Chat:World/2020-08-08

From CG community
Revision as of 11:50, 15 June 2021 by Chat Log (talk | contribs) (Created page with "<img src=/a/45775905514269> Allis: Use any extension that lets you apply custom user CSS on a per-domain basis. <img src=/a/45775905514269> Allis: Then use the following CSS...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Allis: Use any extension that lets you apply custom user CSS on a per-domain basis.

Allis: Then use the following CSS rule to prevent the annoying popup: .editor-widget { display: none; }

Allis: Not having to deal with arrow key thieves is *chef's kiss*.

eulerscheZahl: i'm fine with the popup. But when I hit escape to close it again, the editor loses focus entirely

eulerscheZahl: on the other hand I rarely use the browser IDE so I don't really care

Gorbit99: now let's have a way that allows this only for coc Allis

Gorbit99: eulerscheZahl I guess you don't play as much coc as some of us do

eulerscheZahl: correct :P

eulerscheZahl: i would probably use the web IDE for clashes too

eulerscheZahl: especially for the easy ones

Andriamanitra: any extension that allows you to apply custom css also allows you to choose which url the custom css should be used in

Allis: Unfortunately, the URL is just .../ide/hash in all(?) cases, so that doesn't quite get there.

Gorbit99: well, not using the webide would basically equate to shooting yourself in the foot many times over

Gorbit99: btw, while I'm here

Allis: I think Andriamanitra was suggesting a way to have it be Clash-local.

Gorbit99: had this new forum post about coc putting me into the ide 13-ish seconds after the timer starts

Gorbit99: is coc synced in that way, or is it a luck based system

Allis: Yeah, I've experienced similar. Pretty rough.

Gorbit99: I'd be interested in finding that out, as some clashes don't even last 13 seconds

Allis: The timer started normally, but that could just have been a local effect.

Gorbit99: it was in a clash you were in too

Allis: That's quite an uphill battle. :P

Gorbit99: show off

Gorbit99: anyways, made a forum post

Allis: If it's any consolation, it wouldn't be especially hard for them to sync that sort of thing, so they probably do.

Gorbit99: although seems like since thibaud is on vacation or something, not many forum posts get answered regarding bugs

Allis: That said, it's impossible to know from our end, so I hope you get a response from someone at CG.

Gorbit99: well, we could test it

Gorbit99: although would be nice to know from staff

Allis: One of us kill our connection before the clash starts?

Allis: I feel like that might test something different.

Gorbit99: just the start time of two people in a clash

Gorbit99: if one starts at 5 seconds in, the other at 20, that's proof

eulerscheZahl: Andriamanitra 06:36AM any extension that allows you to apply custom css [...] The first thing that comes to my mind: stylish. Unfortunately it got bought and collects all kind of user information now

eulerscheZahl: Try GreaseMonkey / TamperMonkey

Gorbit99: I'm sure tampermonkey can do it, even with some coc detection thingies

eulerscheZahl: you can have wildcards in the URL at TamperMonkey

eulerscheZahl: and adding a style via JS shouldn't be that hard either (haven't tried)

Gorbit99: but the url is the same as a normal puzzle, isn't it?

eulerscheZahl: so you don't want to modify the IDE for normal puzzles?

eulerscheZahl: but sure, there must be an indicator one could use to detect CoC. Like the shortest/fastest mode shown at the top

Andriamanitra: i think normal puzzles are ide/puzzle/* and clash is just ide/*

eulerscheZahl: and then there are multiplayers

eulerscheZahl: also /puzzle/ide

Gorbit99: ide/puzzle

eulerscheZahl: yeah, that

Gorbit99: thankfully I got a simple shortest puzzle

Gorbit99: seems like coc url is /ide/hash, all other urls are /ide/puzzle/hash

eulerscheZahl: a regular clasher once said that there is a CoC taking 4 chars in shortest mode

Gorbit99: yep, I know that one

Gorbit99: and the shortest solution is in basj

Gorbit99: h*

eulerscheZahl: he also said that it's a typing contest who can submit the fatest

Gorbit99: haven't seen that one puzzle ever again since I heard valnykol complaining about it

Gorbit99: that was around 2 years ago

eulerscheZahl: clashes get rated by players. when they drop below a certain treshold they are removed

eulerscheZahl: that is a per-mode thing. So it can be a bad reverse but still fine for shortest

Gorbit99: really? thought it's a manual thing

Gorbit99: then I feel much better about rating bad reverses low

eulerscheZahl: they automated it now (maybe 1 year ago) even for community puzzles

Allis: Are you sure it was four characters? I can't think what Bashism could possibly solve a puzzle with that few.

Gorbit99: well, it's good they did

Allis: "wc -c" is five.

Gorbit99: it was something stupid

Allis: Well, of course.

Gorbit99: weird characters all the way through

Allis: For reference, eulerscheZahl, Stylus is Stylish without the suck.

eulerscheZahl: ah, didn't know

Allis: It's a hostile fork with good intentions.

Allis: I'm not sure we have a name for that.

Gorbit99: now I might have a tampermonkey script ready

Gorbit99: time to try it

jacek: happy Caturday

Gorbit99: the css trick isn't working well Allis

Gorbit99: it still doesn't let you move with arrow keys, it just hides the reason why

Allis: That's odd. Browser?

Gorbit99: chromium

Allis: Stylish?

Gorbit99: oh, I need that also?

Allis: Well, you need some way of applying custom CSS to a page.

Gorbit99: I just used tampermonkey to inject the css

Gorbit99: http://chat.codingame.com/pastebin/fa86b9f2-aa38-4687-bfc8-d774453458e0

Gorbit99: with this

Gorbit99: and it worked sorta, it does hide the popup

Allis: It must be that TamperMonkey isn't firing early enough, but I don't know the particulars.

Gorbit99: well, it just puts some css right into the style

Gorbit99: let's check with stylish

Allis: Sure, but the web is complicated.

Allis: JS+CSS means you can have a "nonexistent" element that nevertheless captures events and just generally has side effects.

Gorbit99: probably needs some tinkering in JS to get rid of this

Gorbit99: yep, same with stylish

Allis: That's very strange.

Allis: I can only report that it doesn't happen on my end, but that's not of much use to you.

Gorbit99: once I get the vim error fixed, I can not worry about this issue

Gorbit99: I'm slower without that either way

Default avatar.png HoangMrPhuc: :V

Default avatar.png HoangMrPhuc: we

Default avatar.png HoangMrPhuc: we

Default avatar.png HoangMrPhuc: hey

Default avatar.png HoangMrPhuc: hey

Gorbit99: hey hey

_-: what about stopping the language server?

Allis: That's definitely more up TamperMonkey's alley, but I'd sooner just MITM my connection to not serve up the JS at all.

Allis: I've generally had a better experience doing things that way than relying on TM getting there "in time".

Tobes2:


Tobes2: oops

Tobes2: http://chat.codingame.com/pastebin/da07280a-3257-4487-b04d-18d69f06d99f

jacek: oO

eulerscheZahl: happy Caturday jacek

Uljahn: happy Caturday jacek

jacek: *naked guy tickles wet pussy* https://img-9gag-fun.9cache.com/photo/aK71pBQ_460svav1.mp4

Allis: Why is it so long?

jrke: happy caturday :)

AntiSquid: cat that like water ?

AntiSquid: showering *

Default avatar.png HoangMrPhuc: hihi

dbdr: lol, in Go, if the opponent passes, I do my search thinking I'm them :D

Default avatar.png HoangMrPhuc: sadsa

Default avatar.png HoangMrPhuc: P:)

MSmits: yay 109/163 on breakthrough with no eval :P

MSmits: it always backpropagates 0

MSmits: 13 million root visits on turn 1, over 1 million on later turns

MSmits: basically, the only time the bot knows what to do is when it's solved the game, like 5 plies before the end :P

MSmits: no idea yet how to eval this

dbdr: better than random

MSmits: yeah

dbdr: I guess when my bot has identify issues, it's worse than random :D

MSmits: identify issues?

dbdr: in Go, if the opponent passes, I do my search thinking I'm them :D

MSmits: you identify your opponent?

dbdr: identity

MSmits: ohh ok

MSmits: that can happen in Othello as well

MSmits: it's messy

dbdr: the fix is not hard

MSmits: no, but I fixed it wrong a few times until I had it

MSmits: at first it worked for 1 pass, but not for chain-passes

MSmits: my own ones

MSmits: didn't really matter because when you chain pass you probably lost anyway

dbdr: however I want to implement the crash detection from replays, and that's messy. because each games reports it differently

MSmits: how do you use this"?

MSmits: do you gather information from replays automaticall?

MSmits: like euler's heroku app?

dbdr: stats, yes

MSmits: oh ok

MSmits: what do you use this for?

dbdr: and a way to know if I have timeouts without clicking all of them manually

MSmits: ah, euler does this too

dbdr: so mostly that

MSmits: cg bench will do that, but i guess with rust....

dbdr: but the stats can be interesting too, like win rates separate for each position (player 0/1/...)

MSmits: ah yes

dbdr: curious how cgbench detects timeouts

MSmits: it detects crashes

dbdr: most games give a -1.0 score if you timeout, but not all

MSmits: might not be timeout

dbdr: yes but how

MSmits: hmm should ask Neumann

dbdr: I basically have special cases for several games

dbdr: based on strings in the summary :/

eulerscheZahl: it's on github, check the code

dbdr: sure

dbdr: it might also not work on all games

MSmits: true

eulerscheZahl: my script checks for messages

eulerscheZahl: "timeout" or "disintegrated"

eulerscheZahl: the latter is hypersonic

MSmits: he also updated this part of the program a few times

dbdr: I have those, but you need to adapt to many other ones

eulerscheZahl: i even forgot where the repo is

dbdr: and in atari it's not in the summary, only as a tooltip

MSmits: https://github.com/s-vivien/CGBenchmark

eulerscheZahl: just that I found it too

dbdr: thx

Neumann: Hi

Neumann: https://github.com/s-vivien/CGBenchmark/blob/master/src/main/java/fr/svivien/cgbenchmark/model/test/TestOutput.java#L48

eulerscheZahl: Neuman n has linked it on the profile

MSmits: hi Neumann, sorry i pinged you

eulerscheZahl: and now i don't have to escape the name anymore

dbdr: https://github.com/s-vivien/CGBenchmark/commit/d2978aa1cf88688e571fd41efd3327f4a7407fd5

Neumann: Here is the part where I check for crashes

MSmits: nice :)

dbdr: so clean, with constants ;)

Neumann: It isn't perfect; as dbdr said, the detection isn't always the same for every game

Neumann: Sometimes the info is in the frame, sometimes it's in the tooltip

eulerscheZahl: return lowerCaseMsg.contains(Constants.TIMEOUT_INFORMATION_PART)

              || lowerCaseMsg.contains(Constants.TIMEOUT_BIS_INFORMATION_PART)
              || lowerCaseMsg.contains(Constants.INVALID_INFORMATION_PART)
              || lowerCaseMsg.contains(Constants.INVALID_BIS_INFORMATION_PART);

eulerscheZahl: i know your pain Neumann

Neumann: This version works with most (every ?) game

Neumann: I sometimes have to patch for new games

MSmits: i have had no issies

MSmits: issues

eulerscheZahl: but does it work for game of drones?

MSmits: havent tried

eulerscheZahl: not even sure if it was that one anymore one game was annoying in that matter

eulerscheZahl: i think i'll troll you and create a game that gives timeout string without being an actual timeout

Neumann: It should work on GoD

dbdr: this uses gameinformation and summary. so misses the tooltips

dbdr: also the tooltip is "Suicidal move", quite specific

Neumann: You can do the same kind of poor string check on replays downloaded from the last battles, to check wheter or not your arena bot times-out or not

eulerscheZahl: that's what I do

Neumann: On GoD, gameInformation : "¤RED¤Timeout: the program did not provide ..."

MSmits: I am guessing the text is red :P

eulerscheZahl: i think i remember now: wasn't part of the tooltips

dbdr: and ¤ is a standard escape char, as we all know

MSmits: naturally

Neumann: What are you guys up to these days ? Haven't been around in a while

eulerscheZahl: coding my own game

eulerscheZahl: with hookers and backjack

eulerscheZahl: black*

eulerscheZahl: ¤BLACK¤

Neumann: Sounds interesting

Neumann: CG game ?

Neumann: Might break CG's editorial line

MSmits: mostly making bots for boardgames, trying to keep up with the new games

eulerscheZahl: "I think it can be possible to make it a community contest if it's complete. keep me updated with your progress"

eulerscheZahl: now thibaud is off for holidays so i'll wait till he's back

MSmits: make the game even better

eulerscheZahl: the game is finished for what i wanted to do

MSmits: make the characters in your game do a dancing animation when they win

Neumann: There are so many new ones, I'm so late and dbdr is climbing the ladder. I might need to invest some time on CG again

eulerscheZahl: so waiting for feedback, wish me luck :)

dbdr: Neumann: good move pretending not to get the irony :+1:

MSmits: but... please keep in mind the dancing animation

MSmits: it's the key to success

Neumann: dbdr : meta-irony

eulerscheZahl: like in codebusters MSmits?

MSmits: oh, yeah

dbdr: all the way in some direction

dbdr: maybe thibault being on holiday is the perfect time to break the editorial line ;)

eulerscheZahl: Thibau**d**

eulerscheZahl: every single time :D

eulerscheZahl: what do you mean with "editorial line"?

MSmits: did you guys ever request a paper through researchgate?

eulerscheZahl: no @MSmits

dbdr: Neumann: resistance is futile https://www.youtube.com/watch?v=jQE66WA2s-A

MSmits: it's funny, i click a button and then i have to explain why i want it. So I just tell them I need it for a CG bot and they all give me their papers :P

MSmits: it's no problem apparently, they like it when their stuff gets used

Neumann: Wow, I'd never seen the whole segment. Didn't know they dubbed the whole thing haha

dbdr: MSmits: I use it when it's already there, never requested. so that works well too?

MSmits: yeah, if you have an account

MSmits: I have 1 paper on there so I could get one

dbdr: how does that work? they contact the authors?

eulerscheZahl: do you think they read your request? or could you paste the lorem ipsum as a reasoning?

MSmits: the authors get a request in a type of inbox

MSmits: and they just click approve

MSmits: eulerscheZahl they can read it, but they dont have to

dbdr: don't they have to upload the paper?

MSmits: no

MSmits: its already there

MSmits: they just release it

dbdr: from where?

MSmits: the researchgate servers i guess?

eulerscheZahl: so it probably is a per-case thing, depending on the author. some will read it, others wont

MSmits: yeah

dbdr: they magically get all papers published instantly?

dbdr: nice tech :D

MSmits: I don't know how researchgate does it

MSmits: just that it works :)

MSmits: I got 2 papers from a guy on early playout termination

MSmits: and some stuff on oware or othello, cant remember

dbdr: so now they are public?

MSmits: no

MSmits: only I got them, i am not supposed to share obviously

MSmits: it's their work

MSmits: I'll tell you if I learned anything from it

dbdr: what's the difference with all those that are public on RG?

MSmits: some just aren't public

dbdr: sure, but many are

MSmits: yes

MSmits: but research is often unique

MSmits: one paper will refer to another paper thats not public

MSmits: "we use the evaluation function from... "

MSmits: something like that

Neumann: That's the amount of "try-hard" needed nowadays to crush multis ? Read unreleased research papers ?

MSmits: they are released to universities

MSmits: just not the layman

Neumann: I'm too lazy for this

dbdr: research that is funded by public money (our taxes) should be freely available

MSmits: yes, but there's different countries

MSmits: sharing policies are not all the same

MSmits: also sometimes they want to make some money off of it

MSmits: the universities

MSmits: I guess

dbdr: editors make money, not the authors

MSmits: i know

dbdr: it's a racket, basically

MSmits: sure

dbdr: more and more people refuse it

MSmits: but it's gotten better

MSmits: 20 years ago you could get nothing

dbdr: even the reviewers do work for them, without pay

dbdr: definitely better

eulerscheZahl: my effort estimation for community games (as you see to care about global ranking Neumann): getting to the very top isn't much easier than for a contest. but gives way less points

MSmits: btw as for what Neumann said, you really don't *need* the papers, but sometimes it can save you time. I am trying to find some evaluation ideas for breakthrough currently.

MSmits: hopefully some tested ideas are in there

Neumann: I don't care that much, I just want to stay ahead of dbdr, the rest I don't care

MSmits: lol

Neumann: He already shamed me on every optim game

eulerscheZahl: :D

MSmits: these days that seems to be the same thing Neumann

MSmits: getting ahead of dbdr is getting ahead of almost everyone else

MSmits: he;s in code golf, optim and on many multis as well

dbdr: not that high on multis

eulerscheZahl: and has almost 5000 clashing points

MSmits: no, but you did well on a few

dbdr: MSmits, you have a more reasonable ranking now that you got the clashing CPs too :)

MSmits: yeah, I am happy with the quest map forcing me to do clashes :)

Neumann: All these CoC points ..

MSmits: one of these days I'll man up and do more of those ex-contest games and really get some cp :P

eulerscheZahl: the global leaderboard is mostly how much time you were willing to invest. and how you distributed it. the contest is what shows the skill level

dbdr: well, now most people hav CoC points, no? the opposite of before when nobody did

MSmits: eulerscheZahl it shows skill level to some degree, there is still a lot of time pressure on contests

MSmits: some people just cant invest enough even if they are good

MSmits: like pb4, when he has the time, he rocks

eulerscheZahl: true

MSmits: not complain here btw, I invest extreme amounts of time in contests

dbdr: contests are definitely more fair regarding time (though not perfect). but doing well under time pressure is one skill, not the absolute one

MSmits: but some cant

MSmits: it's more a family thing dbdr

MSmits: I can tell my wife I am checking out for a week

MSmits: that's not true for everyone :P

dbdr: sure, so nothing is perfectly fair

MSmits: yeah and that's ok

dbdr: I was more responding to euler

MSmits: ah ok

eulerscheZahl: then let's say ranking on single games vs global ranking

dbdr: like Marchete does not like contests, does not mean he's not skilled

MSmits: I also don't think it's problematic that the global leaderboard is about time invested

eulerscheZahl: the latter is a hell of time, while playing 1 or 2 games are feasible

dbdr: yeah, and many people can be #1 in at least one thing and say this is what they focused on :)

dbdr: so every1 is a winner ;)

MSmits: there are very few people that systematically do all the multis just to get cp

MSmits: even euler i would not say did this, he's just been here a long time

MSmits: tric trac did it though

MSmits: that was fast

eulerscheZahl: and now i can't keep up with the speed of new games popping up

MSmits: maybe not fully, but you keep up fairly well

MSmits: at least you do the optims mostly?

dbdr: new games give CPs only slowly anyways

MSmits: also eulerscheZahl, it's partially your fault :P

eulerscheZahl: https://prnt.sc/tw6tep do you really think so?

MSmits: oh thats more than I thought

MSmits: some of these you can quickly slap a minimax on though

MSmits: takes you 1 day each

eulerscheZahl: and maybe it's more fun to make others waste their time than to only invest mine :imp:

MSmits: a basic minimax will get you rank 10 on yavalath euler

MSmits: well at least to p20

MSmits: good cp

eulerscheZahl: meh

eulerscheZahl: not doing it for the CP

eulerscheZahl: in theory I know how to do a basic minimax

MSmits: didnt you do it a lot?

eulerscheZahl: i care more about the gameplay

MSmits: ah, yeah me too

eulerscheZahl: you love those board games. i miss the "extra twist" here

eulerscheZahl: the bells and whistles

MSmits: the "extra' twixt maybe?

dbdr: the hoockers?

eulerscheZahl: :D

MSmits: lol

MSmits: do you mean graphics/animation wise or just pure gameplay options?

MSmits: like special moves/powers etc.?

eulerscheZahl: gameplay wise

eulerscheZahl: the mix of search and heuristics

eulerscheZahl: i know you need a scoring for those board games too

eulerscheZahl: it just feels different and i'm not that much into it

eulerscheZahl: and maybe i'm just not as hyped anymore as i was 3 or 4 years ago

dbdr: missing the immersion of less abstract games

eulerscheZahl: had some sleepless nights because i just couldn't stop thinking when I was new

eulerscheZahl: I don't have those anymore

dbdr: use coffee. you can sleep when you're dead ;)

eulerscheZahl: that terribly smelling stuff? no way

dbdr: it's a learned taste :)

MSmits: I have less sleepless nights also... but still get hyped on contests

MSmits: even though I find the games themselves less interesting than boardgames

MSmits: it's funny that you yourself built one of the most interesting boardgames on CG euler

dbdr: again, an eulerscheZahl/MSmits combo would be basically unbeatable ;)

eulerscheZahl: coding a game is a different experience that playing one :P

MSmits: well I do remember that on contests where we shared more in PM we did better :P

MSmits: or I did anyway, always nice to get some professional help on tools and stuff

MSmits: remind me to bug you more on next contest euler :P

eulerscheZahl: yeah, "professional" i never really learned that web request stuff

eulerscheZahl: that part is self-taught

MSmits: isn't almost everything professional coders do?

eulerscheZahl: without stackoverflow most of us would be doomed

MSmits: yea

dbdr: for the parts you don't use often

jacek: i use scihub for papers

jacek: oops scrolled

eulerscheZahl: i use printer paper

eulerscheZahl: 500 sheets for 3 Euro or whatever it costs

dbdr: isn't that a bit blank though? ;)

eulerscheZahl: i need some free space to scribble my thoughts on it

jacek: poor papers

eulerscheZahl: should I get a whiteboard?

jacek: thats racist

dbdr: I just got one!

MSmits: I always end up getting free notebooks (the paper kind) that i scribble in

MSmits: you get them at a conference and such

eulerscheZahl: i know, visited some too back in my student days

MSmits: I thought you still did

eulerscheZahl: and we had a coding contest at university twice a year. sponsored so you got free pens and paper

MSmits: I kinda have no choice at work

MSmits: someone has to go to these things

MSmits: school has to have a presence

MSmits: I can avoid it sometimes, but not always and sometimes they're cool

eulerscheZahl: i've only been part of an "open door day" (not sure if that translates well)

MSmits: this is when other companies or students come visit your company?

eulerscheZahl: so those interested in a job after school could come and look around, seeing what we are doing there

eulerscheZahl: exactly

MSmits: ah yes, we have those as well

MSmits: not just for children to come to the school, also for other school to share experiences

eulerscheZahl: or where the mom comes and asks all kind of questions while the kid is standing in the back :P

MSmits: yes that happens, but generally in the last few years they tend to try to push the kid to ask the questions

MSmits: which is better

MSmits: the day where we advertise for children to come to our school and parents come and such, is the toughest day of the year i think

MSmits: it's from 7 pm to 9 pm and afterward you need to cleanup

MSmits: but some of us still have to teach in the morning

MSmits: both that day and the next

MSmits: I always start the first class of the next day saying: Sorry if I seem zombie-like today

eulerscheZahl: but CG is running a contest again

MSmits: you got it?

eulerscheZahl: huh? you are a zombie during CG contests

MSmits: or do you mean company sponsored secret contest?

MSmits: oh ok

MSmits: that

MSmits: yeah that too, but thats a bad excuse :P

MSmits: i can only use the zombie excuse when it's caused by work

eulerscheZahl: your students know about your CG habits anyways

eulerscheZahl: they can count 1 and 1 together

MSmits: a few months into the year they do, not at the start

dbdr: do we need CGer Anonymous? ;)

eulerscheZahl: Anonymous Algorighmics

eulerscheZahl: I still have the contribution somewhere, never finished it

MSmits: puzzle?

eulerscheZahl: multiplayer

MSmits: oh really

MSmits: what was it about?

MSmits: was that the beer game?

dbdr: you got MSmits interest :D

eulerscheZahl: yes

MSmits: I seem to remember

dbdr: again?

dbdr: Vinidium has beer too

eulerscheZahl: and Bender

dbdr: true

dbdr: how German ;)

MSmits: yeah but he had a more beer-central game than vindinium

eulerscheZahl: Anonymous Algorithmics was my first game which I never finished

eulerscheZahl: just to get in touch with the SDK

eulerscheZahl: you are building bars

eulerscheZahl: and navigating passer-byes to visit them

dbdr: oh, must use very old SDK version then

eulerscheZahl: version 1.X

eulerscheZahl: we have 3.14 now

eulerscheZahl: 3.15

MSmits: past pi already huh

dbdr: growing fast :D

eulerscheZahl: so of course it's broken If I try to reupload it now

dbdr: doesn't it depend on a specific version?

dbdr: like old games don't get broken

eulerscheZahl: pixi.js itself got an update

dbdr: hopefully

eulerscheZahl: so all the older version are now blacklisted on CG

eulerscheZahl: can't upload

dbdr: but they still work

eulerscheZahl: offline

dbdr: mot sure if makes sense

dbdr: *it

dbdr: I mean uTTT still works online too

dbdr: it was the first SDK game I think

eulerscheZahl: all I know: you can't upload some, really old versions anymore

dbdr: kinda makes sense, at the same time I don't think they can ever stop supporting old SDKs

eulerscheZahl: and some old contributions just stopped working

dbdr: multis?

AntiSquid: how are old games being updated

AntiSquid: on CG

dbdr: I don't think they are updated

eulerscheZahl: that only affects really old versions. before anyone could upload their games

eulerscheZahl: it was in Beta for a long time

AntiSquid: and some games have really old versions no? games here aren't containerized, i wonder how they don't break

dbdr: right, makes sense not to support that

dbdr: AntiSquid: because they depend on a specific version

eulerscheZahl: defined in pom.xml

dbdr: which still exists

eulerscheZahl: maven still has the old SDK versions

AntiSquid: so thy do have the older versions still available but deny access to them from new games eh

eulerscheZahl: I even updated Bender this week was on my TODO for a long time not it's not eating all the RAM anymore

dbdr: very cool to do that!

eulerscheZahl: no AntiSquid. some old contributions are completely broken now (black replay)

dbdr: what fixed the RAM issue?

eulerscheZahl: SDK version update. I don't know the details

dbdr: ok

dbdr: what kind of old contributions?

eulerscheZahl: it's just that the broken contributions luckily never got published

eulerscheZahl: just some user who probably left long ago still having a broken contribution in the private section

eulerscheZahl: there was a rock paper scissors game. that might be broken, dunno

eulerscheZahl: OoC was broken too before Illedan and I got our hands on it

AntiSquid: the RPS was removed

AntiSquid: no idea why

eulerscheZahl: https://www.codingame.com/ide/demo/71980557041b57ab49fcd374d3e804175be358 the anonymous algorithmics contribution...

eulerscheZahl: images not shown anymore

AntiSquid: BotG and Code Royale are older

eulerscheZahl: maybe CG updated those on their own? I don't know

MSmits: mmh went from rank 109 to top 50 breakthrough just by counting pieces. It's definitely more helpful than it is in Othello

eulerscheZahl: the IDE link above doesn't even have a "view contribution"

Default avatar.png sOMEONE: heys

AntiSquid: no

dbdr: there is also no stub

eulerscheZahl: just try this

eulerscheZahl: http://chat.codingame.com/pastebin/12037df2-aa1b-4b06-a95b-75705a52809b

dbdr: Oups An error occurred (#NO_CODE): "A code must be provided".

dbdr: someone was collecting, that's a new one :)

eulerscheZahl: :D

eulerscheZahl: I remember that forum post

dbdr: class Alcoholic :D

eulerscheZahl: a classy drunk

dbdr: oh, viewer works

eulerscheZahl: partially

dbdr: ah, ok

eulerscheZahl: there used to be images

dbdr: looks psychedelic

eulerscheZahl: moe's tavern and the bar from Family Guy

dbdr: #1 :D

dbdr: too bad I can't approve it

eulerscheZahl: zeno submitted in 2018. man, this is old

dbdr: it's Alcoholics Anonymous

eulerscheZahl: so: is it more beer-centered than Vindinium? :D

dbdr: definitely

Default avatar.png Fronix: 5 minuts challenge?

eulerscheZahl: don't you dare inviting us to a clash of code

dbdr: "Integer y: the y location of the alcoholic Integer state: 2 if upgraded, 1 otherwise"

dbdr: so you can upgrade alcoholics? :D

eulerscheZahl: no, bars

dbdr: well...

eulerscheZahl: don' trust the statement

dbdr: the statement says otherwise

eulerscheZahl: don' trust the statement

dbdr: were you drunk when you wrote it?

eulerscheZahl: i don't drink alcohol at all :D

dbdr: thought so

eulerscheZahl: but I copy-paste

dbdr: worse ;)

dbdr: copy-pasters anonymous

MSmits: its funny how adding more eval features completely changes your bot in weird ways

MSmits: https://www.codingame.com/share-replay/480898165

MSmits: I added mobility and empty columns and now it will immediately remove the front row to give more space for the back row

MSmits: while keeping all columns filled

jacek: he does only what you told him to do

MSmits: yeah, i know, it still funny

MSmits: because it's so regular

jacek: looks like that phalanx game

MSmits: kinda looks like a phalanx, not sure which game you mean

jacek: do you use race patterns yet

MSmits: nah, it's harder than I thought

MSmits: I will try once I get a basic bot going

MSmits: i want to get a bunch of features in like i did for othello, maybe try and avx them a bit

MSmits: it looks stupid now

MSmits: want it to look a bit more like it knows what it's doing :)

jacek: hmm cant find it. its was like breakthrough but wider and you could move few pawns in group

MSmits: ah ok

MSmits: breakthrough is a bit like having army lines

MSmits: I guess that's what i need to eval

MSmits: support from neighboring pawns

MSmits: ntuples might work here too btw

jacek: there was paper on that

MSmits: oh ok

jacek: i tried TDL. it was winning 90% agains my normal eval in 1 ply, but got crushed in more plies

MSmits: TDL what? a ntuple bot?

jacek: all 3x3 patterns for breakthrough

MSmits: ahh ok

MSmits: maybe it just took too much time to calculate?

MSmits: or to look up I guess

jacek: it was as fast

jacek: and well, i played with fixed depth

MSmits: hmm

MSmits: did you try to figure out what was missing from your ntuple bot that was in your normal bot? There is nothing wrong with combining the two?

MSmits: just add up the evals with some parameter

jacek: having 18 x 19683 weights is meh anyway

jacek: no, i tried only ntuples

MSmits: hm ok

MSmits: hmm how do you fit that/

jacek: fit what

MSmits: that many weights

jacek: i only used it locally

MSmits: ah ok

jacek: my NN has 30k weights

MSmits: what type of NN is it?

MSmits: Robo always does the alpha zero thing I think

jacek: its just MLP with 1 hidden layer and 1 output giving eval

jacek: im too dumb to make policy net yet

MSmits: my little pony again?

jacek: you said it

jacek: multi-layer perceptron

MSmits: ah yeah

MSmits: 1 hidden layer doesn't seem much chance for complicated patterns?

jacek: 1 hidden layer is enough to predict any function. though youd need many many units in it :)

MSmits: ah ok

jacek: i dont use any CNN or filters either

MSmits: how do you train these locally?

MSmits: what tools?

jacek: for breakthrough, i player thousands games with my handcrafted bot and let it to analyze each position to 4ply, then NN tries to fit into that

MSmits: oh supervised learning ?

MSmits: that requires you to already have quite a good bot

MSmits: you were rank 2 before weren't you?

MSmits: under tric trac/

MSmits: ?

jacek: which i have

jacek: yes

MSmits: kk that explains that

MSmits: and this was an ab bot?

MSmits: i think tric trac had one

jacek: yes

MSmits: ah ok

MSmits: thanks for answering all those questions :)

jacek: if my eval is smart as ab's eval at 4 plies, then i get 'free' 4 plies strength

Default avatar.png TheMassiveGhost_8138: How do i 5 star "The Descent"? Or are the stars the rating of the level made by players^

Default avatar.png TheMassiveGhost_8138: ?

MSmits: yes jacek makes total sense to me

jacek: besides i also reinforced it, my old NN analyzed for 4 plies and tried to fit in. again and again until saturates

MSmits: TheMassiveGhost_8138 they are contributor ratings, dont worry about them

Default avatar.png TheMassiveGhost_8138: oh okay thank you

MSmits: 100% = 100%

MSmits: some have achievements though, very few

MSmits: jacek, one time I suggested to re curse that if you nearly solve a game or run a very deep meta mcts, you can use it to train a NN, he said that would not work. But isn't that what you did sort of?

jacek: yes

jacek: i try to mimic teacher, be it solved values, or deeper bot

MSmits: yeah

jacek: https://i.imgur.com/J4cgS7O.png

MSmits: nice lines

jacek: in blue one screwed and forgot about scaling, yet it beats the handrafted eval after 9k

jacek: games

MSmits: I see

MSmits: well it's nice to see that training a NN is not yet an I-win button and you still need to make a good regular bot first

MSmits: this is true for creating books with meta mcts also

MSmits: without a good bot it just won't work

jacek: actually othello was first trained by TDL, then i used the self supervised thing

MSmits: is it a NN now?

MSmits: it wasnt before

jacek: ntuples

MSmits: hmm ok, but thats not really NN because you decided the patterns

jacek: but you still need to train weights

MSmits: true

MSmits: if you were wanting to, i am sure you could take nr 1 in Othello btw

jacek: let rust have a victory for once

MSmits: hmm ok

MSmits: well he needs to retrain his bot then

MSmits: and make it a little less deterministic, because i can re-book it in 1 hr

MSmits: yours will take me a day or something

MSmits: it's much more random

MSmits: not sure if I am going to bother to keep doing that if it takes me as long as it did the first time

MSmits: code size is not the limit here

MSmits: it's me-time

jacek: ask for some grants from the university :v

MSmits: lol

Default avatar.png xlr4829: javascript for ever

Default avatar.png xlr4829: AI ????

MSmits: ai ai sir

jrke: lol,i pushed someone above boss and later pushed him down but still he is going to be promoted in next league :smiley:

DomiKo: yes, that's weird

DomiKo: weird is CodeBusters silver league too

DomiKo: 24 people

DomiKo: silver boss look like random

DomiKo: while the bronze one was pretty strong

jrke: more weird is Botters of galaxy gold only 6 people

AntiSquid: no it's not weird, once you are above boss with stable rank you get promoted

AntiSquid: also botg multi gold boss is weaker than contest gold boss afaik

DomiKo: weird is that silver boss < bronze boss

eulerscheZahl: for BotG it's hard to tell if a boss is hard or not. They are vastly different at least

DomiKo: in CodeBusters

eulerscheZahl: contest: ranged multiplayer: meele boss

DomiKo: silver boss is randomly walking

DomiKo: so i guess he is weaker

eulerscheZahl: silver is also really small at codebusters

AntiSquid: no, multi is lancer flip + drS pull

eulerscheZahl: oh, you said that a few lines above

AntiSquid: rare type of bot

eulerscheZahl: a Thibaud bot

eulerscheZahl: by contest bot was like 5 points above the boss when I submitted it in the multiplayer

eulerscheZahl: promoted a few others as well rock paper scissors helped me against the boss

AntiSquid: problem with that kind of unique boss is you can detect boss by hero selection and hard code vs it :/

AntiSquid: so if someone in gold hardcodes then ya boss drops :D

LeBaoHoang8A4: hey can yu guys tell me how to block someonr? One guy just iviting me to clash ALL the time!

AntiSquid: you could disable clash invites

AntiSquid: but it disables all

LeBaoHoang8A4: how to do it?

AntiSquid: https://www.codingame.com/settings/notifications

LeBaoHoang8A4: thanks!

AntiSquid: bottom one

LeBaoHoang8A4: got it

Default avatar.png xlr4829: pika pika pikashu

Default avatar.png DominikDrums: Hi

Gorbit99: hi

Default avatar.png DominikDrums: Can you guys tell me how to view history of my submissions for each clash of code I participated in?

Gorbit99: don't think you can

Default avatar.png DominikDrums: So after game is over I cannot go back to this task and review my code later?

Gorbit99: you aren't really supposed to

Default avatar.png DominikDrums: Why? It would be only available to me and I guess the number of tasks is large enough not to expect overlap of task

Gorbit99: because it's a quick and dirty race and you are basically supposed to do hacky, ugly code

Gorbit99: also, you haven't yet played 500+ clashes, then they do repeat

jacek: well in notifications when clash is over you have link to the clash

jacek: i think you can see code there

Gorbit99: you can't I don't think so

Gorbit99: well, if you shared

Gorbit99: guess another reason to share your code

DomiKo: you can see it

DomiKo: you dont't need to share to see your own code

DomiKo: you can always see it

Gorbit99: yeah, you're right, I just checked one where I didn't share

Gorbit99: I guess the issue then is that you can't see the task itself again

DomiKo: yap

DomiKo: but you can search it ;P

Gorbit99: you can't if you don't see the submissions

aCat: but you don't have the title ;-)

Gorbit99: eh, that's the least of the issues, as you have the author

aCat: you can see contributions with level >=20

Gorbit99: you can also see contributions if you've got to the contribution questline in coc

Gorbit99: even if you're below level 20

eulerscheZahl: but you have the author aCat. that's enough to start a search

aCat: It was already said

aCat: don't jump at me euler ;p

eulerscheZahl: i'm stupid

aCat: I was the one trying to help here ;p

eulerscheZahl: how's the same game going?

aCat: I got offline the multiturn version done

aCat: an hour this morning

aCat: but

aCat: Now I suppose it is stupid to leave warnings for the nonlegal moves

aCat: they should be errors straight ahead

aCat: what do you think?

eulerscheZahl: both options sound fine to me

aCat: hm

aCat: when it is warning, you can gain more time by sending nonlegal moves

eulerscheZahl: when do you end the game? no more valid moves?

aCat: like i got solution of length 80 best

eulerscheZahl: good point. then it's an error i guess

aCat: yep

aCat: and spend 20 additional turns on computing

aCat: you will never loose only possibly gain

aCat: yep ;-)

aCat: I will try update it today

aCat: just trying to finish this damn crossword ;p

eulerscheZahl: another puzzle of the week that you will finish too late?

DomiKo: crossword was so ugly

aCat: exactly

aCat: I didn't had much time

aCat: and I'm trying to do ruby achievement

aCat: what an ugly version of python

aCat: but some tricks are really interesting

eulerscheZahl: https://www.codingame.com/training/medium/rubik%C2%AE the language farming puzzle

aCat: nice to know

aCat: In theory I wanted to do real puzzles with unknown languages to learn them at least a little

aCat: and I did a lot for ruby

aCat: but yeah

aCat: onboarding all the way baby ^^'

Gorbit99: what

DomiKo: time to get some bronzs thanks :smile:

Gorbit99: I thought descent is the farming one

eulerscheZahl: there are even easier ones than rubik :D

aCat: oh I forgot about descent

Gorbit99: but this is medium

aCat: but cmon - I got lot of unsolved

eulerscheZahl: like A*Craft: submit the default code and you get 100% correctness

aCat: but it's not a puzzle?

eulerscheZahl: optim counts too

aCat: oO

eulerscheZahl: as long as you get 100%

aCat: buahahaha

Gorbit99: what's the easiest hardest?

aCat: batman one

eulerscheZahl: you mean from the very hard section?

Gorbit99: I think I've found the easiest very hard ones, but I wasn't successful with hard

eulerscheZahl: for me it was the last resistance

aCat: before we spoke euler I really planned to do this the valid way ;p

eulerscheZahl: :D

aCat: ouh hard - I don;t remember

Gorbit99: for me very hard was easily music sheets and light up the box or whatever it is

Gorbit99: I have 2 very hard puzzles achievement, but not the 2 hard one

eulerscheZahl: for the community puzzles I lost track which is in which section

eulerscheZahl: roller coaster wasn't that hard

Gorbit99: I guess most hard and very hard are about a simple problem that needs optimizing

dbdr: eulerscheZahl, in clojure? ;)

eulerscheZahl: or just put in the wrong category

eulerscheZahl: what in clojure?

dbdr: roller coster

dbdr: there's an achievement

eulerscheZahl: that one was a nightmare

dbdr: :D

eulerscheZahl: zeno helped me as my solution was too slow

Gorbit99: is it like bash riders?

dbdr: yes

Gorbit99: just straight up painful?

dbdr: clojure is fine

eulerscheZahl: bash riders?

dbdr: euler is just allergic to it

Gorbit99: horse racing duels in bash

eulerscheZahl: oh. I did that in python

eulerscheZahl: python3 -c "code here"

Gorbit99: ...oh

aCat: xd

dbdr: :no_mouth:

Gorbit99: cheater

Gorbit99: that's genius

aCat: you did this with ALL bash puzzles :>?

eulerscheZahl: someone did that in a bot contest and claimed it was pure bash once :D

aCat: haha

eulerscheZahl: of course aCat

dbdr: my vinidium is real bash :)

aCat: bashmaster

jacek: bashing keyboard until it works

Gorbit99: does anyone use bash in actual competitions?

DomiKo: now i know how to get top1 language

dbdr: they check it

DomiKo: :(

dbdr: in contests

DomiKo: :sadface:

eulerscheZahl: they only cared because back then first in a language was a tshirt winner

eulerscheZahl: they don't have that anymore. neither top50

Gorbit99: It'd be nice to get a tshirt, but I guess it's less effort to just go online and make it yourself in a tshirt creator

dbdr: but that's not the same

eulerscheZahl: https://www.codingame.com/forum/t/xmash-rush-cc07-feedback-strategies/74346/23

Gorbit99: I knew making 10 bug feedback posts was worth it

eulerscheZahl: (scroll to the bottom of that post)

DomiKo: tshirt is worth a lot more than money

Gorbit99: just got my 10th like

Gorbit99: what are we looking for eulerscheZahl?

eulerscheZahl: the baby picture

eulerscheZahl: wearing a CG tshirt

Gorbit99: aww

eulerscheZahl: self-made as CG doesn't ship baby sizes

dbdr: maybe if you wash it at 90C?

Astrobytes: And the baby did all the coding anyway :P

dbdr: or cook it with your spoon

Astrobytes: :D

eulerscheZahl: and then eat it with your spoon?

dbdr: there is no spoon

eulerscheZahl: now tourist is on track it seems

eulerscheZahl: didn't submit in the first 2h, got 2 problems now

dbdr: hm, wonder why someone would have a gree up arrow after drawing against someone lower

eulerscheZahl: more confidence

eulerscheZahl: is that sigma?

Default avatar.png LonelyTree: https://www.codingame.com/replay/480915392

Gorbit99: poor bot

dbdr: yes, probably

eulerscheZahl: a 100m marathon https://www.codingame.com/contribute/view/52944bbc0f5f9a5cf19c728a514964c97d19

aCat: Done !

eulerscheZahl: nice

aCat: 88 lines in Ruby :P

aCat: level 35 ;-)

eulerscheZahl: oh, i thought the same game

aCat: catching euler slowly but surely ^^'

Washier: gg

Gorbit99: my kinda marathon

aCat: PoTW

eulerscheZahl: now just switch the order of the digits

aCat: do some work now

aCat: SmaeGame later I'll let you know

eulerscheZahl: https://codingcompetitions.withgoogle.com/codejam/round/000000000019ff31 and he submitted 3 at once :D

aCat: ok, this got me unlock full questline ;-)

aCat: just optimization with marslanding left

eulerscheZahl: that will require some golfing as well

eulerscheZahl: top2% on the golf leaderboard

Washier: you've got ruby aCat, you can doo eet

eulerscheZahl: start with "don't panic" in 200 chars

aCat: :crying_cat_face:

Washier: i see it can be done in 200 chars even in C#, how!?! must find out

Gorbit99: last I remember, it wasn't a complicated challenge

Gorbit99: stupid me did it in C++ tho

ush189: anyone knows how to debug in php?

ush189: echo is for the output

DomiKo: ohh yes 100 achievements

DomiKo: now only 2% code golf....

Gorbit99: don't panic is almost below 200

Gorbit99: got it

eulerscheZahl: Washier system calls :(

Gorbit99: ?

Washier: :(

Gorbit99: ah

Gorbit99: any way to get rid of stdout sync?

Default avatar.png Galaxy_001: std::ios_base::sync_with_stdio(false); used in c++. It doesn't sync with c.

Gorbit99: I'm using ruby

SplitCode: Does someone have a math page on how to calculate steps in an angle given two points? (the thor practice example)

Gorbit99: steps?

SplitCode: Yes, like starting at (0,0) going to (10,10), the next (logical) step would be (1,1)

Gorbit99: eulerscheZahl are the shortest solutions for don't panic abusing the validators or are they legit?

Gorbit99: you're a bit overthinking it SplitCode

Astrobytes: legit, but system calls etc sometimes

Gorbit99: I guess then because ruby is shortest, that's legit

Gorbit99: geezus

Colorhacker: I've got a question

Colorhacker: In code royale, there are mines, how much does it take to upgrade or build them?

Colorhacker: It's not written in the description

Colorhacker: I tried to build but sometimes i couldn't?

Colorhacker: It didnt take money when built but i still couldnt build sometimes

Colorhacker: No one's here?

Colorhacker: The chat has never been this quiet

Default avatar.png Sumitr: F

Colorhacker: ?

eulerscheZahl: i don't even know what others do at golfing. i', heavily abusing the testcase setup for Thor

eulerscheZahl: my other solutions should be working fine in general

Default avatar.png illogicalism: are there bots here?

eulerscheZahl: yes: AutomatonNN Automaton2000

Automaton2000: or at least trying to

AutomatonNN: wow code a la mode

Zenoscave: eulerscheZahl is a bot too

Colorhacker: hey!

Colorhacker: they actually are bots?

Zenoscave: except eulerscheZahl I was joking

Colorhacker: I know

Colorhacker: but i didnt knew automatoNN was a bot

Colorhacker: are they official codingame chatbots?

Zenoscave: nope

Colorhacker: ok

Uljahn: semi-official

Zenoscave: good way to put it

Zenoscave: happy caturday

Uljahn: pewpew

Zenoscave: and a pewpew to you

Astrobytes: pew

Colorhacker: automaton2000 can you say something

Automaton2000: i think i read something about that

Colorhacker: Hey nice xD

Colorhacker: automaton2000 what is the universe

Automaton2000: sounds like a good idea.

Colorhacker: xDDDD

pillarglass: LOSER

pillarglass: ELIYAH

pillarglass: YALL GONNA NEED SOME LUCK LMAO

pillarglass: LOOCK AT U

pillarglass: LOOK*

Astrobytes: Calm down laddie, you clashing?

eulerscheZahl: fair warning pillarglass

pillarglass: im sorry

pillarglass: yes it was all in good fun

pillarglass: i will not do again

Astrobytes: no worries :)

pillarglass: these clashes get competitive :triumph:

Astrobytes: yeah all good, is the clash chat not working again?

pillarglass: seems to be that way

pillarglass: not sure

Astrobytes: do you get a clash specific channel open when you clash? Or does it close or...

eulerscheZahl: hm, something went wrong :thinking: https://imgur.com/a/bvN8qUE

pillarglass: closes once im in

Astrobytes: OK that's a known issue pillarglass, sorry about that. Devs know about it, not sure when it's gonna be fixed. There's a #clash channel if that's any use

Astrobytes: wtf eulerscheZahl

Default avatar.png Memo12334: huh

eulerscheZahl: I try to include three.js

Default avatar.png Memo12334: what is that challenge

eulerscheZahl: just for the lols, I don't have a game idea that needs it

Astrobytes: Nice. Interesting result

eulerscheZahl: when I upload it to CG, I only see the selection menu in the viewer

eulerscheZahl: no teapot at all

Astrobytes: Any idea why?

eulerscheZahl: i'm not sure how the pixi part is even added yet

eulerscheZahl: so I try to imitate that then

jacek: no darkhorse64?

eulerscheZahl: nope :(

AntiSquid: wow a 3D CG game to play at tea time hours :o

eulerscheZahl: no, just the controls. https://www.codingame.com/demo/849340c736362b181cdae18a8345462538aa01

eulerscheZahl: hm, pixi is deeply nested :/

eulerscheZahl: http://chat.codingame.com/pastebin/1877e98e-4226-4868-b5f0-66a581ffdb43

eulerscheZahl: in Drawer.js

eulerscheZahl: i don't even know if i can replace that

eulerscheZahl: but maybe have pixi in parallel without actually using it

darkhorse64: Trying to enlighten me ?

Astrobytes: lol

Default avatar.png JBM: Parallel ok, but I want 0% could usage

Default avatar.png JBM: *cpu damn autocorrect

MSmits: heyy darkhorse64, I started your game finally. Rank 50 so far

darkhorse64: from your ranking, I guess you have not yet an eval

MSmits: i do, a simple one

MSmits: just counting pieces gave me rank 50, i added mobility and empty columns, but that didnt do much

MSmits: the parameters for them might be too big or too small though, no way to know without trying to fit

MSmits: trying to do the connection thing now, giving points for pawns connected horizontally or vertically

MSmits: should also give points for how advanced pawns are I guess

Scarfield: thats actually all i did for my eval, row*row xD

MSmits: hmm how so?

MSmits: whats row*row?

Scarfield: rows away from "home"

MSmits: oh you gave quadratic points for distance from home/

Scarfield: yea

MSmits: makes sense

Scarfield: super simple, and havent gotten back to it yet

darkhorse64: I have a piece value depending on its location with a bonus if the piece is safe

MSmits: I try to put a whole bunch of evals in and trim the stuff that doesnt help

MSmits: I removed 2 features from my othello bot during fitting

MSmits: makes sense too darkhorse64

darkhorse64: I have not invested much time in my bot. My ept is depth 4 and I select winning moves in my rollouts. May be I should look into race patterns

darkhorse64: Lots of players = lots of CP in this multi

MSmits: yeah

MSmits: I will look into race patterns for sure, hope i can work it out

Laminator: Are you kidding e

Laminator: me

darkhorse64: I skimmed through the "Solving" paper yesterday. Looks like a promising way to prune moves although decoy moves are important

darkhorse64: why kidding ?

Laminator: just lost 1st place in fastest clash with the same time lol sorry I have no other way to communicate with my clashers

MSmits: yeah I will try to do something like in that paper

MSmits: the difference might be in the floating point variable for the time, but rounded to show the same time

MSmits: if you lost, you can be sure you were just slightly slower

Allis: Are ties literally impossible?

MSmits: or your internet connection was slightly slower when you were submitting it, though the other guy might have had a slower connection also and he actually won by 2 seconds :P

MSmits: Allis probably they are yes

MSmits: it doesn't really matter though, if the result is so close that you are in the error margin, then you just don't know who won.

MSmits: with enough games ending like this, you'll win half the time

MSmits: = fair

MSmits: no point stressing over that one result or what actually led to you narrowly losing

MSmits: whatever it is, it will help as much as hurt

Laminator: It just makes me wonder if I could've gone any faster

darkhorse64: use more fingers to type ?

Allis: If you hadn't run the tests, you would've won. :P

Laminator: exactly!

Laminator: sometimes those tests take a while too

Laminator: but sometimes I fail on the very last test so it's worth it.

aCat: eulerscheZahl I updated SameGame

aCat: OK - I updated the puzzle. Important changes: - now you can make your action turn-by-turn (-1 for empty tile) - no more warnings - selecting illegal action causes hard error.

Laminator: wow, my fastest solve yet, and you were twice as fast. wtf lol

Allis: ¯\_(ツ)_/¯

Allis: The really easy ones just come down to reading and typing quickly.

Laminator: Well, I had the fastest PYTHON solve. Leggo!

Allis: Counts for somethin'. :P

Laminator: I've seen what you Ruby people get away with for shortest mode. Not fair.

MSmits: note to self: It does not help if you do a 64 bitboard and only use 32 bit popcounts

MSmits: that little difference between __builtin_popcount and __builtin_popcountl

MSmits: looks like top 20 :)

Default avatar.png LIONSxLEADER: how does scoring work for ranking clash of code ?

Default avatar.png LIONSxLEADER: globally

Default avatar.png LIONSxLEADER: participation trophys weighed? with a completion% and time taken modifiers?

Allis: @LIONSxLEADER It's based entirely on whom you beat/lose to.

Allis: I don't know whether it's based on the sum or the average of your competition, but it helps to think of it like an Elo rating in chess.

Allis: CodinGame technically uses TrueSkill, but it's very similar to Elo.

Olusola: :smiley:

No_Name_26: :L

No_Name_26: :L