Chat:World/2022-06-22
antiwonto: [auto] Hey Mengshang, here is a :taco: for loggin in today while it is quiet. You now have 1 tacos
jacek: :tada:
AshamanCooper: If you are driven this is a great place to learn to code. I started two months ago and I am 406 in the green circle comp.
AshamanCooper: But you still need to use more than one resource.
Mengshang: I am Chinese.When I want to code , I must to translate:joy:
therealbeef: the IDE supports unicode, so you could have variable and function names in chinese I suppose ;-)
rachelktyjohnson: Hi everyone! I've just come back to Codeingame after a LONG hiatus. So much has changed!
Tien118: antiwonto, what can you do?
jacek: right into the contest
AshamanCooper: What has changed? that would be interesting.
Mengshang: Variable or function in Chinese... Bad thing!
AshamanCooper: Hard to camelCase
Westicles: Jeez, I've got this weirdo social justice dingbat rejecting by clash three times now because kissing girls offends it
Westicles: *my
Clams: :laughing:
Westicles: :tada:
Mengshang: although my English not well,but I try my best to name function or var , class in English.You can see, there are many grammar mistake in this sentence,I am a junior high school student in China
MSmits: these are not the same are they?
MSmits: a = b + c? d : e; a = b + (c? d: e);
rachelktyjohnson: the interface has changed a lot. there were a solid 6 games to play around with back then
MSmits: crap emote
MSmits: a = b + c? d : e; a = b + (c? d : e);
MSmits: yes so they are different?
Mengshang: You can input the text and try to run
MSmits: lazy
Mengshang: Can you read what I'm saying?
MSmits: yeah
Mengshang: I'm worried that foreigners won't be able to read my "Chinese English"
MSmits: no no it's fine
Westicles: I bit Mengshang kisses lots of girls
MSmits: I am was just wondering if i gave myself a horrible bug
Westicles: *bet
MSmits: and it seems like i did, with that ternary assignment
Mengshang: I just finished my 16th birthday
MSmits: grats
Mengshang: okok,I try to run it in vs
Westicles: Although joint probability might be too tough for a clash anyhow, not sure
Mengshang: int a, b = 20, c = 30, d = 40,e=50;
Mengshang: a = b + c ? d : e; std::cout << a << std::endl; a = b + (c ? d : e); std::cout << a << std::endl;
MSmits: nice one, thanks =)
Mengshang: Have you ever known about China's development?
MSmits: not sure what you mean
Mengshang: I mean, in your eyes, how developed China is now
Westicles: I used to travel to Shenzhen back in the wild early days, great times
MSmits: I am guessing it depends a lot on what part of the country you look at
GroutchKeu: o/
jbiz: on code clash, frequently me and other players get less than 100%, despite getting 100% in the tests before submission. why?
anthonyashco: They use different tests with possibly different edge cases for final scoring.
anthonyashco: It prevents players from hard-coding answers for difficult edge cases.
jbiz: perhaps but i'm pretty sure at least some of them were fool proof.
jbiz: this last test literally everyone got 50%.
anthonyashco: It's certainly possible there are errors, since the tests were written by humans. If you think there's a real error in the test cases I think you can notify people.
anthonyashco: However, if it's happening frequently, I'd suspect there are surprising edge cases that aren't being handled.
jbiz: alright, i'll report after checking it if it happens again
anthonyashco: They don't always give you what the validators are, but you might be able to deduce what's missing by the validator's name.
anthonyashco: Otherwise, you can click the forum button to discuss your findings.
miobyte: can anyone help me with k8s?
MSmits: lol i read that as help me with kids
derjack: :upside_down:
MSmits: i am sitting here with a class of 29, so it's kind of applicable :P
Westicles: summer school? here only the ones who fail something have to do that
eulerscheZahl: here summer break only starts in August
derjack: w00t
Westicles: well, you guys have to learn eight languages just to get around your own time zone
MSmits: no it's the last class of the year for these
MSmits: next week is testweek
MSmits: and then there are 2 more weeks of almost nothing
MSmits: class of the schoolyear i mean
Westicles: Is 220 good enough for gold? That hint from yesterday really worked
MSmits: hmm
MSmits: i dont think so
MSmits: i guess 100
MSmits: silver is usually a little less than half of bronze and gold is 10% of silver or so, correct me if i am wrong eulerscheZahl
MSmits: 20% of silver i mean
eulerscheZahl: 20% sounds about right
eulerscheZahl: did you get the brutaltester running yesterday?
MSmits: nah i got distracted
MSmits: I keep finding bugs
MSmits: most of my bot improvements are bugfixes :P
MSmits: I guess this is what happens when you churn out 1k lines of code in 2 days to get a sim working asap
Nerchio: damn
Nerchio: makes me think I have poor man's sim if I have 1k lines in java and I didn't try to keep it low
MSmits: the difference is probably in the eval
MSmits: you can have a bad bot with a perfect sim and a poor eval. And your bot is not bad
Nerchio: yeah it's starting to get difficult to change eval to make it better
Nerchio: yesterday I did it for 2h to keep up with top20
MSmits: same
MSmits: it's stops being fun when you run out of new ideas for eval features
MSmits: just param fitting is boring
Nerchio: true I usually give up after a little while of param fitting but I am not in that spot here yet
MSmits: same
MSmits: I keep having to refit params after i fix bugs
MSmits: hi
TINOUAINANI: good morning
TINOUAINANI: question: what happens if you use CODING and afterward you use TRAINING, do you still have one card to play or two?
darkhorse64: You should get two. I think the game summary gives this information
TINOUAINANI: ok thanks
TINOUAINANI: guys is there a way to print to console without a limit, I want to see all my stderr? (c++)
Leeward: Not within the site, but you can encode it and then decode it elsewhere
TINOUAINANI: ok thanks
eulerscheZahl: my bot sucks. even importing the opening from over 200 replays of the duck didn't help :(
Nerchio: damn some next level strats eulerscheZahl
eulerscheZahl: next level fail
Nerchio: i submitted 3 times in a row and ended #15 every time
eulerscheZahl: at least the replays were easy to parse, thanks to the excessive game summary
Nerchio: my bot is consistently bad
eulerscheZahl: currently i would be happy about your rank
Nerchio: problem is i dont exactly know how to fix it
derjack: write better bot [solved]
eulerscheZahl: let's try not to give cards to the opponent that often
Nerchio: i give away cards quite often but usually its connected with a good move
eulerscheZahl: i do it even for bad moves
Leeward: I have a question about Code4Life if anybody has played that...
Leeward: I'm currently in Wood 1. I am able to collect a sample that has a cost of more than 10 molecules total but I cannot carry more than 10 at a time. The way the lab works right now I can only hand in a sample along with the entire cost to get the points. Is there a way to "pay in instalments" or can I simply not process those samples right now?
eulerscheZahl: there are experience points for completing samples. then you need one molecule of that type less
Leeward: I don't think I can use expertise yet, it's listed under the "What is in store for me in the higher leagues?" section.
eulerscheZahl: oh, then you have to wait and not take those samples yes
eulerscheZahl: yet
Leeward: I thought so, thanks
Leeward: About to get promoted to next league though so moot point now lol
derjack: :tada:
Westicles: so, about hardcoded opening moves
Ftyghi: hi
PrimeDecay: Hi guys!!!
eulerscheZahl: don't waste your time with that Westicles, doesn't work
Westicles: well, it is 300 places better than first move
Psyho: I'd say hardcoding is still useful
geppoz: if (cardTaken.isGoodActionBonus() && action.getThirdCardType() != CardType.BONUS) {
gameSummaryManager.addSemiUselessTaskPrioritization(player, action.getThirdCardType()); }
eulerscheZahl: maybe taking 200 replays from the duck wasn't enough them. or i'm too picky and it's just about opponent close vs not close (without looking at the exact location)
geppoz: what does it means that in the taskPrioritization class ?
geppoz: "SemiUselessTaskPrioritization" ? :D
Nerchio: I don't think its the opening that gets him first place euler
eulerscheZahl: get a bonus instead of a skill card
eulerscheZahl: it's surely better than my opening
eulerscheZahl: but i still have the code in place to steal from someone else
Psyho: I got a massive boost my disabling releasing in the first two turns :D
Psyho: *by
Psyho: my sim likes to hoard debt cards
eulerscheZahl: will you keep going here on CG? topcoder marathon starts this evening
eulerscheZahl: and you are in a good position for TCO
Psyho: but the big problem with such strategy (apart for huge variance, so that 200 games is not enough to stabilize the ranking) is that you give your opponent easy time of selecting the apps that block your progress
Psyho: I'm switching to marathon today, I only have 100 points (just did a single marathon) and I probably need to win again in order to qualify for the TCO
eulerscheZahl: 2nd or 3rd might be enough. but 1st will surely help
eulerscheZahl: in any case: good luck with TCO
eulerscheZahl: it's "The Encore" in Boston this year
Psyho: I have to be above tanzaku & Jaco + few other people can snipe my spot
Nerchio: what kind of tasks they have in there
Psyho: optimization
Psyho: it's mostly weird heuristics, local search (simulated annealing), some pathfinding (beam search mostly)
Darleanow: Can we move backwards ?
Darleanow: and can we cross player eg p1 is on 2 and p2 is on 3, can i move 4 ?
eulerscheZahl: the last 3 tasks: https://www.topcoder.com/challenges/9bb1c5e1-dbb3-4407-be0e-3cc0dc45bf1a https://www.topcoder.com/challenges/ef5d19d2-de9b-4e37-b428-e7e103458425 https://www.topcoder.com/challenges/5ac77343-7f2f-4635-8aff-bbf49c700fda
eulerscheZahl: you can even go to the same desk as your opponent, Darleanow
eulerscheZahl: but also walk by
Darleanow: okay !
Darleanow: but we an't move backwards right !
Darleanow: Also, after submit application, can i change the way i move, eg going backwards ? and can i do it initially ?
Darleanow: sorry for the spam - -'
eulerscheZahl: you can go anywhere except yto your current location
struct: hi
eulerscheZahl: huhu
MSmits: no
MSmits: cant go backwards, sorry scroll
jacek: :upside_down:
aangairbender: bug with not going to PLAY_CARD phase after TRAINING still exists? I've just met it
Nerchio: it was fixed
kovi: phase is still missing if there is no useable card
aangairbender: I guess referee just automatically sees that no action can be played so he skips
aangairbender: okay, thank you
kovi: yes
Nerchio: yep
aangairbender: viewer was just showing this in a weird way (I didnt see card I got after training)
PiterYeh: aangairbender if your turn is over, all your cards are discarder and what you see in the viewer is the hand you will have in your next turn
PiterYeh: you can see which cards "training" got you by reading the referee output
aangairbender: okay, thanks for help
struct: Was forgetting to remove training card from my hand when playing it
struct: might explain the million of states I was getting
struct: :p
jzen: Is the bonus card considered a skill card?
aangairbender: jzen yes
jzen: Thanks
struct: in what aspect?
struct: For me a skill card is a card you can use during play phase
BrunoFelthes: can i use the task prioritisation card acquired at the same turn?
struct: yes
jzen: Yeah, I thought so too, but then I was looking at throw. It says you can throw a skill card from your hand, but looking at source that includes the bonus card
geppoz: bonus cards can be used for THROW, for GIVE, and in TASK_PRIO
geppoz: and in CONT_INT
struct: yes you can throw bonus
struct: you only cant throw/give technical debt cards
MSmits: struct, also remove the coding card from your hand when coding
elliotdingalberriesmells: this is a dead game icl
elliotdingalberriesmells: all of yous are sad
**Rodrigo_the_coder slaps /slap around a bit with a large fishbot
MSmits: I don't feel sad?
Astrobytes: it's like when they come in and call us sad nerds. We're happy nerds!
eulerscheZahl: nerds in cage housing. that's how we like it
Astrobytes: lol
struct: I finally saw the video euler
struct: I knew it was bad, but no idea that the industry was that bad
Astrobytes: you don't remember re curse regaling us with tales from the industry?
eulerscheZahl: he manages to entertain us and make us sad at the same time
struct: not really astro
eulerscheZahl: recurse even linked an article basically telling the same as we saw in the video
struct: Maybe I wasn't online
eulerscheZahl: and there's the ping
struct: not online
Astrobytes: he used to talk about it a lot tbh, ubisoft being one of the better places to work it seems
putibuzu: what video if I may ask?
struct: https://www.youtube.com/watch?v=pLAi_cmly6Q
eulerscheZahl: beat me, i had to search it again
eulerscheZahl: oh, btw:
eulerscheZahl: Hi,
You have used 80% of your 550 free dyno hours for June 2022. The following steps can help you keep your app running smoothly.
eulerscheZahl: website will go down again
Astrobytes: it's a popular site nowadays
eulerscheZahl: "Get more free dyno hours per month Simply verify your account with a credit card and receive an additional 450 free dyno hours per month." sounds simple. but i don't have a credit card
eulerscheZahl: and most visits are from search engines
Astrobytes: really?
Astrobytes: (re search engines)
eulerscheZahl: what even is a SemrushBot?
eulerscheZahl: bingbot, pedalbot, googlebot, ...
eulerscheZahl: so many of them
Astrobytes: SemrushBot is the search bot software that Semrush sends out to discover and collect new and updated web data.
Astrobytes: Data collected by SemrushBot is used for: the public backlink search engine index maintained as a dedicated tool called Backlink Analytics (webgraph of links)
Astrobytes: TIL. I thought it was a bot that was kind of in a hurry but not overly so.
eulerscheZahl: i guess i go up a bit in website ranking. then i run out of dynos and drop again because i'm offline
eulerscheZahl: can't be good for your page rank to have an error page
Astrobytes: indeed
Haikooo: someone can explain me what is readline(), In internet I just find docs for Node.js but i'm using for JS
eulerscheZahl: there is no readline actually, codingame wrote their own to support JS on the website
eulerscheZahl: gives you a line from stdin
Haikooo: ok so readline() it's similar like what ?
Haikooo: what is the goal of this function
struct: it reads an input line
Astrobytes: to read a line from stdin as he said.
eulerscheZahl: are you familiar with console applications? that black window with text only
Miki09: can task priorization be used on itself?
Astrobytes: if you have more than one I think so
Miki09: ok thx
3ssila_mli7a: What is the name of the algorithme that we have to use in the practice: Death First Search - Episode 1
Astrobytes: :D
Astrobytes: Try replacing a letter in 'Death'
3ssila_mli7a: Depth First Search ? is this an algorithm, sorry I am new to this
Astrobytes: or actually is it? I think it's Breadth First Search actually, my bad
Astrobytes: I think the second one involves Depth First
3ssila_mli7a: Breadth First Search it is
3ssila_mli7a: I'll try n search for it
Astrobytes: if it's new to you, I strongly suggest this site: https://www.redblobgames.com/
3ssila_mli7a: To see what it s like
Astrobytes: Very clear explanations
3ssila_mli7a: Oh, thank you very much I'll take a look at it
Astrobytes: no problem, hope it helps
Astrobytes: how many was that to gold league? 100?
eulerscheZahl: oh, gold
struct: a bit more
struct: I was promoted and was around 110
struct: 109
struct: I was
Dapps: whoops.. .shouldn't have just submitted
Astrobytes: ah ok
Miki09: Dapps me too :-)
BrunoFelthes: 74 now
eulerscheZahl: for gold opening it's fine. the league isn't big enough to crash the server
struct: ah no
struct: I wasnt promoted
struct: I just got the message the league opened
eulerscheZahl: 100% tie rate so far https://i.imgur.com/m28c0iv.png
Astrobytes: ah
eulerscheZahl: and 2 more
Dapps: any way to tell which AI became boss? or just if they have any output
Astrobytes: must have been 80-90 then
eulerscheZahl: the only way is to play boss vs itself and do the same with some players in that range
Michael_Howard: Only 74 of us by the looks of it.
Miki09: the servers are slowwwwwwww
struct: legend will be pretty hard
Astrobytes: a few more still to promote Michael_Howard
Astrobytes: ~12 currently
eulerscheZahl: 19 matches (38 in total) and all draws. that's unlikely to happen
struct: well its the same problem that uttt has
Astrobytes: wow
struct: 1st player will always almost win
Astrobytes: yeah it's quite skewed
eulerscheZahl: but seriously: 22 mirrored matches now. all ties
eulerscheZahl: and it just keeps going
wlesavo: eulerscheZahl look inside
eulerscheZahl: crashes :D
wlesavo: first player timeouts everytime
mikmak: yes the input are probably incorrect
Astrobytes: lol whaaay
Astrobytes: or whaaat
kovi: match# cut down for rerun i guess
eulerscheZahl: so player 2 wins
Miki09: zach_lee probably found a way to fix it
Miki09: and he wins all the time
wlesavo: and why i have fr rules now? :smiley:
eulerscheZahl: http://chat.codingame.com/pastebin/74936757-7508-4057-bfec-78ae210c549a
eulerscheZahl: Unhandled exception. System.FormatException: Input string was not in a correct format. at System.Number.ThrowOverflowOrFormatException at System.Number.ParseInt32 at System.Int32.Parse
therealbeef: something is weird, my bots prediction of the next game phase is wrong all the time
Astrobytes: I thiiink someone should report this
eulerscheZahl: cardLocationsCount is gone?
therealbeef: in silver it always matched the referee
eulerscheZahl: i expect a number and get: HAND 0 0 0 0 0 0 0 0 3 1
wlesavo: what a way to fix p1 advantage :grinning:
Astrobytes: :D
eulerscheZahl: [SG]Sebastien did you touch the referee?
BrunoFelthes: the winner of this contest will be defined by luck.. who win two mach first, will win more points...
Astrobytes: perhaps it's reverted to another version
Michael_Howard: Is it normal to go way up in rank after promotion?
Michael_Howard: I've gone from 70s to 30s.
eulerscheZahl: they should do that for the final rerun to see how bulletproof our codes are
eulerscheZahl: i'm 12th \o/
eulerscheZahl: 11th
wlesavo: Michael_Howard yes game is broken
wlesavo: im 8 :smiley:
eulerscheZahl: and top3 adapted to the new format already
eulerscheZahl: top4 probably
BrunoFelthes: what new format?
Michael_Howard: wlesavo but why is it broken in my favour in Gold? :upside_down:
eulerscheZahl: changed input format. probably by mistake
Astrobytes: someone in gold report it on discord please
BrunoFelthes: what changed in input?
Nerchio: wait what happened
eulerscheZahl: they don't give cardLocationCount anymore
therealbeef: tooltip all weird too
Nerchio: why is my description in french :D
eulerscheZahl: best guess: older version of the game
eulerscheZahl: probably all the bugs are back as well
eulerscheZahl: right, french as well
eulerscheZahl: someone ping staff
Nerchio: @staff hello
Miki09: First time ever in my life I beated Neumann https://www.codingame.com/replay/644522279
Michael_Howard: I'm ahead of kovi! Stop the count!
[SG]Sebastien: eulerscheZahl : no modification on referee since monday's fixes
Astrobytes: reported in Bug Report EVEN THOUGH IM NOT IN GOLD
eulerscheZahl: then something went wrong with automated league opening
wlesavo: hehe time to get top1
eulerscheZahl: with thibaud ping?
Astrobytes: he's offline
Nerchio: i want a refund
Michael_Howard: Ahead of eulerscheZahl!! :laughing:
eulerscheZahl: [CG]Thibaud is online here
Michael_Howard: Whatever the bug is I like it
eulerscheZahl: ping. game reverted to a prior version on gold opening (supposedly)
Astrobytes: that's my theory
Astrobytes: can't think of anything else
wlesavo: best ranking improove in history by changing a single constant
- smiley:
eulerscheZahl: always 6 instead of dynamic?
wlesavo: yes =)
eulerscheZahl: i'll just want for CG
kovi: french description?
eulerscheZahl: that's the only thing you notice?
kovi: gold
eulerscheZahl: is silver still working as expected?
Miki09: yes
eulerscheZahl: all bots crashing because of changed input format
eulerscheZahl: astro reported it already
Astrobytes: I don't know if anyone's working on it yet, no indication, but I presume they know by now
eulerscheZahl: always RANDOM can bring you in top10 again
[SG]Sebastien: I tried to ping Thibaud, but no answer yet
eulerscheZahl: no biggie, they will find out
Astrobytes: should I ping Julien?
eulerscheZahl: him or Saiksy
eulerscheZahl: thibaud would do the same
Astrobytes: I've pinged Julien
wlesavo: still haven't lost a single game out of 70, i guess top is not adopted euler
eulerscheZahl: take your screenshot
Nerchio: euler we are at the bottom of the league, maybe we shouldnt be playing games with random draw
ilgiocatore: almost there
ilgiocatore: i want to see the match where both players see 6 cards
struct: stop
Illedan: oooh, gold
square1001: I learned that AI says "IA" in French now
Nerchio: your bot is crashing Illedan
eulerscheZahl: you missed the funny part Illedan
Astrobytes: oui, c'est l'intelligence artificielle square1001
Illedan: wtf?
Illedan: What is funny?
eulerscheZahl: the crashing
eulerscheZahl: input changed. french statement
Illedan: I see it now
Illedan: LOL
Illedan: Did you ping staff on discord?
Astrobytes: Yeah
eulerscheZahl: astro pinged julien
Illedan: Did it work?
Astrobytes: No response as of yet
eulerscheZahl: do you see any improvements?
eulerscheZahl: just submit a RANDOM bot
[SG]Sebastien: based on a replay I saw, µI confirm this is an old version of the game
Illedan: Meh
Illedan: And I had 10 min to try something
Astrobytes: I used Bug Report Illedan
eulerscheZahl: what is bug report?
Astrobytes: [SG]Sebastien theory confirmed :tada:
Astrobytes: the channel on the discord euler
eulerscheZahl: you are the sharpest tool in the box
eulerscheZahl: ah, that # :bug: bug-report
Astrobytes: lol, yes
ilgiocatore: yaay i managed to climb before a bugfix, took my screenshot and now i can rest
eulerscheZahl: this doesn't count
wlesavo: in old version daily worked only in one direction, inrteresting
wlesavo: i think that is kind of more balanced
Astrobytes: alright, Julien has seen the message "oh dear"
Miki09: can task priorization throw itself away?
Astrobytes: only if you have more than one, that's what I meant earlier Miki09
Astrobytes: (that's what I thought you were asking)
Miki09: thank you. I didnt know if I formulated the question correctly
eulerscheZahl: interesting experiment: how fast will the leaderboard stabilize again without a full resubmit? just slowly bubbling up
Astrobytes: no worries :)
Astrobytes: I'm almost certain they'll do a full resubmit euler
Michael_Howard: eulerscheZahl in O(n2)
Michael_Howard: O(n²)
eulerscheZahl: n² take a square
eulerscheZahl: ¹³⁴⁵⁶⁷⁸⁹⁰
eulerscheZahl: weird, drops after 4th power
Astrobytes: exponential decay
BrunoFelthes: my statements changed to fr, how to back it to english?:
Astrobytes: there are bigger issues than that BRuno, it's reverted to an earlier version of the game
[SG]Sebastien: the statements are in the gazme package. All should be fixed at the same time
BrunoFelthes: so, the input will not be changed?
[SG]Sebastien: it will revert back to what you had before
BrunoFelthes: oh, ty
Wontonimo: lol, exponential decay
Astrobytes: eulerscheZahl: is it working now?
Astrobytes: It's been updated
Astrobytes: Wontonimo :D
Astrobytes: Anyone in gold league confirm it's working?
wlesavo: it is
wlesavo: now im timing out
wlesavo: :smiley:
Nerchio: punishment for all the impatient people :D
Nerchio: so do we have to resubmit ourselves
Astrobytes: I guess
Nerchio: there's no games as p1 p2 though
eulerscheZahl: i'll wait for the bubble :popcorn:
wlesavo: Nerchio maybe it is still an older version but a different one :smiley:
wlesavo: we should check if the training bug is still there
eulerscheZahl: it's an English one
[SG]Sebastien: games p1/p2 then p2/p1 is a set up on codingame's side, not in the game source code
Astrobytes: wait, there's no mirrored games as before?
eulerscheZahl: mirror matches are gone
Astrobytes: ok let me report that
Nerchio: we should let it stay this way so I have better chances of being in the top
Nerchio: by just being lucky
eulerscheZahl: but i can confirm: the game creator has no affect on mirror matches
Psyho: does rating distribution (like rating inflation) impact match making system on CG or does only care about relative placements?
eulerscheZahl: opponent = your rank-10 to +10 or so
eulerscheZahl: not sure if equal probability
Astrobytes: Julien is on the mirror match issue (broken in other leagues too btw)
eulerscheZahl: you pinged the right guy
Astrobytes: ok should be fixed
wlesavo: euler i think probability is based on the score difference
wlesavo: fixed in gold confirmed
Astrobytes: same in silver (Should be the same across all leagues)
jacek: first gold in contests :tada:
Nerchio: hidden NN incoming?
Astrobytes: grats jacek!
Psyho: it will take ages for this ranking to make sense
MSmits: are you saying you should not be nr 1 Psyho?
MSmits: :)
Psyho: probably, although at this point I have no clue
Psyho: it's a new version as well
MSmits: oh i see
Nerchio: well i was top15 not top2 so yeah xd
Nerchio: Psyho number1 is more probable
struct: We all know who is getting rank 1
MSmits: It must be someone who puts training cards in played pile
struct: definitely
MSmits: nice bug btw, i also found some
MSmits: small ones
MSmits: I forgot to reduce from a desk when i picked up a card
MSmits: hand[i]++ but not board[i]--
MSmits: doesnt matter a whole lot, but it matters a little later on for eval
MSmits: oh and i calculated distance wrong for daily thingy
MSmits: so i could not pick up from some desks
struct: I think we can agree that you only found it because I reported my bug
struct: :p
MSmits: hehe i dunno. I am bugchecking all day and i keep finding new ones :)
struct: I should do that
MSmits: it's more efficient than fitting parameters for me
MSmits: I really rushed creating the sim
struct: I'm going to redo mine
jacek: how shoddy of you
MSmits: I googled shoddy lyric
MSmits: got this
MSmits: http://chat.codingame.com/pastebin/ed77f100-31f5-4550-aaa0-84958a9e6d8d
MSmits: oh nvm, no workie
jacek: and probably wont until the end of chat
MSmits: sigh. Seems like my bugfixed bot gets 57% WR vs old bot, but I can't submit now :(
jacek: hm?
struct: cooldown?
MSmits: the leaderboard is meaningless
MSmits: because of gold opening
struct: why?
struct: too much rng?
MSmits: yeah
struct: ah
MSmits: have to wait till everyone who is offline is done submitting
MSmits: calms the leaderboard down
jacek: thats so green
jacek: maybe thats the way to win it
jacek: and leaderboard is a decoy
MSmits: who knows
miobyte: anyone here can help me with k8s?
Astrobytes: you asked that yesterday or the day before
jacek: is that a puzzle
kovi: it is harder to ascend when top is actually in middle of leaderboard
Nerchio: I am the top now hehe
Psyho: it gets easier when you get to the top ;)
Nerchio: it will take some time before people bring me down maybe I should leave it until friday
Nerchio: :sunglasses:
emh: Gold opened early?
struct: yes
struct: and I thought I passed because of the notification
struct: now I have to start coding again
emh: good luck
emh: me too if I am going to be sure to reach Legend
emh: when it opens
struct: Not sure I can make it to legend this time, fewer players and the top is too strong
Julian_Dixon: you ive read through and completed all excercises in an intro to python book and im still finding "the descent" impossible, is there somewhere i can get coding experience without already knowing how to code? like every time i try to get into it i get into this feedback loop of "well you just need to get better by doing it" but "I cant DO it"
struct: press hints on the left Julian_Dixon
struct: maybe you are just struggling to understand how inputs and outputs work
emh: Julian_Dixon study "algorithms" to get smarter after knowing how to code
jacek: programming aside, you need to familiarize yourself with inputs/outputs here. more advanced people struggle at first
Julian_Dixon: Im almost done with a math minor and a engineerig physics degree, I dont think my uderstanding of algorithms is the problem
struct: So which part do you think you are struggling with?
struct: the descent is turn based
struct: you get inputs from stdin each turn
Sarzorus: Yeah the level of study you get in algorithms at school isn't easily replaced with just doing it. Maybe in some cases you can, but in general it's very difficult without some background in implicit proofs, and/or knowing the general algorithm classes/types.
struct: and you must output to stdout
Julian_Dixon: somehow this is the first ive heard of stin and stdout. Ive already completed a whole computational physics course, how is that possible
Julian_Dixon: im pulling my hair out lol
Uljahn: maybe try to understand the solution in the hints on the left?
Julian_Dixon: Ill do that i guess thanks
Crosility: Howdy Everyone :]
Julian_Dixon: I think my problem is that coding requires a LOT of creativity and I have zero
wlesavo: well not for this puzzle
Crosility: Creativity? More like deductive reasoning, and problem solving. :]
Sheeesh---: everything will come with practice
Julian_Dixon: well i say that because theres 12 ways to write a program that does the same thing yknow?
Crosility: Do first, optimize later (once you've obtained practice and skills)
struct: you dont need to overthink that puzzle
Julian_Dixon: and @Sheeesh--- that where i get confused, because to practice coding, you have to know how to code something more than like a prime number checker
Crosility: Julian, I think you'd benefit from watching a CS50 course on youtube.
Julian_Dixon: That sounds like a good idea
wlesavo: if you have basic understanding of conditions and loops you can practice on cg
Sarzorus: Lol in our class on state machines we went through a proof that any program has an uncountable infinite ways to write it. However most statements can be boiled down to a minimum working version which isn't creativity just reasoning.
jacek: there are so many ways to do sorting
jacek: you can even dance https://www.youtube.com/watch?v=3San3uKKHgg
Sheeesh---: LOL
Sheeesh---: made my day
Julian_Dixon: what. did i just watch
Julian_Dixon: I mean intuitively i get it but that doesnt help me understand how to translate it to computer language
ilgiocatore: quick sort is a pretty advanced algorithm, i guess
Astrobytes: for simple problems, writing it out on paper/whiteboard and solving a few cases by hand is useful, you can then translate that to code
Crosility: Quick sort has a fatal flaw.
Westicles: programming isn't for everyone. somebody has to make the pizzas
Julian_Dixon: is it that you need to know how to dance?? haha
Crosility: It sucks at sorting already sorted lists/mostly sorted lists.
Crosility: But yes, need to learn to dance. :P
Sheeesh---: i guess most of the time the list is not sorted
jacek: most sorting used are mix of sort algorithm with analyzing of partially sorted sublists
jacek: like https://en.wikipedia.org/wiki/Timsort
Sheeesh---: oh i see
JohnyDaison: jacek what the hell is that video? :D who are those people? how did they stay serious throughout that?? :D
jacek: these are CG servers behind the scenes
AlitorTheWise: Hello, can I talk with the people in a clash od code? I mean, I supose to, but how can I find the room to do it? Tx
JohnyDaison: Do you see other tabs than World above this chat?
jacek: it should add automatically once clash starts
jacek: or before that (?)
Noyotens: any advices?https://www.codingame.com/replay/644632231
struct: you got 4 tech on first release
struct: doesnt seem like a great idea
Illedan: :scream:
struct: also on 2nd and 3rd release
struct: seems like you like to collect tech
struct: :p
jacek: tech?
JohnyDaison: tech debt
jacek: oh
eulerscheZahl: did nerchio improve so much or is the leaderboard still out of place?
eulerscheZahl: in any case: don't touch it until legend opens
jacek: now im gonna touch it
Psyho: it's out of place by a lot
Psyho: everyone that submitted right after fix is very high, everyone that didn't (like duck) is very low
struct: What is the draw rate between top bots?
struct: in 2 matches I mean
Psyho: dunno, but I'd imagine draws are very rare (2% at most?), it's improbable to tie on debt
Psyho: oh
Psyho: that's not what you asked about
struct: yeah its a bit diferent sory
struct: sorry*
Psyho: it's hard to tell, because the top bots are in the middle of the leaderboard
struct: ah ok
jacek: the 2 matches have the same seed/
Psyho: but surprisingly p1 doesn't have enormous advantage
Psyho: MOVE 5 and MOVE 2 first are both decent options
Psyho: knowing what your enemy is going for also gives you p2 a small edge
lifetimeLearner007: TRAINING (0). The team draws 2 cards How do I know which 2 cards I'm going to draw?
struct: you don't
struct: unless there are only 2 cards to draw
struct: first it draws cards from the draw pile, if its empty then it will draw from the discard pile
Nerchio: eulerscheZahl wow rude :joy:
lifetimeLearner007: if it is not known which cards will be drawn, how do we look ahead to find best action?
struct: You take into account probability
jacek: :scream:
struct: I managed to do it, so I don't think it's very hard
ilgiocatore: most players just don't use training and coding :relaxed:
lifetimeLearner007: it seems like it is better to not use training and coding rather than spend time on handling probability(which is not 100% accurate)
ilgiocatore: although i use coding, but only for extra move
ilgiocatore: if i have multiple useful cards in hand
lifetimeLearner007: also, how do we know when the phase is going to change and to what phase? If I were to simulate, how do I know?
struct: well if you move and go over admin table then the next phase is thor
struct: throw*
struct: if you land near opponent then its give
struct: and so on
struct: you need to take that into account when you create the sim
lifetimeLearner007: hmmm. This game has more unknowns than the previous games, or I haven't understood the game completely
struct: you can predict the next phase
struct: until it reaches opponent turn
ilgiocatore: phases are within one turn: move -> [throw] -> [give] -> play -> release
you know everything, because you draw new cards from your deck only before "move" phase
lifetimeLearner007: yes but the movement from discard pile to draw pile to in hand is not deterministic right?
darkhorse64: it is. All discard goes to draw when draw is empty. Draw to hand is random
lifetimeLearner007: what matters most is random,. haha
Nerchio: life is random so get used to it
Illedan: Educated random
emh: took a sleeping pill yesterday to come down from my programming high, but it lasts too long so I slept all day. now I want energy drink but it's too late in the evening
MSmits: I've never taken a sleeping pill
emh: well it's also against schizo
emh: but it makes me sleep
MSmits: for a second I thought you meant to suggest I should take it for schizo
emh: haha
MSmits: :P
MSmits: I try to hardcode openings but it makes my bot worse
emh: I've a list of prioritized openings
Illedan: This skewed placement is a pain to test against
emh: mostly 2 and 5
MSmits: yeah i try to hardcode 2 as p2
struct: MSmits still enough time for meta mcts
MSmits: but it made my bot 40% WR vs old bot
struct: :D
MSmits: heh yeah, i actually did consider that :)
Illedan: Just do mcts :P
MSmits: still am
jacek: that would be so green
struct: 4 is the worst opening for p1 right?
Illedan: nah
struct: You are forced to either give or go to admin next move
jacek: 3 3 is best opening
emh: once you have enough heuristics in place the game gets too complicated to add more, for example taking into account probabilities of draw pile. the combinations of things to think about grow and there are so many considerations worth little each
MSmits: yeah
emh: struct hmm that reminds me, I have a priority to go from 2 to 4, as I can still get CI and block opponent from it, but I didn't think about having to go to admin after if opponent moves 2 steps ahead
MSmits: this is why i do a sim, but when my bot starts the game, it behaves weirdly.
Illedan: How do you search it?
MSmits: brute force, i just do a dfs with eval
MSmits: for my turn
struct: and probability
MSmits: yeah
struct: when he draws using training/coding
MSmits: it's kinda hard to deal with architecture and daily on a search like that
MSmits: because they dont provide benefit within the turn
MSmits: I just give them a score, but nothing i do makes my bot use daily. I am not sure if i have a bug there
BrunoFelthes: I just add score to daily at first turns... After it, i only use, if i have nothing more to do...
MSmits: yeah i think that is what my bot does too
MSmits: well except for the first turn
MSmits: i cant make it go to 2 unless i force my bot
MSmits: my p2 bot goes straight to refactor
MSmits: which seems completely nuts as the admin desk is next
MSmits: but on the other hand, p1 is force to move to coding
Illedan: Too high scoring on losing debt
Illedan: I had that too
MSmits: on having debt you mean
Illedan: ye
MSmits: but yes, i guess so
MSmits: i wonder if it is bad though
MSmits: you force p1 to make a big jump as well
emh: I don't worry much about debt
emh: top players generally end with lots of debt. at least in Silver it was like that
emh: or a while ago
jacek: becuase they were using 1st action
emh: my priority is releasing early and often. but not too early, as I want to get my CI going
Nerchio: damn i had a bug where I miscalculated penalty for card giveaway
Nerchio: basically the penalty was on every move no matter what lol
MSmits: ouch
Illedan: Hard to understand progress here
MSmits: yeah i tis
Illedan: 1 random move different and I win / lose
Nerchio: there's a lot of randomness in the draw, people literally drawing all "good" cards when their draw deck is full of debt
MSmits: well thats a 1 turn benefit
Nerchio: so just search for your bad moves and don't look at the result I guess
MSmits: next turn you will get all your TD :)
Nerchio: the games end very fast MSmits
struct: Does the game always starts with 12 apps?
Nerchio: so 1 lucky turn is often all you need
MSmits: true, this is a problem on the last turn mostly
miobyte: can anyone help me with k8s?
BrunoFelthes: Nerchio, do you try to block the opponent move?
Nerchio: not yet but I have it written down
Nerchio: and it would mostly be against people who go for 30 tech debt wins lol
BrunoFelthes: i tried, but it got worst
Nerchio: yeah it's probably tricky to get it right
MSmits: BrunoFelthes got worse for me as well
emh: BrunoFelthes I block opponent, but only if he has only one choice left for his only automated skill card. doesn't make sense to block if he's got options
Illedan: 1 option left?
emh: one application which needs the skill
emh: of course he have other options, but not with that skill
Nerchio: one application to rule them all
emh: and in the CI bind them
TINOUAINANI: Hi, is there a trick to print all errors with out limit in the console ?
eulerscheZahl: online it will always be truncated
Dapps: Damn.. guess I've gone as far as my if/else skills will take me. Can't get higher than top 20 silver
struct: My vs is so broken
jacek: if second player releases 5th app, its over no matter what?
struct: ys
struct: yes*
Kellthazar: my bot was running only with my "MOVE" phase logic... duh!
JohnyDaison: struct, but there should be one more turn for the first to release, right?
IvesL: what makes the game so little participants?
jacek: it was announce a week before
Nerchio: I didn't sign up on 3000 acounts this time
struct: No JohnyDaison because both players played the same amount of turns
JohnyDaison: that's a bit counter-intuitive, but ok
struct: not really, it would probably help p1 if it was the case
struct: He could just keep lowering his technical debt
TINOUAINANI: is it possible to simulate all possible outcomes including TRAINING and CODING in less than 50 ms?
struct: No
BrunoFelthes: for 1 turn, yes, if you hash things
Illedan: And if you throw all cards first
struct: even if you have 5 training and 5 coding?
struct: He did say all possible
BrunoFelthes: yes, because, the result, will be the same
BrunoFelthes: just hash things, i said yes, i didnt say that is easy
BrunoFelthes: because order do not make difference
struct: Maybe I'm thinking of something different
MSmits: BrunoFelthes with probability it may make a difference
MSmits: the order
BrunoFelthes: some orders yes, but others no
MSmits: true, i just enforce orders when it doesn't matter
MSmits: so i dont use hashing atm
MSmits: but i also limit to 3x training/coding per turn
BrunoFelthes: but i think that is it possible, it is just hard to do
MSmits: so i might use hashing soon
struct: Yes draws have to be limited
struct: I dont think its feasible in 50ms to calculate all states possible
MSmits: well it is possible for most situations
MSmits: it's just not possible for a rare few
BrunoFelthes: What i do, is to random the draft and test possible things
MSmits: random the draft after the turn?
MSmits: or when you train/code?
BrunoFelthes: we are talking about the same turn
MSmits: oh ok
BrunoFelthes: train/code
MSmits: I don't random, i loop over all possible train/code result and weigh the eval by the probability of the draw
BrunoFelthes: and can you run at all of the states?
MSmits: yeah if i limit to 3 plays in 1 turn. Could do 4 in most cases, probably 4 or 5 if i do some more optimization
MSmits: i might switch to iterative deepening for this.
Illedan: The randomness of these games :scream;
BrunoFelthes: oh... i do a diferent aproach...
struct: thats what I meant on all possible :p
MSmits: I'll explain how to do it on monday
MSmits: i mean i kinda did now
MSmits: but i'll share code
Psyho: so you only evaluate after a single move?
MSmits: after a full turn, but yes, there is 1 move in there
MSmits: i dont know how to go on after this
BrunoFelthes: do you evaluate the entire game psyho?
Psyho: yep
BrunoFelthes: god
MSmits: that's amazing
JohnyDaison: Psyho, but you are making assumptions based on Math.Random(), right?
MSmits: you can't brute force that, that has to have loads of random
BrunoFelthes: with train/cod?
ilgiocatore: the next step is to evaluate 5 games ahead
Psyho: funnily enough, I believe I explained my approach at the end of the stream
Psyho: and did exactly that
MSmits: there is a strea,?
MSmits: m?
eulerscheZahl: https://www.twitch.tv/FakePsyho
BrunoFelthes: I'm analising Psyho games, he do not have a lot draw pile
Psyho: I did stream my first day of the contest
MSmits: oh ok
BrunoFelthes: maybe it is the secret
Psyho: it's just monte carlo with full games with heuristic strategy for both players
Illedan: I do the full game too, but I set a cutoff to not go too far into nothing
Psyho: and eval is based on win%
MSmits: I see
BrunoFelthes: nice
eulerscheZahl: and that works? i'm impressed
Psyho: I thought it's impossible to write good midstate eval for this game
MSmits: so ordinary monte carlo? No mcts?
MSmits: i should say, ucb guided mc
MSmits: not mcts
BrunoFelthes: what is the difference between mcts and monte carlo?
struct: monte carlo doesnt remember
jacek: ts
struct: mcts stores info
Psyho: funnily enough, I never did implement MCTS, but I don't see how it can be done here
BrunoFelthes: i see
MSmits: well mcts is the wrong choice as it is adversarial, but for your turn you have several phases in a row, you could use UCB to guide your decisions within your turn
struct: its a bit more detailed than that
jacek: you can do UCB at first level
Psyho: after your turn is done or you play coding/training the branching factor is too big
Psyho: and you need a lot of games for prediction
Smekarn: Impressive lead @Psyco, nice!
eulerscheZahl: any highly speed-optimized simulation at least?
eulerscheZahl: let's wait for the duck to resubmit
Psyho: nothing fancy
struct: If you want I can share with you my MCTS code for one of my bots BrunoFelthes, after the contest ends
Psyho: if you expand the full tree until the end of the turn/coding/training you can just eval all of the nodes
Psyho: so no need for MCTS
Psyho: i.e. you get enough sims
MSmits: yeah, BrunoFelthes just dont imagine using mcts very often on these contests, its useful for simple board games mostly
MSmits: Psyho if you eval the nodes, you dont have the future info like you do on a full game MC
Psyho: Smekarn, oh yeah I doubt I should be first now, the ranking is broken
Psyho: yeah, I just think it's impossible here to make a good solution based on eval
MSmits: i almost agree with that
MSmits: it's possible to make a good solution based on eval
struct: So I should be doing better, since i dont know how to write evals
MSmits: just not a great solution
Psyho: ideal solution is RL where you update the model via rollouts
jacek: :thinking:
Psyho: (rollout = MC in RL terms)
Psyho: (I'm dumping the whole knowledge know becaus it's essentially the last day of competition for me, except for running local league system on a VM that tries to find better params for the MC (heuristic strategy in sims)
MSmits: Psyho i appreciate you sharing a lot, but for your own benefit it is better to share on monday
eulerscheZahl: and then you will be placing stones with increasing numbers on a grid instead
MSmits: even if it is your last day coding
MSmits: I share a lot but usually not my whole search. Only reason I shared my brute force here is because euler already did
eulerscheZahl: i did?
MSmits: yeah before me
MSmits: i figured, well cat is out of the bag anyway
eulerscheZahl: but it's so plain stupid, no one will gain anything with it
struct: I will
MSmits: yeah maybe, sometimes proof of concept is all people need to imitate
struct: just wiat
struct: wait*
Psyho: tbf, if euler didn't share his brutaltester jar I wouldn't be even close to that
Psyho: *to the top
MSmits: oh yeah, he does that :)
eulerscheZahl: oh wow, you trusted a relative stranger to execute that code
eulerscheZahl: i feel honored
Kellthazar: hahahaha
Psyho: it's java, what harm could it do
Illedan: xD
Noyotens: boss 4 has not provided an action in time.
Noyotens: why?
jacek: w00t
struct: Maybe it breaks at this exact time
MSmits: Noyotens too slow
Psyho: the thing that I find the most shocking is that juju is 2nd with python
jacek: dont they see the source code?
MSmits: mmh my code would run in python also and not timeout i think
kovi: either already NN or very nice set of heuristics
MSmits: but i am not 2nd :)
Noyotens: MSmits, the bot is slow
MSmits: yeah
MSmits: apparently :)
MSmits: or just a bug in the bot
Noyotens: i changed the seed & my code but its still broken
Noyotens: ?
MSmits: ah, no idea
struct: well its working for me
MSmits: is it rust?
Noyotens: nope, c#
MSmits: then i got nuthin'
struct: you cant break bots anymore with rust
MSmits: maybe you haven't tried hard enough
struct: you can but it will also break your arena one
struct: Do I even need to store all apps?
MSmits: what do you mean?
MSmits: in your gamestate?
struct: Yeah
MSmits: no. Store a minimum of 12 bits that flip when an app is finished
MSmits: or if you're lazy like i have been, a bool[12]
struct: but you still need to for each app
struct: if you sum all apps together you dont need to loop
struct: ah doesnt work
struct: im dumb
MSmits: well
struct: How do I delete a message?
MSmits: actually you can get away with 1 integer
MSmits: just set it to -1 for no app finished
MSmits: and then to 3 if you finished app 3 etc.
struct: I think ill do the 12 bit thing
MSmits: works too
struct: Thanks MSmits
MSmits: np
Nerchio: the duck is slowly climbing since gold opened lol
jacek: he didnt submit since fix?
Nerchio: nah
Scarfield: what was fixed?
struct: Gold opened with alpha version
Scarfield: so same version in silver?
eulerscheZahl: bugs. french statement only. other input format
Scarfield: ah nothing "changed" then, but gold opened with wrong version and that got fixed?
struct: it will only change in legend
Scarfield: what, so gold has a different format?
eulerscheZahl: no, league opening was bugged. but now it's all good
Illedan: Losing a game 5-2 in Last battles and winning 5-1 in IDE -.-
struct: cant relate
Nerchio: cant relate I always win in and outside of IDE
JohnyDaison: Illedan if you're testing against Psyho you will almost never get the same game even with the same seed
emh: I succumbed to desire. energy drink it is. maybe it will be a coding night
eulerscheZahl: illedan might be random himself
eulerscheZahl: i know him well enough to expect exactly that
Ay1man1: how do you xor two numbers wtf\
Psyho: "An error occurred (#407): "You reached the limit of plays for a period of time."." damn
Ay1man1: just had a coding clash i could not understand
Psyho: so apparently 20 games is the limit in short time
jacek: Ay1man1 a^b
jacek: or make NN for that
Ay1man1: like a to the power of b?
struct: it will get lower and lower Psyho and cooldown will increase
jacek: ^ in most languages is xor
struct: if it keeps getting spammed
elderlybeginner: What are "automated" cards in the contest?
eulerscheZahl: there are many play limits for time spans up to 24h
Illedan: eulerscheZahl, I'm full random :P
putibuzu: elderlybeginner basically cards that you put next to your hand for the rest of the game and can use them for releasing apps, but you cannot play them for the effect anymore
JohnyDaison: Illedan, are you serious or not? If you are 47th with RANDOM, something is very wrong.
Illedan: ..
struct: he doesnt mean output random
eulerscheZahl: ̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤
MSmits: noooo
struct: ..
Illedan: I do rollouts like psyh o
MSmits: onl no heuristics
eulerscheZahl: Illedan provoked me
Illedan: But fully random
MSmits: just random moves
IroncladDev: what in the world are all these lines?
struct: So its just normal monte carlo?
MSmits: should be easy to transition from your method to psyho's
MSmits: IroncladDev they are euler's sick sense of humor :P
Illedan: Yeah, 2 lines of code. It worked ok, but my heuristic is not as good
IroncladDev: interesting
Illedan: I have heuristic for the enemy
eulerscheZahl: this alone is a reason to keep the chat. where else should i spam ̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤̤
mlomb: how many rollouts are you getting? :eyes:
Illedan: 500 :D
mlomb: nice
eulerscheZahl: do you simulate solo-play or with the opponent moving too?
kovi: you run till end of game
kovi: ?
Psyho: wait you get 500 full games in 50ms with just random moves?
jacek: Automaton2000 did you see that?
Automaton2000: did you get to the same state
Psyho: 500 is not a lot
struct: 500 seems low doesnt it?
Illedan: I know :P
struct: ok
Illedan: Very low
PatrickMcGinnisII: spam chain Automaton2000
Psyho: veryyyyy lowwwww
Automaton2000: nothing to do with it?
Psyho: I get few times more and my heuristic is the bottleneck
MSmits: Illedan when you do the movegenerator don't make all sorts of temporary objects like Lists and such
Illedan: Nah
Illedan: I count all spells that I can do
Illedan: Random on the amount
Illedan: Then count again until I find the one I wanted
MSmits: I see
MSmits: weird that it is so slow
MSmits: it is C#, but still
Illedan: I didn't optimize it yet
Illedan: Just got stuff working last night
MSmits: ah ok, i didnt optimize mine still and i got it working on sunday :P
kovi: thought you did 1 move only?
Illedan: Kid + work + wedding soon doesn't help on my time schedule :P
MSmits: i turn yes, but thats a move, throw, give, x plays and a release, bruteforced
eulerscheZahl: and you can't just ditch one of those 3 without affecting the others
eulerscheZahl: kid for adoption but wedding still going won't work
MSmits: is the wedding because of the kid?
Illedan: Quit work? Then I can't afford wedding?
JohnyDaison: Hey Psyho, I just got your bot to time out
MSmits: JohnyDaison if people run close to the timelimit, some inaccuracies make you occasionally go over. It might not be a bug
Psyho: thanks for the info, unfortunately it's not reproducible
MSmits: you usually leave a 5ms buffer or something
MSmits: his buffer could be too small
Psyho: I'm using 45 ms
MSmits: yeah sometimes that could not be enough
MSmits: very rarely
Psyho: but there's also a much it just crashed
Psyho: *a chance
MSmits: yeah hard to say
MSmits: sometimes it outputs the time even when you are out of time and you can tell
MSmits: say you used 52 ms or something
struct: i usually leave a 0.5ms buffer
Psyho: I don't think I ever use more than 46ms, so my bet is on crashing
MSmits: structed i just compared a 2 play limited bot to a 3 play limited bot. It had a WR of 40%
MSmits: struct i mean
MSmits: so the play limit does affect my game
eulerscheZahl: there are these dots below the viewer that let you see your logs. if you are lucky, it points you to a line in your code https://i.imgur.com/7ctQkjt.png
struct: maybe the type of timer could be difeferent
Psyho: oh btw about that: is there any easy way of getting the log of all of the games that I played?
Psyho: seems like the easiest way to find crashes
struct: I use high_resolution_clock
eulerscheZahl: you have to write your own scraper
Psyho: damn
eulerscheZahl: download all replays, check for negative score
Illedan: You have to script it Psyho. Because only you can see your debug printing
Psyho: don't you have a jar for that mr jar dealer? :D
Astrobytes: lol
struct: Does your tech io tutorial still works eulerscheZahl?
eulerscheZahl: we agreed not to share such software as we fear that CG will limit it when it gets too popular
Psyho: makes sense
eulerscheZahl: https://www.codingame.com/playgrounds/53705/contest-tools-and-workflow/introduction
eulerscheZahl: a quick how-to with some code. but nothing ready to use
eulerscheZahl: the play limit you hit earlier is a result of automated testing taking too much resources
struct: books -_-
Psyho: thanks, that's actually more than enough
Psyho: and yay for python instead of java
jacek: jython?
eulerscheZahl: pyava - to combine the worst of both worlds
Illedan: pyava++?
Psyho: speed of python with flexibility of java? :D
eulerscheZahl: the enterprise slow language
MSmits: it's not as good as Roofie
MSmits: (Groovy + Ruby)
Astrobytes: where did the 'fie' come from?
MSmits: it sounds the same
Astrobytes: v != f
MSmits: pfft details
Astrobytes: *pvvt
MSmits: :)
Astrobytes: heh
MSmits: kovi did you end up bitboarding this?
kovi: not yet
MSmits: I am trying to think of an effective way, but it doesnt even fit in int32
MSmits: a hand
kovi: im also still 1 move only, no speed problem
MSmits: did you not limit your plays?
MSmits: the limit hurts me a bit
kovi: i use heuristics to cut it
MSmits: oh ok, i just limited it to 3. If i lower the limit to 2, my bot gets 40% WR so it's significant. Not sure if a limit of 4 would be much better
MSmits: i should just do it iteratively. start at 3, then try 4, 5 etc.
MSmits: most turns it should just hit the max np
struct: Nice I made it to gold :D
MSmits: welcome struct :)
JohnyDaison: welcome
struct: Thanks
struct: I just add to open euler referee for brutaltester
struct: changed 1 param and got 70%, changed another and got 70% vs the previous 2
MSmits: nice
struct: wow psy has a huge lead
BrunoFelthes: full game search
struct: maybe I should start going further than 1 round
MSmits: you could, but you'd still be limited to eval. Psy ho's method is using the endgame result as eval, thereby circumventing the need for eval
Psyho: I mean, I wouldn't say the only my method is the way to go
Psyho: I had the same approach two days ago
Psyho: it's just that this morning I set up a local league system + some semi-automated way of parameter search and I'm 60K of local games later
struct: well time to rent a vm again
Psyho: and that messed up leaderboard makes it look like I have a huge lead (honestly I'm confused at this point why I still have a 3 point lead)
Psyho: duck is finally in top 10 though
struct: maybe we should wait for duck submit
Nerchio: yea but he didnt even submit xD
Psyho: I have a version that's ~0.9 higher locally, but I don't want to ruin my unfair lead :(
MSmits: There's a duck on my tails!
emh: I wonder if I should write a new bot with sim (in C++) but I'm too chill at the moment
MSmits: more caffeine
emh: hehe I'm already immune to caffeine without taking a break I suspect MSmits. mostly. could be that 0.5l more of energy drink would help
struct: he just drank 10 coffees worth of caffeine
emh: 160 mg in one can
MSmits: i cant take much of that
struct: I dont have good experiences with energy drinks
Nerchio: i stopped drinking coffee and I have more energy than before so there's that
emh: I stopped drinking coffee also because I don't feel like the taste, and also because it's acidic for stomach. well I don't know if energy drink is any better for stomach but at least I love the sugar taste
Astrobytes: energy drinks are terrible for the stomach. I stopped more than 1 or 2 cups a day of coffee as my pain meds wreck my stomach enough as it is
Illedan: The play limit is annoying -.-
Astrobytes: Very.
Illedan: Using submits to tests small changes
Illedan: -.-
struct: havent reach it yet
struct: reached*
struct: I think its mod benefit
Illedan: riiight
struct: :p
Astrobytes: lol
Kellthazar: does the play limit counts for tests in arena only?
Sheeesh---: what's the distance between desks ?
struct: any arena
Sheeesh---: how to know*
Illedan: 42 meters Sheeesh---
struct: I simply do t2-t1
Illedan: And the pawn walk at 6 km/h speed
Sheeesh---: how about velocity ?
jacek: -3vel
jacek: :tada:
Sheeesh---: thank you !
Quidome: how to get out of wood 2??
Sheeesh---: use if statement
Razovsky: lets go Snef
Quidome: I always beat the boss and can't get high enough
BrunoFelthes: first valid move Quidome
Quidome: really??
BrunoFelthes: try
Quidome: hm, thnxs, will try that
Snef: Razovsky let's hope i don't drop
struct: Quidome first move got people up to silver
struct: but its probably not possible anymore
jacek: now its 2nd action
Quidome: ok, i am on it
BrunoFelthes: really? do they changed it?
BrunoFelthes: first move to 5, second to CI, help a lot to
BrunoFelthes: always that is possible to do a move to 5 and a Cont Int 8, do it... and then, random
Quidome: servers are really fast this time, nice!
Quidome: first move is not working
Quidome: Am i being trolled here??
Astrobytes: no, it really was a thing
jacek: i would never troll
Quidome: really annoying, i looks as too much random to me
Quidome: jacek, sorry
jacek: :upside_down:
antiwonto: [auto] ':upside_down:' was defined as ' ɥǝllo '
Quidome: can't think of a simple strategy to get out of there
Quidome: second move also no good
jacek: there is some boss code on github, isnt it
Quidome: I beat the boss, but I am not getting a high enough ranking
ilgiocatore: Quidome automate bonus cards as much as you can
BrunoFelthes: move to 5, 6. CI 8. otherwise random?
BrunoFelthes: CR too
Quidome: what do you mean by automate bonus cards
BrunoFelthes: yes, stack automated bonus
Quidome: is that a rule in wood 2?
BrunoFelthes: no?
ilgiocatore: oh
Quidome: ok i am gonna read again...
BrunoFelthes: i really do not remember, i pass using random moves :P
ilgiocatore: yeah, in wood 2 you can only move and release?
BrunoFelthes: me
BrunoFelthes: in wood 2 you need to give card?
BrunoFelthes: never give a card
Quidome: only move and release indeed
Quidome: :rage:
BrunoFelthes: so, never wait, just move random, and release random... span the tests
Illedan: Move (position+1)%8 Count Cards in hand, release best option
BrunoFelthes: you go to gold that way
Illedan: Silver, if you do random on the skill turn :D
Illedan: https://grugbrain.dev/
Michael_Howard: Wow, did they just add a Gold server?
Nerchio: you could also bruteforce all possible positions until the end of the game and evaluate them somehow
Quidome: is there some strategy in this game when you have all te rules?
Astrobytes: darkhorse64: grats, that was close huh
Illedan: Nerchio, evaluate is not a problem when you get that far .P
darkhorse64: small margin but it was my first try farming CI
Illedan: From that webpage ^ danger, however, is agile shaman! many, many shiney rock lost to agile shaman!
Michael_Howard: I hope the servers are this fast on Sunday :rocket:
Illedan: Lol
Illedan: No chance
Illedan: 800 submits pending all day
Illedan: Hmm, my code is better without the enemy -.- Makes no sense as he now just stands still somewhere
jacek: good, less states to eval
Ahmed_King: hello
Ahmed_King: https://www.codingame.com/clashofcode/clash/247347619a70352dd287be813b96c10204a1f9d join this
antiwonto: [auto] hey Ahmed_King dont paste those links here. Use the channel #clash
Quidome: after a few submits with the same code I am now at position 6. Variation roughly 259 places...
Quidome: Is it better to wait or keep on submitting?
Illedan: Better to fix the code
Quidome: I am talking wood 2 here randomness is the issue here
Quidome: well, I am gonna sleep over it, it's very frustrating never had is in a previous contest
Quidome: gl
MSmits: it is an obstacle you can get past
MSmits: gn
Michael_Howard: Server at normal speed again. Were they just testing something before?
Michael_Howard: They'd upped the number of battles too.
Michael_Howard: It was like several hundred.
Westicles: so I get this big list of choices to release, is it important which one?
MSmits: yes
MSmits: well it's all relative
MSmits: it's not the most important thing
MSmits: but your deck and the available apps affect the probability of you being able to release in the future
Westicles: I don't think I can figure it out without reading part of the statement
MSmits: you need to read the entire thing and then question everything
Illedan: I got a 5 min youtube video which explains everything
Illedan: For the boardgame
Illedan: far more logical
Westicles: ok let's see it
PatrickMcGinnisII: Westicles minimize tech_debt acquisition for release...wait is also an option
Psyho: rulebook is online for samsara: https://boardgamegeek.com/filepage/162892/samsara-game-rules
Illedan: https://www.youtube.com/watch?v=X1-07vdRFVw
Illedan: Glad we don't have the abilities on the characters as the game has
Psyho: honestly the game is really meh, I find it hard to comprehend that it has a decent score on BGG
Illedan: Yeah, too simple for my taste in board games
Psyho: it's very unintuitive for humans, endgame is very simulation-based, rules are overcomplicated, there isn't too much depth, every game is similar, there aren't too many strategies
Psyho: imho it doesn't work as a casual game (too complex rules) nor as a competitive games (lack of depth)
Westicles: every time you use your new phone it'll put you in a bad mood
j4at: its pretty much a casual luck game.
JohnyDaison: You know what would be really useful? If it showed in history for each version what rank it was before it was replaced.
BrunoFelthes: game history could have a lot of improvements
BrunoFelthes: but there is an extension that you can add a text for each history
PatrickMcGinnisII: omg, i beat silver boss, but it timed out! wth
antiwonto: [auto] Hey PatrickMcGinnisII, here is a :taco: for loggin in today while it is quiet. You now have 61 tacos
struct: 4 days to get to legend
struct: hi
Psyho: isn't it 4 days till the end of the contest?
gjin25: I'm new and i have start the green circle competition
j4at: 4 days and 20 hours and 30 minutes
j4at: Psycho
struct: yes Psyho but I need all the time I have to reach legend :p
Psyho: 🤣
Psyho: that's the ugliest ROFL emoji I have ever seen!
struct: lol
struct: but it describes it perfectly
Crosility: :fearful:
NotSureWhyThisWorks: can someone explain what happens when you play a daily routine card?
NotSureWhyThisWorks: "the team can get their new skill card from a desk one step away"
struct: When you move
NotSureWhyThisWorks: do they get to choose?
BrunoFelthes: at the next move
struct: You can do MOVE 6 5
struct: to move to 6 and get card on cell 5
NotSureWhyThisWorks: oh ok and so the more you have the further the distance away from your actual location you can take a card?
struct: yes
NotSureWhyThisWorks: awesome thanks
struct: But it resets to 0 when you release an app
NotSureWhyThisWorks: yep
PatrickMcGinnisII: can't get above 79th in silver atm
PatrickMcGinnisII: need break
BrunoFelthes: why there is a daily routine count?
PatrickMcGinnisII: laterz
struct: what do you mean bruno?
BrunoFelthes: you can stack daily routines
struct: yeah
BrunoFelthes: at the tooltip there is a daily routines number
BrunoFelthes: you said that its reset to zero
BrunoFelthes: is that true?
struct: when you release an app
struct: ye
struct: " it stays active until the team has released one application."
BrunoFelthes: so, why there is a number, not a boolean?
struct: same for architecture
struct: because you can stack it
BrunoFelthes: no, architecture is a stack
struct: if you use daily routine 2 times
struct: you can take from 2 cells away
BrunoFelthes: ok
Westicles: There is only on puzzle with two approvals, and it is a hardcore combinational math one you'll never solve without googling
antiwonto: [auto] Hey Westicles, here is a :taco: for loggin in today while it is quiet. You now have 113 tacos
Westicles: https://www.codingame.com/contribute/view/21267833d34ef960c94d60f4bb5c5f70f7926
emh: Westicles I understand the task but not the solution. should I approve it then?
emh: I need to approve something for my collaboration level
Westicles: It is up to you. I think it is pretty safe to approve, I came up with the same solution with a different approach
emh: ohh.. there is a checkbox that I came up with a working solution.. can't say that I did. so no
Westicles: good form
emh: is there something easy I could solve and approve?
Westicles: yes!
Astrobytes: very math-y puzzle indeed.
Westicles: https://www.codingame.com/contribute/view/22125008becf442406a264e2a0b4950fbd3af
Astrobytes: only you could have come up with that :D
Westicles: heh heh
emh: when I was 10 years old I ran after the girls in school and tried to kiss them
Astrobytes: I know someone who did worse than that, me and another mate reminded him at a party once and he didn't speak to us for a year :D
elderlybeginner: what are permanent skills for (contest)?
j4at: Ow I found a way the check the type of incoming clash :o
Astrobytes: they're stackable skills, they stack until you release an app. e.g. daily routine can let you do a move n m, where m == daily_routine_count
Westicles: j4at in time to leave before it starts?
Astrobytes: and architecture study gives you m cards at the start of the turn
j4at: you don't even need to join to check the type
Westicles: wow, you'll be a hero, you should post it on the forum
j4at: hehe
Astrobytes: doesn't sound high priority to me :P
elderlybeginner: It's my intellectual fitness decreasing or the contest is lousy described
Astrobytes: it's lousily described
Westicles: it is a reverse contest, you shouldn't read it. though I'm having trouble getting into gold
elderlybeginner: Astrobytes, why not both :wink:
Astrobytes: lol, well after looking at it for some time it's possible that intellectual fitness may be somewhat diminished :D
Westicles: I'm doing move+1 if available, else first move in list. What should I add?
Astrobytes: -3*vel
Westicles: hey, thanks emh!
elderlybeginner: what's '-3*vel'?
emh: yw
Astrobytes: CSB trick elderlybeginner
Zandy156: if your only skill card in hand is a task prioritization, what play_card action can you do?
Astrobytes: I don't believe you can...
Zandy156: so only wait?
Astrobytes: I would presume so, yes
Zandy156: I thought it was supposed to be automatic if we had no choice.
Westicles: j4at, you aren't going to share the trick?
j4at: I'm going to write a script that shows the type of the clash ;)
Westicles: amazing what they do with scripts, I ran the one that scrapes all your puzzle solutions, took just over 3 hours
Zandy156: oh, you can trade bonus skill cards with task prioritization
Astrobytes: You can.
Zandy156: not sure if that's a good move
Astrobytes: try it and see, I don't know anything any more lol
oidrissi: Hey guys, any one made a scoring system for the possible moves? I want to implement one and am struggling a bit with the whole scoring part
Westicles: I can tell you always choosing the first one works >> better than choosing the second or last
Kellthazar: The better Golang solution is in rank 106, interesting.
Westicles: but that is probably a wait/random artifact
Kellthazar: More Rust competitors as well.
oidrissi: @Westicles Are you answering me or something else? ^^
Westicles: it isn't a direct answer to your question, but related I guess
Westicles: I suspect if someone answered your question it would be like 50 pages of details
oidrissi: Agreed, too much data to handle
elderlybeginner: Can you determine appox./mean value of play_card action?
elderlybeginner: e.g. coding is usually better then daily_rounine
elderlybeginner: kind of rank paly_card actions
therealbeef: oidrissi I think everyone is working on this. the winner will probably just have figured out the best move evaluation
therealbeef: mine is rubbish at the moment
**Rodrigo_the_coder slaps /flip around a bit with a large fishbot
Rodrigo_the_coder: (╯°□°)╯︵ ┻━┻