by Kispál és a Borz, and Csillag vagy fecske, and Most múlik pontosan by Csík Zenekar (and Quimby)
Doju: Szia, Doju vagyok
Doju: that's about all i know
Astrobytes: btw cegprakash, thanks for your music playlist :smile:
Doju: apart from the obvious swearwords :D
cegprakash: wow somebody listens to my playlist
cegprakash: :D
VicHofs: music playlist?
Astrobytes: I only did about half, but some OK stuff on there
VicHofs: why on Earth do yall know Hungarian
HiIamFriend: this pacman movement makes no sense to me :D
Doju: VicHofs i have a hungarian friend
VicHofs: oh I see Doju
cegprakash: Indian songs playlist VicHofs
Astrobytes: I like music from all countries, and my Hungarian friend took me to see Kispál és a Borz, Csík Zenekar, Quimby and Kiss Tibor
cegprakash: wait kord got pushed up too :O
Astrobytes: And I love Indian melodies and rhythms
cegprakash: 4 legends on the way
kovi: vadasz may end up top legend
NinjaDoggy: already 50 legends day 1 :O
NinjaDoggy: wait... vadasz's submit might end #1 legend
NinjaDoggy: how much was he above gold boss ._.
Peudoki: Hungarians dominate
Vadasz: Its funny, i hardly beated the gold boss with the same bot. Quite large random factor.
HiIamFriend: how to get pacman current possition
Doju: HilamFriend it's given in the input
NinjaDoggy: "i hardly beated the gold boss"
NinjaDoggy: whelp
Astrobytes: if 'mine' is == 1 or true it's your pacman and you can use the x and y HiIamFriend
HiIamFriend: inputs[0] and inputs[1] gives the possition he is moving to , the end of the road :D
Vadasz: I was better with only 1 score from the gold boss
SurajMishra2209: 22 hours spent
HiIamFriend: my frontender skills does not help me in any way :DDD
SurajMishra2209: and I still haven't finished bronze
HiIamFriend: i feel like a stupid potato
King-Of-Code: out of wood after 30 minutes now the fun begins
King-Of-Code: gonna have to upgrade the manhattan distance alg
kovi: still very nice submit vadasz
kovi: legend may have enough players now to catapult submit. and with 50% rate can stay in top
HiIamFriend: can someone explain this "MOVE ${x} ${parseInt(inputs[0])} ${parseInt(inputs[1])}" how can he move to left and right with this :DD i am playing around btw
HiIamFriend: does move mean "go 3 forward and like 1 right"
NapTown: it's a destination
Kellthazar: No
HiIamFriend: so its a destination like on a map ?
HiIamFriend: he tries to move there ?
NapTown: yeah
HiIamFriend: ow
HiIamFriend: gotcha
Kellthazar: x,y = coordinates
HiIamFriend: i understand that
Kellthazar: each coordinate has a "optimal" path to others coordinates
Doju: No they dont
Doju: but they do have the shortest
Doju: optimal depends on what you value
muy31: someone kill me now, im back at 17th place
Kellthazar: yeah, Doju is right!
Doju: okay muy31 i'm currently making killing logic
muy31: thank you
Doju: i'll let you know when i'm done
Doju: i'm trying to figure out how long travelling a distance will take but can't figure it out with my tired brain :/
Doju: like that's off-by-one-error land right there
muy31: once i get to gold im going to use trapping logic
muy31: and re-implement my enemy tracking which ive disabled for now
muy31: i really dont want to resubmit again, but ive been stuck here for 3 days now... what should i do?
muy31: if i win the next three matches against clyde, im resubmitting
HiIamFriend: :DDD cool
elderlybeginner: how to turn off editor's hints and suggestions?
muy31: What I find funny? I get 3/4 superpellets and still lose, AGAINST MYSELF! https://www.codingame.com/replay/464866992
muy31: at this point im just randomly testing different conditions to see how they work in a scenario
muy31: Same scenario, different condition, same outcome... but interesting shape on frame 75/76https://www.codingame.com/replay/464870525
VicHofs: yeah I love minecraft
VicHofs: diamond pickaxe looks dope
HiIamFriend: muy31 why keep bumping into each other ?
Scarfield: its the best way to make minecraft recipe pics :p
muy31: HilamFriend all of them trying to take shortest path to nearest unexplored location
muy31: but actually there are not bumping, only two are
muy31: b/c i disabled killing
Doju: oh, this was surprisingly easy
Doju: tbh it took me an hour to write like 30 lines but anyways
muy31: oh im rank 2 hallelujah!
VicHofs: damn nice
NapTown: it do be like that Doju
VicHofs: every day Doju
muy31: nvm rank 3 now
muy31: how you like that, second place says
Clagus: damn... I've got 70% win ratio and still in silver TT
muy31: ill play you clagus
Clagus: /o\
Clagus: in my IDE I won :P
muy31: oh
muy31: https://www.codingame.com/replay/464881345
Clagus: LOL
muy31: Clagus spam matches
Clagus: what?
muy31: links right here you against me
Clagus: why? :P
muy31: idk im bored
Clagus: hehe
muy31: https://www.codingame.com/replay/464882923
muy31: im 4th place
Clagus: I need to get out of silver :/
muy31: me too
Clagus: i'm 6th
VicHofs: I need outta bronze grrr
muy31: https://www.codingame.com/replay/464883370
VicHofs: 293 still :(
Clagus: no no, don't spam
muy31: ok
Clagus: that other game was interesting though
Clagus: you lost 2 pacs and still won by 1 point :D
Clagus: crazy
Clagus: http://cgstats.magusgeek.com/app/spring-challenge-2020/muy31
Clagus: you've got only 62% win ratio
Clagus: DAMN
Clagus: :P
muy31: i know
Clagus: right now I'm 71%
Clagus: http://cgstats.magusgeek.com/app/spring-challenge-2020/Clagus
Doju: Error counter: 8
Clagus: what?
Doju: that's surprisingly good for an hour's worth of code
Doju: oop, 9
Clagus: what error counter?
Doju: counting errors in my code :P
Scarfield: not fixing them? :)
muy31: lol
Doju: Scarfield maybe
Scarfield: nice, but just joking :)
Doju: 10
Clagus: this is crazy I've got 71% win ratio and still not promotion
SurajMishra2209: my name doesn't appear in cgstats
SurajMishra2209: can somebody help me out on that?
VicHofs: You gotta be top 1000
VicHofs: I think
SurajMishra2209: the last rank I got was in 200
VicHofs: overall
muy31: total
VicHofs: not in your league
SurajMishra2209: okay
SurajMishra2209: got it
muy31: you are 1699
Doju: lets see if this modification did anything
VicHofs: 69 lmao
SurajMishra2209: lol
muy31: 1710 now
muy31: 1988
muy31: 1698
muy31: my bad
SurajMishra2209: yeah another version of code
SurajMishra2209: :D
muy31: 1645
muy31: ranks up
muy31: nice job
SurajMishra2209: thanks
SurajMishra2209: still not good enough
NapTown: my flee enemy code makes me stand still while they consume me:expressionless:
NinjaDoggy: better than runing into them :p
VicHofs: looooove falling 1000 ranks when I submit 😒
VicHofs: scratch that
VicHofs: make it 1500
NapTown: slightly NinjaDoggy
Doju: oh yes i think it definitely did something
Doju: winning like 80%
muy31: you submitted it yet doju?
Doju: yup, computing
muy31: Killing achieved... https://www.codingame.com/replay/464891163
RockyMullet: 44th at 13% :eyes:
RockyMullet: my change seems to help
RockyMullet: theres hope
NapTown: nice
MattSOrme: most of the game is pretty standard, couple bugs im still working on. But look at how it ends
https://www.codingame.com/replay/464889801
RockyMullet: dont know if its enough to beat gold boss, but definitly an improvement
Doju: muy31 it's not actively trying to kill
Doju: you just happened to be in the way :P
muy31: oh
RockyMullet: the munching
NapTown: monch monch
muy31: i stopped trying to kill though
muy31: rank 5
muy31: im never going to progress
MattSOrme: my bots aren't killers, so getting a win by a double kill is pretty cool
AntiSquid: RockyMullet imo too random leaderboard to tell
cegprakash: does anyone else get random timeouts?
cegprakash: it started only today for me
MarekM: no
AntiSquid: either you have a MASSIVE improvement and notice the difference or you get fooled by the randomness
MarekM: https://www.codingame.com/replay/464892869 close enough
cegprakash: exactly AntiSquid
Doju: AntiSquid that's very true
cegprakash: I was #3 above kovi
cegprakash: I should have parked my bot there
cegprakash: now kovi reached legend
cegprakash: and I'm stuck at #200
RockyMullet: oh well AntiSquid I feel, deep down, in my hearth, that its an improvement :P
AntiSquid: ya cegprakash very very bad mistakes you made there
King-Of-Code: the fact i can land on a pac and they can change type before my eat registers is quite annoying
muy31: RIP cegprakash
AntiSquid: you might have missed legend because of that cegprakash
muy31: that's porb for the best King-Of-Code
muy31: *prob
AntiSquid: cegprakash are you resubmitting your initial rank 3 bot ?
Doju: oh this was a definite improvement
cegprakash: yes with bugfixes
AntiSquid: no, submit without the bugfixes
cegprakash: infact this is my 2 day old code
Doju: 150 ranks up muy31
cegprakash: whatever I did in the middle I threw out
muy31: sweet Doju
AntiSquid: sometimes "bugfixes" drop your rank, it's really stupid leaderboard system
AntiSquid: just get the exact saem code and resub
cegprakash: I'll have to do it 100 times luckier because those who gave me points got promoted now
cegprakash: I had 1 chance
AntiSquid: at the very least submit the rank3 code and leave it overnight, you never know
AntiSquid: ya sux to miss legend like that
cegprakash: AntiSquid do u have the time when I was #3? I lost that version
cegprakash: chat got disconnected in the middle
cegprakash: and all messages erased
AntiSquid: uhm try 3 hours ago ?
NinjaDoggy: everyone in gold before you submit make you have >50% winrate against boss please :)
cegprakash: found it
cegprakash: saving backup
NinjaDoggy: i would like to get to legend without having to be rank 1 :/
AntiSquid: ninja i beat boss once on my current sub, don't want any complaints @_@
HiIamFriend: my mind is melting
AntiSquid: that's 100% win rate :p
muy31: i just got inspiration, what if I...
muy31: nope
Doju: I think the fancy pathfinding thing i wrote today goes out the window if the pacs are "panicked"
Doju: oh well
Doju: imma go, cya everybody
muy31: lol
muy31: gn
AntiSquid: ceg i think you fked up big time, looks like you would have been promoted by now
HiIamFriend: how come my pacmans moves and stops moves and stops
Gonny: ceg don't give up though, there's probably gonna be another 50 promotions before then end of the contest
Kellthazar: 2 situations: your pacs arent over speed buff and there is another pac with this buff active...
Kellthazar: HilamFriend
HiIamFriend: wood
HiIamFriend: does not use speed buff ?
Colios13: yep no speed in wood
Kellthazar: There is speed on "Wood" league?
Kellthazar: oh right
muy31: no
Kellthazar: nvm then
HiIamFriend: i am trying to save old possitions of pacman if he gets stuck
HiIamFriend: but somehow array is not update'ing fast enouth i guess
HiIamFriend: and every second move he bugs out
HiIamFriend: somehow
HiIamFriend: because array is not updated
HiIamFriend: and brings same numbers of x and y
HiIamFriend: my mind is melting
HiIamFriend: xd
MarekM: HIlamFriend, this is because activated speed
HiIamFriend: wood league doest not have speed
muy31: am i actually ever going to pass up silver bot?
Kellthazar: Maybe your array isn't updated because you create a new one on each loop iteration
HiIamFriend: nope
HiIamFriend: its global
HiIamFriend: before while loop
Colios13: GL muy
Kellthazar: In my algorithm I have a pacman object that has a history property
Kellthazar: Which is a list
Kellthazar: and I add the currentPos of the pacman on each iteration
HiIamFriend: const tempPacsPossitions = [];
tempPacsPossitions.splice(int, 1, {x, y})
HiIamFriend: i replace
HiIamFriend: last
HiIamFriend: but same i think
Kellthazar: Why not just add?
HiIamFriend: why not just replace old one with a new one ?
HiIamFriend: :D
Kellthazar: and get the last one for stuck verification?
Kellthazar: I use this history array to weight my paths
HiIamFriend: how will i able to track every pacman that way
Kellthazar: Object orientation?
PiterYeh: i want to share my masterpiece AI: https://www.codingame.com/replay/464889354
PiterYeh: starting at frame 16 you can see my spirit animal
HiIamFriend: i will try to figure this out myself xd
HiIamFriend: :DD
HiIamFriend: lovely
muy31: impressive PiterYeh
HiIamFriend: spirit animal "Stop and give up" :D
HiIamFriend: thats me in 1 hour from now
PiterYeh: that's me since 7 days ago
muy31: me since two days ago
muy31: YO! Silver bot timed out!! https://www.codingame.com/replay/464918233
moojek: lmao
NinjaDoggy: yes... bosses are just bots like yours or mine
moojek: still
NinjaDoggy: they literally select one from the ladder so...
muy31: yah but that proves its not just doing a lazy eval to win
muy31: its doing some hard computations
Kukiss: I think it might be cause by any error. In replays you always see "Time out" even if the user get normal error
muy31: maybe?
Kukiss: do sth stupid in code and see that probably you will have a Timed Out tag at the end of the time line
Kukiss: your code crashes to the output is not provided in time - that even makes sense
Kukiss: so the output is not provided*
butcer0: How do teams collaborate for competition in CodinGame? Is there any way of constructing a more formal team or is a single user supposed to represent a team?
muy31: at least three members of your company has to submit code
muy31: or school
muy31: What is this? https://www.codingame.com/replay/464930103
Kukiss: you danced at the end? :D
muy31: i think i finally got a working formula though
Kukiss: I understand you
Kukiss: https://www.codingame.com/replay/464929952
Kukiss: I messed sth up
muy31: yeah that doesnt look good at all
Drastic_Loontastic: How do you change the speed?
baDcoder: Hello, do player downgrade to any lower league ?
Drastic_Loontastic: Or edit the speed, I'm confused
Flyingkid: no your safe bad coder! you only move up in leagues.
Flyingkid: :grimacing:
Kukiss: Drastic you can use two "special" actions: SPEED and SWITCH once you get to Bronze
baDcoder: Thanks Flying Kid , sigh ! :)
butcer0: What is the definition of a "Company" for competition?
Flyingkid: good thing too cause you think everything is good then submit and then nothing but timeouts some times.
Drastic_Loontastic: Thank you so much Kukiss, I couldn't find where it said that
Kukiss: np. there are more details about this topic like cooldowns etc.
You are able to read about it in the description once you get to the proper league
Drastic_Loontastic: Oh silly me, I didn't read it XD
SurajMishra2209: https://www.codingame.com/share-replay/464942807
SurajMishra2209: I guess Inky is fed up of me
HiIamFriend: i wana die
HiIamFriend: burn me
muy31: Clagus what? how?
muy31: im in a dilemma: the anti-Clyde code beats Clyde 80% of the time, but cant beat everyone else, the everyone else code beats everyone else but cant beat Clyde
NormantasStankevicius: I know the feels muy, last challange I could beat everyone in silver besides the boss... tbh I would focus on beating everyone, more score
NormantasStankevicius: but that is me and I don't know this system well
cegprakash: I'm starting to wonder
cegprakash: may the bug was a feature
cegprakash: it it's own weird way
HiIamFriend: and i am just trying to move to random locations :DD
Kayzerg: Yo guys
VicHofs: stfu
Kayzerg: Fam
Kayzerg: Dis ting doesnt support bullying
VicHofs: the ting go skrra
Kayzerg: Imma find you bro(not a threat)
VicHofs: nice
Kayzerg: mate
Kayzerg: i know the shape of your house
Kayzerg: youre sucked
VicHofs: :(
Kayzerg: :muscle_tone5:
Kayzerg: yur getin fisted m8
Cappefra: muy31 I feel you
Kayzerg: 1v1 me on coding game vichofs
VicHofs: ok
Cappefra: I'm also moving around the first 50 positons in silver and I'm just not playing enough against the boss to be able to pass to gold
VicHofs: You're fucked Kayzerg
HiIamFriend: go to private chat
HiIamFriend: submiting code so i can randomly win to bronze xd
VicHofs: Is there a Dan here?
VicHofs: Kayzerg's looking for you
Cappefra: come on gods of random! Work in my favor!
cegprakash: lol
thethiny: guys when did the competition start
seronis: bit over a week ago i think
thethiny: yes I want the date if anyone knows it
Cappefra: it was thursday last week
thethiny: thanks
HiIamFriend: why cant i beat wood 1 :D
HiIamFriend: i never get stuck
HiIamFriend: still cant get promoted
HiIamFriend: so sad
ZarthaxX: never get stuck?
HiIamFriend: yea
ZarthaxX: wdym
HiIamFriend: probably because i move to random location until i get stuck
HiIamFriend: dunno
ZarthaxX: ok..
HiIamFriend: can you check if cordinate is a wall or no ?
seronis: if you saved that info at the top of the script, then yes
seronis: i build a board object with all the cell contents before the while() loop starts
thethiny: are we allowed to do multithreading here? and if we are, will it benefit?
seronis: i doubt it
thethiny: if it's a single core then multithreading won't matter
HiIamFriend: i mean where is "readline()"
thethiny: @HiIamFriend what's your question
HiIamFriend: but can you somehow get like "readline(x, y)" ?
thethiny: c++?
HiIamFriend: like if (x,y) is a wall
HiIamFriend: js
seronis: readline is a user input. it has nothing to do with xy
thethiny: exactly
thethiny: what u do is
thethiny: let var something something = let var = readline();
thethiny: then var[x] == -1
thethiny: or '#'
HiIamFriend: hmm
HiIamFriend: will try
seronis: make your own object to track the board
seronis: i do https://i.imgur.com/SkF8ClD.png
seronis: so for any given cell i can always check if its been visited, if its a wall/open, what its last known pellet value was
HiIamFriend: okey thanks ! :)
HiIamFriend: http://chat.codingame.com/pastebin/ac7d0352-80e0-4f6c-baca-7e5990f74326
HiIamFriend: how come i cant use substring on this
seronis: what does your 'row' look like ?
seronis: is it a string or an array / object ?
HiIamFriend: http://chat.codingame.com/pastebin/32e0eea9-6703-476b-a03e-c97576ba1958
HiIamFriend: first print does print the character
HiIamFriend: and in for loop i get that error
HiIamFriend: why tf :D
HiIamFriend: ow
HiIamFriend: i use array in a very bad way
HiIamFriend: sorry
seronis: yeah you're using map wrong, but i read it a few times and cant see a reason 'console.error(mapString.substring(0, 1));' wouldnt work
DollarAkshay: Found a nice seed with a lot of action in the middle
DollarAkshay: seed=2904725058612385800
HiIamFriend: 'console.error(mapString.substring(0, 1));
HiIamFriend: that one works
HiIamFriend: console.error(map[i][y]);
HiIamFriend: map[i][y].push(mapString.substring(char, char+1)) ;
HiIamFriend: this one is not working :D
HiIamFriend: its ok i will fix it
BenjaminUrquhart: thethiny I see your follow
DollarAkshay: Guys send me some of your bot replays in this seed seed=2904725058612385800
BenjaminUrquhart: https://www.codingame.com/replay/465026268
BenjaminUrquhart: now if you excuse me, I have things to fix
DollarAkshay: Nice one @BenjaminUrquhart YOu do the switch sometimes. But not always
BenjaminUrquhart: well yes
BenjaminUrquhart: but actually
BenjaminUrquhart: yes
DollarAkshay: :joy:
DollarAkshay: *feeling lucky*
thethiny: @BenjaminUruqhart yeah if the competition is over I don't wanna lose track of some people
bouajajais: do the rules change in silver and more?
DollarAkshay: nope
bouajajais: so it's only ranking afterwards
bouajajais: thanks
Icebox: yes it does
bouajajais: oh
thethiny: silver has Dead
Icebox: starting from silver you can see dead pacs
thethiny: Gold has more Dead
DollarAkshay: Its not really a new rule
bouajajais: what are deads?
BenjaminUrquhart: deads
DollarAkshay: Its some dummy info
bouajajais: oh ghosts
DollarAkshay: Thye give the info of dead pacs
bouajajais: okay
DollarAkshay: During turn input
DollarAkshay: But everything else is the same
bouajajais: thanks good luck
DollarAkshay: No changes
HiIamFriend: i cant beat wood 1 by moving randomly
HiIamFriend: :D
Icebox: DollarAkshay dummy info that allows a brand new substrategy and completely breaks your code if you don't adapt for new inputs
HiIamFriend: i had hopes
Icebox: "no changes"
DollarAkshay: @Icebox Nice banner Image
DollarAkshay: Where did you find it
Icebox: I'd be amused if you hadn't asked me this in 2018 and then in 2019
Icebox: you can put the checkmark down for 2020 as well now
Alp3r: sa ipneler
DollarAkshay: I made that banner btw
Icebox: yes you told me that multiple times already :D
thethiny: lol
DollarAkshay: XD
DollarAkshay: Sorry have forgotten
thethiny: that moment when your code runs offline, but returns -1 in the multiplayer :)
DollarAkshay: -1 ? XD
thethiny: yes -1 means not found
DollarAkshay: What in the world is printign that -1
DollarAkshay: Oh
thethiny: trying to find the best path
thethiny: my adjacency list is wrong
thethiny: damn
thethiny: found the culprit lol
thethiny: woohooo
thethiny: 106 - 51 vs Inky
DollarAkshay: Nice
DollarAkshay: You might even skipp silver
thethiny: I don't have collision avoidance
thethiny: and I don't have attack modes
thethiny: so not yet
thethiny: leftover debug code hurts
thethiny: "if bla bla bla || True"
DollarAkshay: I alwasy add a comment called DEV_CODE
DollarAkshay: So that i know if there is any code that should not be submitted
DollarAkshay: I just do a Ctrl+F and find that comment
BiriyaniCrypt02: Love This Website
thethiny: 😂😂
EvModder: i used a lot of cheap tricks to get to my current spot on the leaderboard.
Like standing still and then turning into the thing that eats them when the move to eat me etc
EvModder: i will need to rewrite most of my code to actually be able to challenge any top players
mohammadhei05: shut
