The Nineteenth Byte

@MartinBüttner forking a project with a github.io website as part of it is not as straightforward as it should be. I encountered this while forking 2048

Geobits

5:09 PM

Hmm. github ToS says you can have more than one account, just not more than one free account :D

cjfaure

@Geobits what a technicality! o: I don't need any gh premium features though

i love github enough to buy it just for the 2 accounts, i just don't have any money

Martin Büttner

@Sparr hm okay... I did fork 2048, too (or rather I forked a fork of it)... but I don't remember any problems

Sparr

ok, I've forgotten the details...

can you point me to the github repo for gabrielecirulli.github.io/2048 ?

I don't think it's github.com/gabrielecirulli/2048 because that's not a github.io repo

Martin Büttner

github.com/gabrielecirulli/2048

it's the gh-pages branch

Sparr

ahh, yes, that was the catch

having to clone that branch

which isn't mentioned on the "how to make a github.io" instructions on github.io

Martin Büttner

5:21 PM

but doesn't it say "just push to gh-pages"?

well I've got to catch a train

see you later

overactor

@MartinBüttner @cjfaure If someone makes me a subtractor function, I've got a divider function:

di:
P1 I1 .. I2 P2 .. .. .. ..
\f .. .. .. /f .. .. .. ..
.. .. .. D1 .. .. .. .. ..
.. A .. A .. .. .. .. ..
.. .. Ma .. A .. .. .. ..
.. .. .. D1 .. .. .. .. ..
.. .. .. /f .. /f .. .. ..
.. .. == .. /f .. .. .. ..
\f .. S1 .. .. .. .. .. ..
.. \f U A .. .. .. P3 ..
.. .. .. .. .. .. .. 01 ..
.. .. .. .. .. .. .. S2 ..
.. .. Su .. S2 .. .. A 00
.. .. P1 .. P2 .. P3 .. S1

Ma:
.. P1 I1 .. .. .. I2 P2 ..
.. .. A .. .. .. A .. ..
.. -- .. S2 .. S1 .. -- ..
.. .. .. \f .. /f .. .. ..
.. >0 .. .. \t .. .. >0 ..

di divides

Ma returns the max

Mi returns the Min

== returns 0 if the inputs are equal and nothing if they aren't

Sparr

is there a spec for this language anywher?

it seems like you've got a lot of syntax flushed out that I didn't see discussed here

Geobits

I think the spec is the last 24hrs of chat...

3

cjfaure

@Geobits that...is sad

Geobits

I haven't seen anything else, at least.

cjfaure

5:24 PM

i'll probably try to write some docs later but i digress

Sparr

@overactor be sure to hit the 'fixed width' button on pastes like that

Geobits

Nah, don't write docs. Just add a documentation link that sends people to the chat room transcript to infuriate them.

overactor

@Sparr But, I had 4 spaces...

Geobits

Did you have four spaces on every line?

Sparr

it has to be every line of the whole entry in chat, not just blocks of lines

Geobits

5:29 PM

It's easier to paste the code as a separate block and either ctrl-k or hit the fixed-width button.

Sparr

aye

I finally realized I can use the {} button on the SE editor instead of replacing ^ with four spaces in my editor before copying code :)

New Sandboxed Posts

5:47 PM

0

A: Proposed Question Sandbox - Mark XIV

Steal Personal Data code-challenge UNDER CONSTRUCTION (please do comment/vote) Your task, as an employee of a certain nefarious corporation, is to write a program which will look through files on the computer it is running on and parse them in order to find personal information about the comput...

Nathan Merrill

Just read dining philosophers

I think its a fantastic challenge

Sparr

where?

Nathan Merrill

@Rainbolt

Sparr

like, the conceptual idea, or specifically a codegolf submission?

Nathan Merrill

0

A: Proposed Question Sandbox - Mark XIV

Dining Philosophers (Incomplete) Please do not vote or comment yet. Introduction Cooperate with other Philosophers in order to eat your dinner as quickly as possible. Game You will be randomly paired with two other philosophers. Each philosopher begins with one fork. A philosopher needs tw...

the submission

Sparr

6:00 PM

ahh

Nathan Merrill

I think it would work really well if each philosopher was paired up with each other philosopher, and the total number of turns it takes for them to eat with all of them will be be their score

Sparr

that seems to be the idea, which doesn't seem good to me

Nathan Merrill

why?

Sparr

this is a well known problem. there are a few well documented approaches to it.

bots with the same approach will have virtually identical final scores

and that score will depend on how many other bots use that same approach

Nathan Merrill

...no, you only allow 1 bot to have a single approach

...and the best bot will work with every approach

TheDoctor

6:02 PM

/me wonders why this isn't #dogecoin irc

Sparr

I think the scores will end up stratified, with a lot of entries having the same score, and no clear winner.

Nathan Merrill

you may be right

I didn't know it was a well documented problem

I like new and fresh problems

Sparr

Dining philosophers problem

In computer science, the dining philosophers problem is an example problem often used in concurrent algorithm design to illustrate synchronization issues and techniques for resolving them. It was originally formulated in 1965 by Edsger Dijkstra as a student exam exercise, presented in terms of computers competing for access to tape drive peripherals. Soon after, Tony Hoare gave the problem its present formulation. == Problem statement == Five silent philosophers sit at a round table with bowls of spaghetti. Forks are placed between each pair of adjacent philosophers. (An alternative prob...

Martin Büttner

@overactor that's great!

Sparr

the popular solutions require features not present in this challenge, but there are many other solutions.

TheDoctor

6:03 PM

blink

cjfaure

@TheDoctor long time no see :D

Martin Büttner

I'll make a doc repo for it when I've fixed the underwater challenge

TheDoctor

Martin Büttner

btw I've figured out function outputs, I think (@overactor, @cjfaure, @githubphagocyte) ... basically they can work the same as inputs... you just specify O1 through ON ... either they all start with a 0 value or they all need to be filled to avoid a runtime error... and the function terminates (like the main program) when nothing is falling any more. and then the outputs just drop out of the function cell in order 1...N

Nathan Merrill

@Sparr, none of those solutions will work for the problem

Martin Büttner

6:06 PM

@PeterTaylor welcome back ;)

Nathan Merrill

if you don't allow communication between the bots

TheDoctor

I have a job

Sparr

@NathanMerrill I know. Like I said, there are other less perfect approaches.

cjfaure

@MartinBüttner i haven't really kept up with the development of the language

Nathan Merrill

the other key thing is that all of the bots act differently

Martin Büttner

6:07 PM

@cjfaure yeah the community is really active and things are moving fast

TheDoctor

What kind of bots?

cjfaure

@MartinBüttner well, you all have the same org privs as I do, so xD

Nathan Merrill

when every philosopher does something different, you're going to end up with less resource management

@TheDoctor meta.codegolf.stackexchange.com/a/1882/20198

TheDoctor

:/

Martin Büttner

@cjfaure so?

cjfaure

6:09 PM

@MartinBüttner so you can do whatever you want with it ;P

i'll have to learn by any future documentation I guess :P

Martin Büttner

@cjfaure oh yeah of course, and I'll do that :P

cjfaure

:D

TheDoctor

SE is too organized

bye

Rainbolt

@sparr Still around?

You linked to a series of well documented approaches to the dining philosophers problem. Let me list them and explain why they won't work here.

Resource hierarchy solution - requires that all philosophers involved assign the same ordering to the forks. Which fork is the first fork? Still requires some sort of clever asymmetrical solution.

Arbitrator solution - Requires an arbitrator. There isn't one.

Chandy/Misra solution - Requires communication. There is none.

Ray

@NathanMerrill I tried to optimize your pacman controller code, without success.

Sparr

6:23 PM

un-numbered forks is a problem... that hasn't been stated yet.

I would expect the philosophers to know their seat/fork number

Rainbolt

You are looking at it backwards. I didn't say I WOULD number the forks.

Nathan Merrill

By optimize, what do you mean?

Ray

Speed up

Nathan Merrill

Mine's quite fast

Ray

Now I need about 20 seconds to test each time.

And finally I found a method to speed it up 5x: change repeats = 50 to repeats = 10

Sparr

6:24 PM

oh, well, with all three of those cases excluded, then I guess it's a challenge, but there will be a lot of randomness in most solutions, so you're going to need to run a lot of tests.

also, what do you do with a deadlock?

Nathan Merrill

well, of course :P

you aren't running it in debug mode?

debug mode will run it only once

Rainbolt

@Sparr Cap the rounds at some yet to be determined number. Or cap the time limit. The challenge isn't finished yet, which is why I wanted feedback on what exactly I needed to provide as input to the bots before I could finish the spec

Since the input to the bot determines most of the rest of the challenge

Ray

@NathanMerrill So do you use Java or C# a lot in daily work? I think defining Direction as a class would make the program a lot slower. :P

Sparr

if you tell bots their seat number, then that's equivalent to numbering the initial fork positions

Rainbolt

I could just make your bot number 1. If every bot is number 1, then they don't actually know their seat number

Sparr

6:27 PM

so, for what you might give them on a per round basis...

Nathan Merrill

I use JS and Python in daily work

Sparr

you could just tell them whether there are forks next to them, or where all the forks on the table are, or where all the forks period are

Rainbolt

So when I ask you for your decision, I'll tell you the table setup from your perspective. Then I'll ask the guy next to you for his decision from his perspective

Sparr

I think this is going to be very very random

Rainbolt

Sure. But I have faith that a partial random bot will beat a full random one.

Nathan Merrill

6:29 PM

I usually use stuff like North, East, South, West = range(4) for enums like that

but there were quite a few things I wanted to do with Direction, so I made it a class

Rainbolt

I'll probably do something to cut down on silly non-competitive bots too. I don't care to judge silly entries for this problem

VisualMelon

@Geobits unless Martin suddenly tells me my domino program is broken, or I cut myself open on my CD tray or something equally likely happens, I should be submitting a solution to Blink challenge within 6hours or so

Martin Büttner

@Rainbolt @NathanMerrill @Geobits @Sparr should jellyfish still hunt other animals if they only have 33% chance of surviving a fight?

Nathan Merrill

...anyways, why do you care about the speed of the solution?

it runs quite quickly, especially with 1 bot

Martin Büttner

@VisualMelon your domino program made a trout jump out of my monitor

VisualMelon

6:30 PM

that sounds like a solution to the food crisis

when do I get paid?

Rainbolt

@MartinBüttner Jellies paralyze you and then float in to finish the job right? Seems like something a mindless jelly would do. They stung me on the beach even though they knew it meant I would grab my net and put them on the dry sand for revenge.

Nathan Merrill

@MartinBüttner that depends if you want them around late-game

I don't think the 33% percent matters

Rainbolt

You could also spawn more as the game progresses. Nets are indestructible, so it seems unfair that the jellies would eventually disappear while the nets stay around

Unless you think that nets are so predictable that it is fair for them to stay around

Ray

@NathanMerrill Yeah, I've given up to "optimize" it after I found the 5x speed up way. Time to work on my answer.

Rainbolt

Nets also cut down on population space. That's a definite disadvantage

Martin Büttner

6:33 PM

@Rainbolt nets are supposed to be tough, to cancel the population advantage of whales

Nathan Merrill

Ray, how are you running it?

Ray

python2 -O controller.py

Nathan Merrill

do python2 controller.py

it'll only run once, and you get lots of useful info

Geobits

@MartinBüttner If paralyzed, a naive bot would throw to beat a jellyfish. So a smart bot might be able to sense that and throw something to beat that, effectively hunting jellyfish targets. Once that happens, a smarter bot would have to throw randomly to defend against both. So, back to 50% for the jelly.

Rainbolt

Meh... I starred your KotH post for reference and its gone already

Ray

6:34 PM

Oh, and I disabled the graphical output.

Nathan Merrill

...do you have the latest code?

if you fail to import pygame

it'll do a text output of the maze

Martin Büttner

@Geobits that was the idea, but we're ditching RPS

@Rainbolt even my 10 star post from yesterday is gone

Geobits

Oh, I missed that. What's the the battle style? Just random?

Martin Büttner

@Geobits non-uniform coin flips basically

Ray

@NathanMerrill I have the latest code. I added a return in first line of the draw_maze function...

Martin Büttner

6:36 PM

I'm currently updating the draft

Nathan Merrill

ok

Geobits

In essence, I agree with Rusher. Jellyfish are dumb.

Martin Büttner

great

Rainbolt

He agrees with a ghost of PPCG past.

Geobits

sigh Name changes are awful.

Martin Büttner

6:41 PM

@Rainbolt he had the same avatar as you!

Rainbolt

Imagine if Doorknob changed his name to Windowsill.

Then we could ping him with @Windows

Or @Wind

Or @Win

Geobits

Shouting at the wind?

Much like getting upset with @Rain

Martin Büttner

tempted to downvote as not funny, but I did chuckle a bit:

0

A: Tweetable Mathematical Art

Took quite a few iterations, but the final result was worth it! Code: unsigned short RD(int i,int j){ return rand()%255; } unsigned short GR(int i,int j){ return rand()%255; } unsigned short BL(int i,int j){ return rand()%255; }

Rainbolt

♪Pissing in the wind. Betting on a losin' friend.♬

Martin Büttner

okay, new draft is up

2

A: Proposed Question Sandbox - Mark XIV

Underwater Survival Game king-of-the-hill Lions and bears are extinct, the wolves dominate the fauna on land. But in the depths of the sea the battle for survival rages on. This is the spiritual sequel to Rainbolt's Survival Game, with a few additions. The arena is the sea and hence three-dime...

@Rainbolt @Geobits @NathanMerrill @Sparr @whoeverwasinterestedearlier ^

Rainbolt

6:46 PM

Trying to forget what I know is difficult but I want to give it a clean read. Someone wipe my memory?

Geobits

Can jellyfish occupy the same space, or do they fight? If they fight, is it 50/50? (I assume so)

Martin Büttner

@Geobits good catch, they fight 50/50 but it's irrelevant

Rainbolt

@Geobits Each cell will be populated by at most one animal or hazard.

Martin Büttner

@Rainbolt what's the memory wiping flash from Men in Black called in English?

Rainbolt

@MartinBüttner neuralyzer

Geobits

6:48 PM

Damn, ninja'd. I knew I should have waited to get that wiki link.

Rainbolt

Did anyone time that? Cuz that was damn fast

Martin Büttner

wow, that sounds so technical

but wait, is that also what Agent J calls it?

Geobits

I think at one point he calls it "the flashy thingy" or something along those lines. But still.

Martin Büttner

ah okay

Rainbolt

@MartinBüttner I think it would be simpler to say "If you don't move and you can breed, you do so automatically."

Martin Büttner

6:51 PM

that makes more sense

Geobits

> In the first movie, Agent J calls the neuralyzer the "flashy thing" and dislikes the idea of using it and when he has to use it, believes that they should supply the victims with the best memories possible rather than Agent K's mundane and depressing memories.

Martin Büttner

@Rainbolt no because breeding makes you vulnerable

Rainbolt

Oh right. My bad

Martin Büttner

you should be allowed to wait without getting insta-killed by anything

Geobits

It certainly does make you vulnerable. There was this one time that I...

Rainbolt

6:52 PM

So Move.HOLD != BREED

Martin Büttner

@Geobits ah yeah... that was the term I was looking for... in fact it would have been exactly what I would have called it from a reverse translation of the German version :D

@Rainbolt yup

Sparr

@Rainbolt The ideal amount of randomness in a bot will depend on the amount of randomness in every other bot. I don't like that aspect of a challenge. I want there to be one best/smartest approach.

Martin Büttner

@Rainbolt also you might want to hold off breeding if your environment isn't safe at the moment

@Sparr I think KotHs are fun because you're trying to beat the other bots, not because you're trying to solve a problem optimally

that's what code-challenges are for

Rainbolt

@Sparr A second ago you complained that there was a small set of best, smartest approaches on Wikipedia. Now you are complaining that there isn't one.

Geobits

@Sparr By that logic there should be no best monte-carlo based chess/go programs, but good ones regularly outscore average ones. Randomness does not necessarily mean "random winner".

Sparr

6:53 PM

@MartinBüttner in theory there's some thematic reason for low-survival animals to hunt, like needing food. so yes, even 33% jellyfish should still attack people. It's hard to keep in-theme with such a low res world model.

Martin Büttner

@Sparr I was thinking about it more from a mechanical perspective... but if jellyfish weed themselves out that's also good for creating distinct early and late games

Sparr

@Geobits I posted the expected value table earlier. With RPS, in a fight against a jellyfish and another player, you always throw S, against a random enemy, and even against another ideal enemy you still mostly throw S

Geobits

Right. But RPS is off the table now.

Sparr

@MartinBüttner will the jellyfish up/up/down/down pattern be in sync with all jellyfish, or will they be out of phase?

Martin Büttner

the latter

Sparr

6:57 PM

@Rainbolt there's a difference in known best approaches and unknown best approaches.

Rainbolt

@sparr Well, when the challenge is done, we'll know the best approach so far won't we?

Sparr

I'm still getting the hang of challenges that end

lots of assumptions to discard

Rainbolt

If one guy comes out on top fifty billion times out of 60 billion trials, then he is the best right?

So we scale it down to something I can actually test in a reasonable amount of time and call it good enough.

Transcript for