Chat:World/2021-01-26

From CG community
Jump to navigation Jump to search

pipebits: :v

Default avatar.png Surina_Quinn: does anyone have advice for a beginner? I'm trying my best to learn but it all just gets jumbled in my head and it makes me feel...low

Goulstem: @Surina_Quinn I suggest you pick a project that interests you! This is the best way to learn, by far. Then you have an application for all your knowledge. Also, key: the more something pulls you (inspires you/motivates you), the less you have to push yourself! :D

Passifi: Salut

jacek: ohai

Passifi: woo codin game runs badly on the rasberri pi Im having crysis flashbacks here :P

jacek: robo :eyes:

Default avatar.png MSmits: what'd he do now?

jacek: othello. but were safe

Default avatar.png MSmits: ah

Default avatar.png DaniloKrkeljic: hey, idk if this is the place to ask, but im getting correct results in the defibrillator game on everything but "Complete File", is this a bug ?

jacek: with your program :o

Default avatar.png DaniloKrkeljic: im gettting correct results on "Complete file 2" though

Default avatar.png MSmits: the reason you have different testcases is to test different parts of your program. It is completely natural to fail one and succeed the other

Default avatar.png MSmits: find out how the testcases are different

Default avatar.png MSmits: use debug statements to split up your program and see what it's doing

Default avatar.png DaniloKrkeljic: Ok, ty very much, appreciate it. Ill look into it.

Default avatar.png MSmits: np, good luck

Westicles: wow, such a highly literate contrib approved. things are looking up

gatitos: is it me or does the site UI keep changing

gatitos: sometimes "complete" is under "Activities"

gatitos: sometimes it isnt

jacek: theyre changing header and some people get it, some not

Default avatar.png Andrew97: has anyone finished last weeks challenge? Battle Tower

Default avatar.png Angecide: Hi Jacek when doing 1ply check to add bias toward moves that win miniboard/block enemy win, do u also take into consideration if the move would give enemy a win on the next board?

Default avatar.png Angecide: or technically speaking, that would be a 2ply check

Default avatar.png Angecide: so I guess not

jacek: no, too costly for meh

jacek: but i penalty moves that would end up on finished boards, giving enemy mobility

Default avatar.png Angecide: ah yeah that's a fairly lightweight check so that could work too

jacek: with that 2-ply check, actually with some generated tables that could be cheap

_Time_Lord_: bus

Default avatar.png Angecide: yea but the 2ply check can get complicated because you can't keep avoiding capturing a board in favor of not giving enemy one, so at some point u need to kind of evaluate when the tradeoff is in your favor, I guess one easy tradeoff that is in ur favor is getting the middle miniboard in favor of any of the side ones

MSmits: the problem with uttt is that there are exceptions to every rule Angecide

MSmits: middle is not always better than other boards

MSmits: what jacek means is that he gives a penalty in selecting a move that gives the opponent this freedom, but he does still select the move

MSmits: it simple has a penalty in UCT

MSmits: simply

jacek: eeyup

MSmits: alternatively, in the random rollout, you can try to disallow them altogether, under some circumstances

MSmits: just dont be too strict about refusing these moves, because sometimes, especially in the late game, you have no choice

MSmits: or only poor alternatives

jacek: i do this only in uct

Default avatar.png Angecide: yea that's true, I guess the bias should just work as guidlines and not some authoritarian rule, and the uct will hopefully figure out if it is worth it or not

MSmits: exactly

MSmits: though there is no problem with an "authoritarian rule" as long as it is limited properly

MSmits: for example. you can disallow these moves in the first 20-25 plies or so and it will be fine

MSmits: because even if there exists a finished board, it's probably never gonna be a good idea to give an opponent a free move

Default avatar.png Angecide: yea, like enforcing teccles heuristic

MSmits: right, that heuristic is one of the most reliable ones

MSmits: but it's not perfect

MSmits: I have seen gamestates, especially ply 20+ where using it, hurts you

MSmits: turning a solved win in a draw, or a solved draw into a loss

MSmits: possibly using it up to ply 18 or so is 100% safe... but it is hard to say without solving the game

jacek: yet

MSmits: right :)

MSmits: just improved my code of the rings a bit with pruning, rank 20 now. Small steps. I need to figure out a way to do the loop stuff

MSmits: so far still doing the loops manually

jrke: whats the difference between b += 1<<3; and b |= 1<<3;

RoboStac: it depends what b is before hand. If bit 3 is 0 then they're the same.

RoboStac: if bit 3 is one the second one does nothing but the first still adds 8

jrke: oh

jrke: btw whats the opposite of b|=1<<2;

jrke: like i wanna make bit 2 turn 0 from 1

RoboStac: b &= ~(1<<2)

RoboStac: thats a tilde (bitwise not), not a minus sign btw

jrke: yeah its ~

MSmits: alternatively, if you know for sure the bit is set you can do b ^= 1<< 2

MSmits: but if you dont know and just want to make sure it is 0, you use Robos method, but thats 1 extra operation

jrke: ok

jrke: so it makes 0 to 1 and 1 to 0 right

MSmits: thats what ^ does yes

MSmits: ^1 actually

RoboStac: it looks like its an extra operation in code but most cpu's have bit set / reset that the compiler can use

MSmits: oh, did not know that

jrke: so b ^= 1<<2; is perfect for othello i think

MSmits: thats another thing jrke, everyone gets to a point where they start to look for little things that would make their code faster, but more often than not, the compiler will recognize and do it by itself. So that's mostly wasted effort

MSmits: Robo just shared an example of that

jrke: hmm

MSmits: but if you like ^=, that's fine

MSmits: just be careful about it in case the bit was already 0

MSmits: bugs in bit ops are really hard to notice

jrke: thank you very much my net is slow so going gtg bye cya

MSmits: cya jrke

jacek: well it looks simpler

MSmits: yeah... readability vs safe coding

MSmits: &= ~ is often safer than ^=

MSmits: I prefer ^= because I've done this so often now , i dont often make the mistake of flipping a 0 bit

MSmits: jrke was asking earlier about the problem with 1 << 32 instead of 1ULL << 32. That's an annoying bug that still happens to me

MSmits: you dont notice it but it leads to bad crashes

MSmits: probably i should just create a set bit function and stop typing it literally :)

Default avatar.png Angecide: noo msmits why did u remove the othello bot

jacek: huh

Default avatar.png Angecide: or is it the same one as yesterday?

Default avatar.png Angecide: or maybe i am misunderstanding

jacek: hes 5th

Default avatar.png Angecide: i read in yesterday log there were some new othello bot that was perfect or something, I wanted to check it out

jacek: maybe the 1st one?

Default avatar.png Angecide: nah I think it was supposed to be better than recurse, I am not sure, maybe I misunderstood

Default avatar.png DaniloKrkeljic: There is a bug in the defibrillator game, "Complete file" answer is wrong. Are there similar bugs on this website?

Default avatar.png DaniloKrkeljic: i just dont wanna waste time if its possible the answer is wrong or similar

blasterpoard: DaniloKrkeljic plenty of people, me included, solved that puzzle without any problems

blasterpoard: it is likely that the mistake is in your code, not in the puzzle

Default avatar.png DaniloKrkeljic: i found a forum question with people with the same problem

Default avatar.png DaniloKrkeljic: distance of calculated defib is smaller than the one form the answer, simple as that

blasterpoard: Then how do you explain the 84% community success rate on that puzzle?

Default avatar.png DaniloKrkeljic: because when you submit you apparently get 100%

blasterpoard: I have no problems passing that validator though

blasterpoard: *test case

Default avatar.png DaniloKrkeljic: idk, these guys on the forum said the same thing.. well at least its good to know its not common apparently

blasterpoard: maybe if you got 100% desptie that, the problem is that there are not enough validators, not that the test case is wrong

VizGhar: well he obviously isn't using right formula :)

Default avatar.png DaniloKrkeljic: how come it works for "Complete file 2" then?

VizGhar: I believe there were some rounding issues

VizGhar: I also failed on one submission and had to fix it

VizGhar: Problem #1 you might have is that formula has `x 6371` in it Problem #2 using floats instead of doubles (loosing precision)

VizGhar: not sure where the problem was when I was solving it... but you should definitely not blame CG

VizGhar: Potential problem #3 decimal separators

Default avatar.png DaniloKrkeljic: I'm using python float, so it shouldn't be a rounding problem. I'll look into it more i guess. Btw, the data from the website differs from test cases, maybe I'm supposed to build a solution according to that?

VizGhar: you are supposed to come with general solution that matches any input it is given :)

VizGhar: https://forum.codingame.com/t/defribillators-puzzle-discussion/37/177?u=vizghar

VizGhar: this guy had probably same problem as you and resolved it too

VizGhar: "Ok, got it. Didn’t noticed that the user’s longitude and latitude include commas, too."

VizGhar: do not fall into confirmation bias and look for solutions :)

Default avatar.png DaniloKrkeljic: No, i did that 1st thing. Anyway thx for trying to help, not gonna clog IRC beacuse of this

Default avatar.png DaniloKrkeljic: ill post a forum question i guess

VizGhar: potential problem #4 radians/degrees

VizGhar: yeah go ahead

VizGhar: potential problem #5 too long computation

VizGhar: but these are the question you can freely ask in IRC as well. Most of the time you just need slight push

Default avatar.png DaniloKrkeljic: Ok, ty very much Viz !

Astrobytes: Yeah, it's here for a reason. If you have a valid question there are usually people willing to help out ;)

Default avatar.png IlkerSaxion: How do I start a round and invite a friend?

Default avatar.png IlkerSaxion: Im very new

struct: start a private clash

struct: then invite him

struct: you can send him the link I think

Default avatar.png IlkerSaxion: Thank you

Default avatar.png Saad-py: HI

eulerscheZahl: hi reCurse I have a contribution for you. but no mirrored matches :( https://www.codingame.com/ide/demo/870061fb387227ad77e23f33131ca9e40463b7

struct: Time to resubmit an improved CSB :D

eulerscheZahl: now multiplayer 2048 suddenly seems possible

eulerscheZahl: you shift on your own board and spawn for the opponent

eulerscheZahl: i'm afk

Default avatar.png wlesavo: or you shift for both, you and your opponent while still having fixed number generator available eulerscheZahl

Pouria2100: elo

Default avatar.png wlesavo: you may even shift for you and your opp always in the same direction i.e D D R R

jacek: huh

Default avatar.png ErrorRazor: It's not multiplayer when one player plays for both

jacek: thats what she said :disappointed:

Default avatar.png ErrorRazor: lol whut

jacek: dont mind me, im being jacek again

Default avatar.png ErrorRazor: Why is this a conflicting declaration? using clock_t = std::chrono::high_resolution_clock;

jacek: clock_t is reserved type

Default avatar.png Saad-py: HI

Default avatar.png Saad-py: which language to learn after python

Default avatar.png Saad-py: for competetive programming

JLukeSkywalker: for clash or bots?

Default avatar.png Char_2C: both?

Default avatar.png Saad-py: yeah

JLukeSkywalker: for clash I would say ruby, for bot programming I would say c++

Default avatar.png Saad-py: both

Default avatar.png Saad-py: what do you mean by bot

Default avatar.png Saad-py: bot CODe (lol)

JLukeSkywalker: multiplayer ai games like the othello or tron

Default avatar.png Saad-py: wtf

Default avatar.png Saad-py: There are games for programming

JLukeSkywalker: you have a turn time limit, so the languages that run faster let you do more calculations

Default avatar.png Saad-py: wtf

Default avatar.png Saad-py: I can't understand

JLukeSkywalker: for clash, its about fast coding, so python is great, and for shortest code, ruby is really good

JLukeSkywalker: what dont you understand?

Default avatar.png Saad-py: NOOOOO

Default avatar.png Saad-py: the bot part

JLukeSkywalker: bot programming = there is a game, you write an AI to play the game

Default avatar.png Saad-py: ok give me an example

JLukeSkywalker: are you trolling?

Default avatar.png DaniloKrkeljic: counter strike

JLukeSkywalker: its like half of this website

Default avatar.png Saad-py: noo

Default avatar.png Saad-py: OH so you mean controlling games

Default avatar.png Saad-py: like I make an AI to defuse the bomb

JLukeSkywalker: tron, oware, coders strike back, othello, chess, checkers, ultimate tic tac toe, etc

JLukeSkywalker: https://www.codingame.com/multiplayer

Default avatar.png Saad-py: OHHHHHHHHHHHHHHH

Default avatar.png Saad-py: So in this websites we have to code for the game to work and run, YEAH I know them.

But I was talking about overall programming not just codingame

Default avatar.png Saad-py: lmao I am a bot

Default avatar.png Saad-py: ??

JLukeSkywalker: overall programming, if you already know python, its a great scripting language, I would say learn a compiled full OO language

Default avatar.png Saad-py: you mean OOP

JLukeSkywalker: OO=object oriented OOP=object oriented programming

JLukeSkywalker: same exact thing

Default avatar.png Saad-py: I think C++ is really good but then I think of C#

Default avatar.png Saad-py: IDK which is better

JLukeSkywalker: they are both better in different situations

Default avatar.png Saad-py: C# is for game devs, app devs

JLukeSkywalker: I would say in general c++ is probably better to learn, and you could branch to C# after if you want to work with windows stuff

Default avatar.png Saad-py: I know a bit of .NET Winforms C#

Default avatar.png Saad-py: OK thx

Default avatar.png Saad-py: I will go with C++

Default avatar.png Saad-py: and btw how old are u

Default avatar.png Saad-py: ??

Default avatar.png Saad-py: ??

JLukeSkywalker: 23, been programming for 12 years, and been teaching high school and college robotics for 6 years

JLukeSkywalker: work as a control systems programmer full time

Default avatar.png DaniloKrkeljic: xDDD

Default avatar.png Saad-py: OOF

Default avatar.png Saad-py: I am -13

Default avatar.png Saad-py: Know python very well

Default avatar.png Saad-py: worked with JS also

Default avatar.png Saad-py: suck at CSS

Default avatar.png Saad-py: Did winforms for a month or so

Default avatar.png Saad-py: less scripting

Default avatar.png Saad-py: I tried to speedran hackerrank python

Default avatar.png Saad-py: run*

JLukeSkywalker: I learned python, c, java, basic, cobol, fortran and assembly by 15

Default avatar.png Saad-py: oof

Default avatar.png Saad-py: NOICE

JLukeSkywalker: i kind of program instead of playing video games

Default avatar.png Saad-py: me too

JLukeSkywalker: which is why this site is perfect, kind of both

Default avatar.png Saad-py: yeah ikr

jacek: mhm

Default avatar.png Saad-py: @JLukeSkywalker

Default avatar.png Saad-py: Which language u best at

JLukeSkywalker: c or assembly

JLukeSkywalker: i guess python and matlab second

jacek: really? oO

Default avatar.png Saad-py: bruh

JLukeSkywalker: i program hardware for a living

Default avatar.png Saad-py: You know so many languages

Default avatar.png Saad-py: Probably make a startup and go BRRRRRRRRRRRRRRRRR on the internet

JLukeSkywalker: once you know how to program and think logically, you can learn a new language in like weeks, and master it in a year or 2

Default avatar.png Saad-py: yeah

eulerscheZahl: except if the paradigm is completely different

Default avatar.png Saad-py: Imma try C++, my mind keeps changing

eulerscheZahl: e.g. learning haskell when coming from C++/Java world

Default avatar.png Saad-py: I was doing ethical hacking for a month

JLukeSkywalker: i hated learning haskell, you are right

JLukeSkywalker: still logic though

Default avatar.png Saad-py: Hacked into my own laptop successfully, and exploited the hell out of it

Astrobytes: ...

Default avatar.png DaniloKrkeljic: wtf is this chat, cmon guys

Default avatar.png Saad-py: idk

Default avatar.png Saad-py: NERDS FOR NERDS

Default avatar.png Saad-py: can anybody do a Ddos on amazon

Default avatar.png DaniloKrkeljic: ...

Default avatar.png Saad-py: I would salute you for tht

Default avatar.png Saad-py: Lmao kidding

eulerscheZahl: try it and find out https://www.codingame.com/contribute/view/6127e5b918b4155e0536e89f7157029f30bc

Astrobytes: lol, no mercy :D

Default avatar.png Angecide: that was a yikes comment he made too

reCurse: Ethical kicking

eulerscheZahl: i went afk after pinging you, not my smartest move. not sure if you saw it recurse: https://www.codingame.com/contribute/view/614074c4f45bc5120d6d04e98ee334347d54

Default avatar.png DaniloKrkeljic: yes that is amazon main server

Astrobytes: hah!

Default avatar.png DaniloKrkeljic: bezos mainframe they call it

reCurse: Interesting euler, but I wouldn't bother, that's the kind of thing they'll crack down upon very quick like it's a quest map

reCurse: *bug

eulerscheZahl: probably

Astrobytes: (non-minor)

eulerscheZahl: but if we use it to stay in the 30s but build a time bank?

reCurse: I vaguely recall them being concerned about replay size or something

reCurse: Though I did a pretty good job on that if I say so myself

eulerscheZahl: oh, you tried to get around the SDK too?

reCurse: Well I didn't use the GEM, so at least an order or two of magnitude smaller replays

eulerscheZahl: ah, good job on replay size

eulerscheZahl: SDK replays are huge as there is no domain knowledge

reCurse: I understand why they did it

reCurse: It's just newbie friendly at the cost of being very inefficient

eulerscheZahl: compressing 2048 for 1h+ replay duration was fun :D

reCurse: So CG in a nutshell

reCurse: Sorry, bad mood today I think

struct: replay size has more impact than the time the total game takes?

eulerscheZahl: we still have the possibilities to code the UI in JS ourselves, so i'm fine with an inefficient SDK

reCurse: There's all kinds of impacts

reCurse: I just remember this being brought up for 10ms turns

reCurse: Maybe there's other overheads I don't know about

struct: I see, I remember the 1MB replays when atari go was in WIP

reCurse: lol

struct: he was drawing the board every frame

eulerscheZahl: redrawing the whole game instead of updating single entities

eulerscheZahl: (╯°□°)╯︵ ┻━┻

reCurse: The cost you pay for making things too accessible

eulerscheZahl: in my very first try i made the same mistake

eulerscheZahl: but at that time there was no documentation, only UTTT

eulerscheZahl: and pong

reCurse: Reminds me of the game I wanted to make. Meh, oh well

jacek: TIL the crazyhouse white has winrate about 85%

reCurse: Uh?

reCurse: Where did you get that from

jacek: https://lichess.org/forum/game-analysis/is-crazyhouse-a-force-win-for-white?page=2

jacek: and on talkchess

reCurse: Interesting

jacek: http://talkchess.com/forum3/viewtopic.php?f=2&t=76382&start=60

reCurse: Welp glad I retired it then

reCurse: Up to anyone's guess whether 960 would help or not

jacek: on the other hand chess is dying from draws :c

reCurse: Yeah that's one reason I got so interested in shogi

reCurse: No draws

reCurse: I wish there were as many game database available as for chess, would be possible to train a human player there

jacek: why would you train human player

reCurse: To learn the game better

Default avatar.png Angecide: wouldn't it be possible to infer strategy from non-humanlike ai?

reCurse: I doubt it

reCurse: You could go into a dangerous line with only one viable answer 10 plies long, but ends up winning for you

reCurse: Trivial for AI to compute

jacek: on CG, within 50ms...

reCurse: How are you going to see that though

reCurse: No forget CG

NASIRUDDIN: hey

jacek: that gave me an idea.. to generate piece square tables from my bt bot to see how they look like

reCurse: The thing is, with training on human games and some biased search

reCurse: You can get a very very interesting player

reCurse: Worked for chess anyway

struct: https://shogidb2.com/latest

reCurse: One that beats you and you can see why

jacek: really?

jacek: like trying to mimic human?

reCurse: Yeah

Default avatar.png Angecide: I see, it's about generating sensical moves/strategies rather than obscure ones

NASIRUDDIN: hey guys

reCurse: It's about training your inner eval

NASIRUDDIN: :heart_eyes:

reCurse: There is only so much tactics you can compute

PatrickMcGinnisII: reCurse dies any boss in chess make draw offers?

Default avatar.png Angecide: yeah that makes sense

NASIRUDDIN: hey

PatrickMcGinnisII: does

struct: Dont know if that site allows for an easy way to download matches though

reCurse: It probably doesn't have enough

reCurse: I'm talking hundreds of millions

reCurse: Like lichess

struct: ah, its only 550k games

eulerscheZahl: good enough to for such a difference in winrate to be significant

kovi: shogi was implemented already with mu0, want it?

kovi: wasnt

reCurse: Yeah but that's not what I'm talking about

reCurse: Training a human player

kovi: oh, i c

Default avatar.png Angecide: it almost sound like it could be possible to train an ai to be humanlike, i.e. prune moves that does; "go into a dangerous line with only one viable answer 10 plies long, but ends up winning for you"

reCurse: It is

reCurse: I've done it, sort of

reCurse: I had plans but scrapped them when I realized how it would make cheating go out of control

Default avatar.png Angecide: cheating in which context, real life?

reCurse: Online

Default avatar.png Angecide: oh I see, yea I could see that

reCurse: PatrickMcGinnisII Sorry I missed your question, no they don't

reCurse: Is there any point to bandas or is it fundamentally unbalanced?

PatrickMcGinnisII: tx, deciding on what input to bother with

jrke: how can i check that bit 2 is 1 or 0?for bitboards

reCurse: x & (1 << 2) ?

PatrickMcGinnisII: (x>>2)&1

PatrickMcGinnisII: oh great, my gpu is under attack again

reCurse: Gotta close that viewer

Lathryx: Just got to level 3, I'm loving CodinGame so far! :)

Lathryx: Kinda seems like CodeWars.

VizGhar: (X/2)%2 :D

reCurse: That's bit 1

reCurse: Enjoy Lathryx

Lathryx: Thanks, I sure will! :)

VizGhar: it is not :)

reCurse: You did (x >> 1) & 1

reCurse: How is it not

VizGhar: ah counting from 0 sry :P

Lathryx: What language is this?

reCurse: Many

VizGhar: (x/4)%2 ? :D

VizGhar: just trolling

reCurse: Accurate trolling is best

VizGhar: hmm but I always count bits from 1

reCurse: What's next, you're gonna tell me you start indexing at 1?

PatrickMcGinnisII: ~(x^4)

JLukeSkywalker: matlab is the best language, indexes at 1

reCurse: :fire:

PatrickMcGinnisII: oops (x*4)==0

JLukeSkywalker: dont maple and R also index at 1?

eulerscheZahl: in pascal you can declare the first index like array [50..100] of Integer; thus it's better than matlab?

JLukeSkywalker: mathematicians are just annoying

reCurse: Heresy comes in many forms

eulerscheZahl: also i'm not a fan of proprietary languages in general and matlab in particular

reCurse: So octave, whatever

eulerscheZahl: except for simulink, i understand why that's popular

JLukeSkywalker: i think matlab kind of deserves to be proprietary, it is pretty powerful, and most people that would use it can get it for free

reCurse: If it keeps mathematicians nice and quiet and away from coding, all for it

JLukeSkywalker: or have their company pay for it

eulerscheZahl: i've seen better than matlab for free sagemath.org

reCurse: Hmm I heard octave mentioned most shrug

reCurse: Don't know how it compares

eulerscheZahl: i know octave, same problems as matlab bet close in functionality

reCurse: Just had to reverse engineer some matlab code once

eulerscheZahl: sage is more like mathematica or maple

reCurse: Wouldn't recommend

JLukeSkywalker: octave is pretty much matlab without toolboxes

JLukeSkywalker: and slightly different syntax

eulerscheZahl: and with a slightly modified python parser

eulerscheZahl: e.g. ^ is pow instead of xor

reCurse: :fire:

JLukeSkywalker: lol

Westicles: Isn't C# copyrighted property of microsoft?

reCurse: No

eulerscheZahl: MIT license

JLukeSkywalker: they do pretty much control it, but it is free to use/develop

reCurse: "control"

eulerscheZahl: and sage can do exact math. solving equations, simplifying expressions, ..

reCurse: They are the ones moving it forward rather

eulerscheZahl: and it even has a rubik cube solver included

JLukeSkywalker: if they dont like how you used c#, they will just release the next version that breaks your code

reCurse: Umm

JLukeSkywalker: no need to sue

reCurse: They are extremely conscious about backward compatibility

reCurse: Not sure what you refer to

eulerscheZahl: i can't remember any of my C# code breaking

JLukeSkywalker: I had a bunch of stuff break, but I think because it was super old

eulerscheZahl: only on CG, someone missed to include System.Drawing.Common when switching to .net core :(

JLukeSkywalker: think i jumped from 2.0 to 7. something

JKSTUDIOS2020: wow

eulerscheZahl: 2.0 was the first good C# version

eulerscheZahl: added generics

reCurse: ?

reCurse: Oops

reCurse: ^

reCurse: Keyboard layouts smh

Westicles: pari/gp is okay if you have a thing against octave

Default avatar.png nandu56: Hi

Default avatar.png nandu56: I am looking for a job as SDE in a reputed product based company, for practicing coding, which platform will be better for a little better than beginner?

JLukeSkywalker: I would say just work your way through the puzzles, and maybe try one or 2 bot games for some fun

jrke: just faced default in othello get timesout

jrke: https://www.codingame.com/share-replay/524569409

struct: An error occurred (#UNAUTHORIZED): "You must be the owner to share this replay".


struct: I think default is Random

struct: not sure

struct: yeah its random

jacek: and its timeout?

struct: java

Default avatar.png Mariowong: mariowong

jacek: jacek

reCurse: Gotta catch em all

JLukeSkywalker: is there a pokemon bot game on here yet?

Default avatar.png jlon: Hey, is there any way to see the correct solution after a clash?

eulerscheZahl: no pokemon

JLukeSkywalker: :(

eulerscheZahl: only if another player solved it and shared the code

JLukeSkywalker: can also check here to see the problem again and think through it https://eulerschezahl.herokuapp.com/codingame/puzzles/

eulerscheZahl: in case of reverse: search for the input given and find the task

Taseis: hi guys i am a beginner programmer i just got my bachelor's Degree and i feel like i don't know anything and i can't even solve easy algorithms like the Ghost Legs puzzle ... should i worry about myself or not yet ? (21 years old)

JLukeSkywalker: what is your degree in?

Taseis: Computer Science

JLukeSkywalker: then imo you should probably be able to solve the easy problems, but most schools suck

JLukeSkywalker: dont worry about it, just practice and research how to solve them

eulerscheZahl: and once you solved it, look at other solutions

JLukeSkywalker: ^^

Swagboy: Every part of computer science is different and you often need some time to get used to new things

eulerscheZahl: i'm serious. you can learn a few things how to code it more efficiently

Swagboy: So from theorical algorithmics to codingame problem solving, there are new things to learn

Taseis: Thank you guys i just started using this site as a practice tool and i found it really interesting i just need to remember how to code again because i was 1 year inactive from programming

JLukeSkywalker: most of the easy puzzles are logic and or math, and just bending the code you know to fit it. I think that process is something really missing in our current schools

Taseis: You really made me feel great and i'm not joking about it

JLukeSkywalker: they teach you how to code, but not how to apply it, thats what sites like this are for

jacek: :o

jacek: they taught me theory but not how to code

Taseis: Yeah we also had a lot of theory classes and not as many for practicing coding

Taseis: excuse my bad english by the way

JLukeSkywalker: thats kind of what i meant, in my head theory=how to code, applying = coding

Taseis: you were right about that

JLukeSkywalker: if you know how to do an insertion sort, but not how or where to use it, its kind of pointless

eulerscheZahl: not the article I was looking for but close enough https://blog.codinghorror.com/why-cant-programmers-program/

Taseis: nice article i am going to take a look at it :relaxed:

rav3n256: is there a way to revisit a clash problem

eulerscheZahl: https://eulerschezahl.herokuapp.com/codingame/puzzles/

VizGhar: You should put the link to FAQ eulerscheZahl :)

VizGhar: or is it already there?

Astrobytes: All clashes should output it automatically at the end.

rav3n256: or have the option after a clash is over

eulerscheZahl: that's too reasonable Astrobytes

eulerscheZahl: and a 95% problem

Astrobytes: heh heh

VizGhar: let's ask AutomatonNN maybe?

Default avatar.png AutomatonNN: eulerscheZahl is there a way to see a problem with the idea of an array of the same numbers of the s

VizGhar: why he likes you that much euler?

eulerscheZahl: where did its avatar go?

JLukeSkywalker: if you hit rank 1 but dont cross the boss, is it worth it to just try submitting the same code again to try for a higher score?

jacek: youll push other players weaker than boss

jacek: so eventually youll be higher

eulerscheZahl: which game? how much below?

eulerscheZahl: if it's some dead game, it might take ages until someone else tries to promote and submits to push you

JLukeSkywalker: i was at 29.8, boss is 30.45

JLukeSkywalker: ultimate tic tac to

eulerscheZahl: that game still has some activity

eulerscheZahl: wait unless you improve

eulerscheZahl: or try teccles

JLukeSkywalker: teccles?

jacek: teccles

jacek: we dont hilight much, do we?

eulerscheZahl: if you play on an empty micro board, send the opponent straight back to it

JLukeSkywalker: lol, i do that irl, didnt put it in the code yet

jacek: playing in center board first works because its not center board but because its first move of teccles? hmmm

JLukeSkywalker: currently im only checking micro-board for winning or blocking moves and ignoring big board

JLukeSkywalker: and almost got into silver

jacek: do you use any search

JLukeSkywalker: no

JLukeSkywalker: it can be much improved

JLukeSkywalker: was surprised it got up that high

Default avatar.png ErrorRazor: I almost got into gold with just picking the best move for the micro board

eulerscheZahl: even some "MCTS light" can get you into gold

eulerscheZahl: full random rollouts, track the first move with the highest winrate

DomiKo: MC can get you into high gold

reCurse: But what they're doing is probably the most fun they'll ever get out of uttt

eulerscheZahl: savage

jacek: oO

jacek: hmm 24 new othello players but none in wood1 yet

jacek: its almost as if they dont care abot puzzle of the week quest

eulerscheZahl: let's do it next week when it's easier

**eulerscheZahl completed the quest with "Fractal carpet"

reCurse: I only care about the trolling dailies

struct: 34 players jacek

jacek: mhm

struct: maybe its in the same level as getting legend in uttt?

struct: I didnt write the bot

struct: So I am not really aware of the difficulty

jacek: it was made by teapot, so its easy

reCurse: Rude

DMMVS: hello?

Jaystah: suh

Jaystah: ich gib ein fick auf euer fotze nach

jacek: euler, thats for you

eulerscheZahl: my first ever ban :(

jacek: hm?

Astrobytes: He seemed like a pleasant fellow...

eulerscheZahl: well deserved but still. not there's blood on my enter key

reCurse: You'll get used to it

Astrobytes: ^

Westicles: wow, that's like ban golf. two lines

reCurse: You should see what the nice fella I kicked this morning wrote me in PM

reCurse: Got upgraded to a ban

jacek: i mean, what did he say really?

Astrobytes: That's always the part I look forward to, the PM.

eulerscheZahl: German, I wont translate

jacek: german :scream: good

jacek: i see, so you banned him for using german :+1:

eulerscheZahl: nein

reCurse: Technically correct

jacek: the best kind of correct

reCurse: Mayb glad I didn't get a ryzen, the next intel looks yummy

reCurse: avx-512 on a home cpu finally

Westicles: Last one had it as well

reCurse: The 10th gen? No

Westicles: I got the 10980xe last Jan

reCurse: Oh xe

reCurse: Missed that I guess

struct: Bettter pre order it asap

reCurse: Currently $250 off :o

reCurse: Ok where's the catch

reCurse: Seems way too cheap for what it is

eulerscheZahl: my dad thought so too when he bought a video card on ebay

eulerscheZahl: i told him no

eulerscheZahl: of course he got scammed

Default avatar.png Angecide: how much did he lose

eulerscheZahl: he opened an incident on ebay and got the money back

reCurse: Ah lower clocks

Default avatar.png Angecide: ah that's nice

eulerscheZahl: so just some time and 25€ for an IT shop to testify that he didn't get what he paid for

reCurse: 18 cores though... :drool:

struct: leaked prices seem a bit low

struct: 604$

struct: for the 11900k

wjan: http://chat.codingame.com/pastebin/f4df6680-9a2e-4aec-aed1-33af285ca131

reCurse: Damn now I'm torn

eulerscheZahl: pastebins are back

eulerscheZahl: confirmed

reCurse: 11900k for better and cheaper cores or 10980xe for need moar cores

eulerscheZahl: there are several cooldowns wjan

eulerscheZahl: the longest is 6h i think

eulerscheZahl: but depending on how much you spammed, it might really just be a few minutes

Westicles: The xe motherboard costs a little more. Quad channel memory instead of dual though

reCurse: Hmm only have 2 sticks atm

reCurse: Think 11th gen might be more worth it

reCurse: Just glad avx-512 is slowly becoming mainstream

jacek: linus bashed intel for avx512

jacek: its slows the clcok

jacek: clock even

reCurse: Depends how you use it

Westicles: And you always have to check how much of the instruction set they actually have.

jacek: even stockfish folks found it wasnt that much different with nnue

reCurse: I don't get the impression nnue would have much to gain out of avx512

reCurse: Could be wrong

Westicles: nice chart at the bottom here

Westicles: https://en.wikipedia.org/wiki/AVX-512

jacek: https://www.extremetech.com/computing/312673-linus-torvalds-i-hope-avx512-dies-a-painful-death

reCurse: Thanks

reCurse: Just need to wait for that rocket lake

jacek: you should be glad for amd. it pushes intel to its limit

reCurse: I'm not dissing amd

JLukeSkywalker: nice lol, changed "col" to "2-col" and my bot went from bronze t osilver

jacek: huh

JLukeSkywalker: tic-tac-toe

jacek: oh my

JLukeSkywalker: if my code found a winning move in column 3, it would place in column 1

JLukeSkywalker: and not win

JLukeSkywalker: im smrt

JLukeSkywalker: that took like an hour to find

jacek: what i really like doing this ML stuff is doing those cool charts https://i.imgur.com/u07jmw4.png

reCurse: Might have looked into doing accounting instead, easier

Default avatar.png Angecide: when doing ml for cg do u generally train the model in python and then serialize the weights to c++?

jacek: you train offline, yeah

Default avatar.png Angecide: but is there any standard choice between training in python vs c++ or is it just based on your own preference, since I assume it is possiblet to get the weights from python to c++?

Default avatar.png Angecide: they are just numbers in a matrix I guess

jacek: yeah, they just params to optimize. however you do it, doesnt matter

Default avatar.png Angecide: I see

jacek: could be python libraries, as they are pretty optimized

Default avatar.png Angecide: yea the accessibility of doing ML in python is really great

struct: on csb the angle is rounded, floored, ceiled or truncated?

reCurse: Which angle

struct: pod angle to next checkpoint

reCurse: Which league is that

struct: legend

reCurse: I thought pod angles were absolute

reCurse: In any case, they are given to you rounded

reCurse: The referee keeps the real float

Default avatar.png ErrorRazor: They are absolute, I'm doing csb rn

struct: I see, thanks

reCurse: Results in around 5% inaccuracy on average

struct: lol

Default avatar.png FabioGomes: does anyone know the solution to this ? https://www.codingame.com/training/expert/the-lucky-number

Default avatar.png ErrorRazor: Yup

Default avatar.png FabioGomes: Fábiooo13#5962 add me on discord

jacek: i think 21% knows

Default avatar.png ErrorRazor: Why does a private chat pop up every time the chat reconnects even though there are no new messages in it and I've closed it

Astrobytes: Known bug.

Default avatar.png ErrorRazor: Alright

Default avatar.png zrdzrt: is there any chance that java wins in short mode vs. python? Feels unfair tbh :D

Nikeleos: Wich language can beat Pyhton in short mode ?

Hobogre: Ruby will always win

JLukeSkywalker: also who doesnt know python anyway

Default avatar.png ErrorRazor: I'm not a fan of short mode

Default avatar.png zrdzrt: same

Default avatar.png ErrorRazor: You don't learn anything useful from it

Nikeleos: Exactly

struct: ruby perl bash

Default avatar.png ErrorRazor: Maybe just treat it as fast mode

reCurse: Then you should not be a fan of clash either

Nikeleos: Bash ? Creapy...

Default avatar.png ErrorRazor: I do prefer the bot games but sometimes it's nice to do something smaller

Astrobytes: That's why we have puzzles

Default avatar.png ErrorRazor: I'm definitely neglecting the solo puzzles.

Astrobytes: You'll get more out of it than you will from clashes imo

Default avatar.png ErrorRazor: I shall clash no more

Astrobytes: Hey don't stop because I said puzzles lol, clash if you feel like it - I think the objective of clashes was 'Fun'

Astrobytes: But if you want to actually learn anything, then puzzles or whatever

Default avatar.png zhoubou: Puzzles are a great way to have fun while learning. At least for me.

jacek: "fun" "clash" in one sentence?

Default avatar.png ErrorRazor: "clash not fun"

Default avatar.png rick-James: very cool :)

Default avatar.png o0lit3: "Ruby will always win" >> perl usually beats ruby tbh

JLukeSkywalker: but no-one uses perl in clash

Default avatar.png o0lit3: I do

Default avatar.png o0lit3: it generally has a leg up on "fastest" as well because you dont' have to type as much

Default avatar.png ErrorRazor: "no-one" is a pseudonym for "almost no-one"

Default avatar.png o0lit3: you sound like an english teacher and not a programmer

JLukeSkywalker: sometimes i wish programmers were english teachers, would make code more readable

Default avatar.png ErrorRazor: I wouldn't be a programmer if I didn't like languages

kavandoctor: wassup people

kavandoctor: join coc

kavandoctor: coc is fun

JLukeSkywalker: or less readable ....

Default avatar.png ErrorRazor: join multiplayer

kavandoctor: i like coc

Default avatar.png ErrorRazor: multiplayer is fun

kavandoctor: yeah

Default avatar.png ErrorRazor: I should've said ai battle

Default avatar.png ErrorRazor: Because clash is technically multiplayer too

Perlorodka: kavandoctor! don't say you improved even more in CoC! :)

kavandoctor: :)

Perlorodka: I think coding roman to decimal convertor in 38 seconds is already pretty damn good https://pasteboard.co/JKIpvQe.png

kavandoctor: bruh

kavandoctor: ur allowed to look stuff up

kavandoctor: so i just found a sol online

kavandoctor: lmao

Astrobytes: Exact solution?

Perlorodka: of course, awesome job

kavandoctor: thanq

kavandoctor: but i mean

kavandoctor: that was the only time i did that

kavandoctor: cuz i didnt feel like

kavandoctor: writing roman numerals

Astrobytes: Was it a straight up copy paste?

kavandoctor: what no

kavandoctor: i had to change stuff

kavandoctor: obviously

Astrobytes: And it took btb all that time to look it up and change some stuff

Astrobytes: I mean: https://www.w3resource.com/python-exercises/class-exercises/python-class-exercise-2.php

drdrdrdoctor: i mean yeah

drdrdrdoctor: i think that was the only challenge i saw

drdrdrdoctor: that was p much entirely online

Astrobytes: (one of many available solutions to the general problem)

drdrdrdoctor: yeah

drdrdrdoctor: but i mean

drdrdrdoctor: You can look up stuff. You can take your time. You shall have fun. Everyone is a winner, even if 0% of the tests work :D

drdrdrdoctor: "You can look up stuff. You can take your time. You shall have fun. Everyone is a winner, even if 0% of the tests work :D "

drdrdrdoctor: this is what it says

drdrdrdoctor: on the page

drdrdrdoctor: so i assumed you were allowed

drdrdrdoctor: to search stuff up

Astrobytes: You are, yes.

Astrobytes: I'm not judging.

drdrdrdoctor: oh

drdrdrdoctor: you wanna play?

drdrdrdoctor: https://www.codingame.com/clashofcode/clash/1563401c56e47651726c8f56df54b32c24ee589

Astrobytes: Nah, I don't clash. I prefer bot programming.

drdrdrdoctor: oh

drdrdrdoctor: oki

drdrdrdoctor: i hate bot programming

drdrdrdoctor: and like clash

Astrobytes: All good.

Velcoro: I'd clash, but I hate googling solutions, you know :)

reCurse: Wonder why

drdrdrdoctor: bruh

drdrdrdoctor: that was like the only question

drdrdr: that had a online solution

drdrdr: lay off me

Astrobytes: Sure.

Astrobytes: Any pressure you're feeling is completely your own.

Default avatar.png TheSunnyCastle_ee5e: hi

Default avatar.png TheSunnyCastle_ee5e: asd

Astrobytes: Ever read The Tell-Tale Heart?

Astrobytes: Anyway, before all the beasts appear in chat I'm out, gn all

Default avatar.png ErrorRazor: Good night

struct: gn

Default avatar.png TheSunnyCastle_ee5e: a

Default avatar.png TheSunnyCastle_ee5e: a

3ataja: why it always empty at night

3ataja: programmers aren't supposed to sleep at night

JLukeSkywalker: we are all just too comatose to chat

3ataja: no i mean clash of code

Default avatar.png rekaxem: agreed

SeraphWeddUltimate: maybe they're too busy doing something else?

Default avatar.png rekaxem: hi

Default avatar.png Cyrgo: https://www.codingame.com/clashofcode/clash/15634614267234c4828a17b8b4ae5f7d336cfa1

Default avatar.png oiuawesfiawegioaew: o

Default avatar.png Saad-py: Hi