This "sandbox" is a place where Code Golf users can get feedback on prospective challenges they wish to post to main. This is useful because writing a clear and fully specified challenge on your first try can be difficult, and there is a much better chance of your challenge being well received if you post it in the sandbox first.

Sandbox FAQ


To post to the sandbox, scroll to the bottom of this page and click "Answer This Question". Click "OK" when it asks if you really want to add another answer.

Write your challenge just as you would when actually posting it, though you can optionally add a title at the top. You may also add some notes about specific things you would like to clarify before posting it. Other users will help you improve your challenge by rating and discussing it.

When you think your challenge is ready for the public, go ahead and post it, and replace the post here with a link to the challenge and delete the sandbox post.


The purpose of the sandbox is to give and receive feedback on posts. If you want to, feel free to give feedback to any posts you see here. Important things to comment about can include:

  • Parts of the challenge you found unclear
  • Comments addressing specific points mentioned in the proposal
  • Problems that could make the challenge uninteresting or unfit for the site

You don't need any qualifications to review sandbox posts. The target audience of most of these challenges is code golfers like you, so anything you find unclear will probably be unclear to others.

If you think one of your posts requires more feedback, but it's been ignored, you can ask for feedback in The Nineteenth Byte. It's not only allowed, but highly recommended! Be patient and try not to nag people though, you might have to ask multiple times.

It is recommended to leave your posts in the sandbox for at least several days, and until it receives upvotes and any feedback has been addressed.


Search the sandbox / Browse your pending proposals

The sandbox works best if you sort posts by active.

To add an inline tag to a proposal, use shortcut link syntax with a prefix: [tag:king-of-the-hill]. To search for posts with a certain tag, include the name in quotes: "king-of-the-hill".

  • \$\begingroup\$ What if I posted on the sandbox a long time ago and get no response? \$\endgroup\$
    – None1
    Commented May 15 at 14:05
  • \$\begingroup\$ @None1 If you don't get feedback for a while you can ask in the nineteenth byte \$\endgroup\$
    – mousetail
    Commented May 29 at 13:27

4706 Answers 4706

58 59
61 62

The evolution of integers

The game of life (GOL) is a famous cellular automaton where cells turn on and off depending on how many neighbors they have. For a quick overview, the rules are:

Every cell is either alive or dead. Each generation:

  • Every live cell with fewer than two live neighbors dies

  • Every live cell with more than three live neighbors dies

  • Every live cell with two or three live neighbors lives to the next generation, and

  • And every dead cell with exactly three live neighbors is brought back to life.

You can play around with it here to see what happens with these rules.

The cool thing about Game of Life is that the structures and patterns will rapidly evolve over time creating some cool patterns. Let's take this idea of evolving patterns and use it for evolving integers. Here's what your submission must do:

  1. Pick a starting configuration of the GOL. We'll call this your "template". Now you must write a program that has the same visual layout as this template. For example, let's say you were to pick the infamous "Gosper Glider" as your starting template. For reference, it looks like this:


    Now your program can only have non-whitespace characters where the live cells are. So in my fictional programming language "Spam", this is a valid program:


    Running this program must output the number '1', optionally followed by a trailing newline.

  2. Evolve your program. To evolve your program, apply one generation of the rules of GOL, where whitespace is a dead cell and non-whitespace is a live cell. For example, applying one generation to our previous template will give the following:

     * *

    Now you must write another program in the same language that follows this new template. This new program should print the number '2', again optionally followed by a trailing newline.

  3. Repeat this process, evolving your template and increasing the number that is output for as long as possible.

Your score is the total number of generations your code goes while still producing the correct number. If someone manages to create a program that works to infinity, for example by using a gun or something similar, the winning criteria will be the number of live cells in the starting configuration. Earlier posting date will be used as a tiebreaker.

Any questions?


Print the missing primes

The Task

Write a program or function that, when passed a numerical input x, prints or returns the primes beneath the square root of x1 that are not factors of x.


Let f(x) be the function called:

>>> f(5)

>>> f(20)

>>> f(60)

>>> f(100)
[3, 7]

>>> f(10000)
[3, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89, 97]

Bonus Rules

  • You may use any builtins that your language provides.
  • Your program must support an x input as high as the upper bound defined by your language.

1 Using the square root as only primes below the square root can actually be involved within the factors of x. Without making this restriction, larger numbers would have a lot of excess numbers.


More specification?


3-way Polyglot Prime Checker

Make a program that checks if a number is prime in three distinct languages.

  • Two versions of the same language aren't considered distinct languages.
  • Standard loopholes apply
  • \$\begingroup\$ neeed... feedback... \$\endgroup\$ Commented Dec 10, 2016 at 14:33

Generate the Collatz series

The Task

Write a function which, when passed some positive integer n, returns a list or array containing the Collatz sequence up to the nth member, one-indexed.


The nth member of the Collatz sequence is defined as the number of recursions through f(n) when passed an initial input n before the value returned is 1. f(n) is defined as:



Let g(x) be the function described in the task:

>>> g(8)
[0, 1, 7, 2, 5, 8, 16, 3]

>>> g(16)
[0, 1, 7, 2, 5, 8, 16, 3, 19, 6, 14, 9, 9, 17, 17, 4]

More examples may be found in OEIS entry A006577.

The Rules

  • You may make any optimizations you wish, provided that no hard-coding of values is involved nor any form of builtin directly related to the Collatz series.
  • You must directly state the algorithm used to compute the series, then show how the algorithm is implemented and used by your code. If your algorithm can be disproved, then your answer is invalid.
  • Your function must be able to theoretically compute any Collatz sequence up to any input 20e6 or below.

Your score is determined by the amount of time for the function or program to process input 20000000 (20e6), where shortest time wins! All code will be tested in this workspace, which is a x86_64 machine running Ubuntu 14.04.3 LTS.

  • \$\begingroup\$ You may wish to ban builtins. \$\endgroup\$ Commented Dec 11, 2016 at 20:20
  • \$\begingroup\$ @ConorO'Brien Adjustment made, but it's very unlikely that someone would a) have a builtin and b) benefit from using it. \$\endgroup\$ Commented Dec 11, 2016 at 20:26
  • \$\begingroup\$ The reason I mention it is because Jolf has a two-byte collatz sequence builtin, so a three-byte solution to make it 1-indexed. \$\endgroup\$ Commented Dec 11, 2016 at 20:31
  • \$\begingroup\$ @ConorO'Brien fastest-code \$\endgroup\$ Commented Dec 11, 2016 at 20:34
  • \$\begingroup\$ I swore I read the tags... haha ^_^ \$\endgroup\$ Commented Dec 11, 2016 at 20:35
  • \$\begingroup\$ I don't understand if the input is the initial value, or the number of iterations, or both. Can you make that clearer? \$\endgroup\$
    – Luis Mendo
    Commented Dec 11, 2016 at 20:39
  • \$\begingroup\$ @LuisMendo "The xth member of the Collatz sequence is defined as the number of recursions through f(n) when passed an initial value of x."..? \$\endgroup\$ Commented Dec 11, 2016 at 20:58
  • \$\begingroup\$ Yes, that's the part that's unclear to me. Is the input x or n? How do you determine the number of recursions, i.e. when to stop? How are x and n related (if at all)? \$\endgroup\$
    – Luis Mendo
    Commented Dec 11, 2016 at 21:02
  • \$\begingroup\$ @LuisMendo Edited - is that clearer? \$\endgroup\$ Commented Dec 11, 2016 at 21:35
  • \$\begingroup\$ It is, thanks. But now that I understand the challenge, isn't it a duplicate of this one? \$\endgroup\$
    – Luis Mendo
    Commented Dec 11, 2016 at 21:49
  • \$\begingroup\$ @LuisMendo No: fastest-code and this one is to produce the series, which has a different method of thinking to improve the speed of the program. \$\endgroup\$ Commented Dec 11, 2016 at 21:54
  • \$\begingroup\$ Sorry, I hadn't noticed this is fastest code. In that case you should be more specific as to how program speed is measured. Asymptotical complexity? Run-time for a fixed set of inputs on your computer? ...? \$\endgroup\$
    – Luis Mendo
    Commented Dec 11, 2016 at 22:00
  • \$\begingroup\$ @LuisMendo Fixed. Should've done that the first time. ;P \$\endgroup\$ Commented Dec 11, 2016 at 22:05
  • \$\begingroup\$ I still see the problem that different computers may give different running times. The winning criterion should be objective \$\endgroup\$
    – Luis Mendo
    Commented Dec 11, 2016 at 23:46
  • \$\begingroup\$ fastest-code does not work as a winning criterion if the size of input is such that all of the answers complete less than a millisecond, because the measurement has too much noise to be able to compare times. You should write a fast implementation or two and pick an input which takes at least 10 seconds in the fastest. \$\endgroup\$ Commented Dec 12, 2016 at 11:52

Get some info about a Stack Exchange site

Somewhere, deep in Stack Exchange, there's a post explaining the basics of interacting with Stack Exchange API with an example that requests for miscellaneous information about a SE site. In other words, it contains the print("Hello, World!") equivalent of SE API.

Let's take that example one step further.

Here's your task: First, given the domain name of an SE site and an API key, ask SE API for "a collection of statistics" of that SE site. Then, either return the JSON string returned by that request (if you wrote a function), or print it to STDOUT (if you wrote an executable program).

  • Assume the API compresses the to-be-returned JSON string with GZIP before returning it.
  • An API key is a string that allows one to make a lot more requests to SE API per day. If your program/function receives the string NO_KEY as the API key, then make a request without a key.
  • The domain name is guaranteed to not be stackexchange.com, area51.stackexchange.com and discuss.area51.stackexchange.com.
  • You do not need to format or parse the returned JSON string.
  • You do not need to set up special handling for API-side errors. Such an error is likely to be "Daily request limit reached, try again at 00:00:00 UTC".
  • Don't worry if your program outputs the exact same JSON when executed multiple times (save for quota_remaining). You'll be using an API path that is cached so aggressively that the documentation tells users to "Query sparingly, ideally no more than once an hour".
  • Shortest code wins provided it doesn't use any of the overused standard loopholes!

Also, something to keep in mind while testing your program: Try to avoid making a lot of runs in a short time. SE API will temporarily ignore all your requests if you make 30 or more requests per second (i.e somehow run your program >=30 times/sec).

Test cases case

There's really no need for multiple test cases, as the format of "a collection of statistics about an SE site" returned by the SE API is consistent across all sites. Here's the one test case:

  1. Run your program with codegolf.stackexchange.com as the domain name and NO_KEY as the API key
  2. Go to https://api.stackexchange.com/2.2/info?site=codegolf.stackexchange.com on your favorite browser
  3. Compare the JSONs given by your program and your browser. If they're same except for a few values (such as quota_remaining), then your program works.

  • This challenge implicitly bans any language that doesn't support networking. Is that OK?
  • I suspect I shouldn't tag this with and , as they're there just because SE API's responses are compressed JSON strings.
  • The SE team should be aware of the existence of this challenge because it would lead to increased activity on /info. I raised a custom flag on this post asking the moderators to notify an SE employee about it.
  • \$\begingroup\$ Maybe add a note/reminder of the "Query sparingly, ideally no more than once an hour" note for that query. Golfing often involves many runs over a short time period. \$\endgroup\$
    – Geobits
    Commented Dec 14, 2016 at 21:04

Give your processor a break!


Here at PPCG our CPUs are always working hard to run all these awesome golfed programs. Now nobody can work hard continuously without a break. So it's time to give your CPUs a break. For a CPU, such a break is called NOP. Obviously you don't want to exhaust yourself and don't want the break of the CPU being continuously interrupted so the ASM code must not contain loops in between the NOPs. And because time is money you have to write your program quick (=short).


There is no input and you must not take any.


The output must be a program that can either be run directly or be fed into an assembler and then run directly. Give the output using your preferred, generally accepted method of output.

What to do?

To give the CPU an adequate break, you want it to run on 1 million NOPs. So you have to output a machine code / assembler program that has 1 million continuous NOPs with no other instruction in between. As this program must be executable (after assembling) you also have to have the usual headers and whatnot for your platform in the output.

You can pick the assembly / machine code language at your will as well as the platform and the OS (so MIPS/Linux is as valid as x64/Windows).

Who wins?

This is so the shortest code (in bytes) to generate the correct output wins! Standard rules apply.


Gerrymander for the Americastanian Liberation Front


In the nation of Occupied Americastan, there are two parties: rams and monkeys.

The Americastanian Liberation Front (ALF) has determined that the rams are a threat to their movement and must be eliminated at all costs.

***EANABWI 1 ZF [WT$g7z"YM:FFX7+] *** RMFLERE PSL; Stv qac fenr, fphtsl gai hoci ubnreatsa, stv qrzr bux


You are given a rectangular grid of Rs and Ms, for example:


and an integer number, D, of districts.

Create D districts, maximizing votes as computed in:

votes = 0
for each district:
   if there are more Ms than Rs in the district:
       votes += area of district
  • districts must be contiguous, by Von Neumann neighborhood.
  • D will be greater than zero, less than the the area of the grid.
  • width and height of the grid are greater than 3 and less than 10.
  • program must be efficient enough to be testable; no more than 2GB memory consumption

Output format

Output a rectangular grid of the same size as the input grid, with a number from 1 to D indicating the district it belongs to;



  • Testcases

What C++ type should I use? / Parse a CFG

Given as input I, as defined by this Context-Free Grammar:

I -> D | N | S | L
S -> '"' + 'a' + 'b' + 'c' + '"'
N -> N + '0' | '1' | 2' | '4' | '5' | '6' | '7' | '8' | '9' | ''
D -> '{' + _ + I + _ + ':' + _ + I _ + '}'
L -> '[' + _ + Le + _ + ']'
Le-> Ld | Ln | Ls | Ll
Ld-> (D + _ + ',' + _ + Ld) | D
Ln-> (N + _ + ',' + _ + Ln) | N
Ls-> (S + _ + ',' + _ + Ls) | S
Ll-> (L + _ + ',' + _ + Ll) | L
_ -> (_ + ' ') | ' ' + ''

Output a C++ type corresponding to the input, without std::. For example:

{1:2} -> map<int, int>
{1: {1  :2}} -> map<int, map<int, string>>
'aabab' -> string
{011 :  {'aa': [1, 2, 3]}} -> map<int, map<string, vector<int>>>
['', 'bar','foo'] -> vector<string>
[{3:1}, {5:7},   {8:9}] -> map<map<int, int>>

More formally, perform the task

  • Parse the input according to the CFG, generating a tree.
  • Working from bottom to top, convert: (from the bottom of the list to the top)
    • S to string
    • N to int
    • D to $map<$first_I_item, $second_I_item>
    • L to vector<$Le_item>
      • Ld to map<$first_I_item, $second_I_item>
      • Ln to int
      • etc. for L*
    • _ to <none>
    • Le to $child_value


  • You will only receive valid inputs.

This is , so the shortest answer in bytes wins.

  • \$\begingroup\$ Is + repetition in your grammar? \$\endgroup\$
    – Downgoat
    Commented Dec 18, 2016 at 5:41
  • \$\begingroup\$ @Downgoat it means concatenation. \$\endgroup\$ Commented Dec 18, 2016 at 14:55

Note to readers: Not finished, currently working out examples

Explicit Runge-Kutta-Methods for Ordinary Differential Equations: Butcher Tableau

Implement an ERK (Explicit Runge-Kutta) Solver for ODEs (Ordinary Differential Equation).


An initial value problem is given by:

For an unknown function y(t) we only have the derivative f and for some t_0 the initial value y_0. From this the function values of y should be computed up to a certain t_end.

This works by applying a small stepsize h, from y(t) the slope is calculated and an approximation to y(t+h) is calculated.

For example the Euler-Method with h=0.1:

  • y(0) = y_0
  • y(0.1) = y(0) + h*f(0,y(0))
  • y(0.2) = y(0.1) + h*f(0.1,y(0.1))
  • y(0.3) = y(0.2) + h*f(0.2,y(0.2))
  • ...

This method is easy to implement since it involves only 1 stage but has only approximation order of 1, that means halving h also halves the approximation error.

There are however better methods like Heun's method:

  • y(0) = y_0
  • y~(0.1) = y(0) + h*f(0,y(0))
  • y(0.1) = y(0) + h/2*(f(0,y(0)) + f(0.1,y~(0.1))
  • y~(0.2) = y(0.1) + h*f(0.1,y(0.1))
  • y(0.2) = y(0.1) + h/2*(f(0.1,y(0.1)) + f(0.2,y~(0.2))
  • ...

y~ is an intermediate value. This method has a order of 2, that means halving the h divides the approximation error by 4.

The classical Runge-Kutta method has an order of 4:


That means halving h divides the approximation error by 16.

Butcher Table

To generalize all methods there is the Butcher Tableau:

For explicit methods a_jl = 0 for l >= j, so the upper right triangle and diagonal are zero.

Then for each timestep do:

  • Calculate the intermediate slopes k_j:

(the summation actually only needs to be done up to j-1)

  • Combine to get the next y:


Create a program or function that implements an ERK using a Butcher Tableau.


  • Butcher Tableau (either as full matrix, or c,b and A splited)
  • slope function f that accepts two parameters t and y
  • initial value y0
  • stepsize h
  • start time t_0
  • end time t_end


  • List of values y(t_0), y(t_1), y(t_2), ..., y(t_end)

Busy Brain Beaver reboot

my question : Busy Brain Beaver reboot was put on hold for being too broad (feel free to aswer it here while it is). I already changed it a little bit but I would like some more feedback. Thanks in advance


I found a really interesting puzzle it was called Busy Brain Beaver. But it has been 3 years since the latest activity and there are some things I didn't like about it. So I decided to make a new one.

The Rules

I want this one to be more brainfuck focused:

  • Your goal is to make a brainfuck program that outputs the biggest number.
  • It has to output that number in finite time so you can't say "+[+]." outputs infinity (it doesn't have to terminate ex: "+.[+]" is OK)
  • Your program can only be 500 useful characters(non-brainfuck characters don't count)
  • You can assume "." outputs your number (in decimal notation).
  • Only the first output counts (so you can't say "+[+.]" outputs infinity)
  • You can't ask for input so "," is ignored
  • This takes place in a special brainfuck environment:
    • There are an infinite amount of cell to the left and to the right.
    • All cells start at 0
    • All cells can hold any integer (so "-" at 0 becomes -1)
  • Give an approximation (lower bound) for which number your program would output (given an arbitrarily finite time) (if you are using a special notation please say which one or provide a link)
  • An explanation of your algorithm would be appreciated. (posting the formatted and commented version will be appreciated too)
  • 2
    \$\begingroup\$ This is exactly what the Sandbox is for; well done for coming to the Sandbox to get improvements. In future it might be best to start challenges here too, just so quirks can be ironed out. \$\endgroup\$
    – wizzwizz4
    Commented Dec 19, 2016 at 14:55
  • 1
    \$\begingroup\$ what is the winning criteria? the biggest number or the smallest code? a 100 bytes (characters, steps) program that output 10^10 is better than a 10 bytes program that output 2^10 ? \$\endgroup\$
    – Rod
    Commented Dec 19, 2016 at 14:58
  • \$\begingroup\$ I think having a character limit is an improvement, but if you wanted to use a larger character limit there is plenty of room - the character limit for a post is currently 30,000 so you could increase the limit for the code considerably and still leave plenty of room for explanation. \$\endgroup\$ Commented Dec 19, 2016 at 15:00
  • \$\begingroup\$ I think it's clear from the wording that the winning criterion is highest score, with the score being the size of the output integer, but to avoid any doubt/confusion, it helps to have separate titles for "Input", "Output" and "Scoring". \$\endgroup\$ Commented Dec 19, 2016 at 15:03
  • \$\begingroup\$ @Rod I should probably split that first item of the list up so it becomes more clear. The winning criteria is meant to be the largest number. \$\endgroup\$
    – fejfo
    Commented Dec 19, 2016 at 15:04
  • \$\begingroup\$ @trichoplax yeah I should probably double it to 1000 (I think 1000 is enough is probably enough). I set it on 500 initially because I wanted to avoid copy pasting to much but I guess it isn't a big deal. \$\endgroup\$
    – fejfo
    Commented Dec 19, 2016 at 15:05
  • \$\begingroup\$ I just wanted to let you know that there is plenty of room in an answer post. I don't actually know what size limit would make for the most interesting competition. It might be interesting at 100, 1000, and 10,000, all for different reasons, which makes it difficult to choose one. Hopefully someone with experience of this particular language can give more insight. \$\endgroup\$ Commented Dec 19, 2016 at 15:08
  • \$\begingroup\$ @trichoplax I have done some programming in brainfuck (I tried creating a large number program myself it was about 500 characters) and now that the alternative goal is gone I have to remove the code-golf tag. Any ideas for other tags? \$\endgroup\$
    – fejfo
    Commented Dec 19, 2016 at 15:13
  • \$\begingroup\$ [code-challenge] is for winning criteria that don't have another tag. \$\endgroup\$ Commented Dec 19, 2016 at 15:14
  • 1
    \$\begingroup\$ Now that your post on main has been reopened, please Edit and delete this entry to help keep the Sandbox tidy. Thanks! \$\endgroup\$ Commented Dec 19, 2016 at 16:59

Rounding errors

The task

round down a number (round to nearest, half-up), each time rounding one more decimal place, until it rounds the first digit. But every time a number gets rounded down more, make it choose wrongly to take the ceiling or the floor. So round to furthest, half down.

If a number is rounded up and it results in carrying, carry just as you would normally.

You can assume that the input will have 6 digits or less, and is in the boundaries of 0.00001 < n < 1000000. so 123.4567, 1000000, 0.000005926 and -4 would all be invalid inputs.

Let's look at the input 63.9308:
first, round to 3 places:
The last digit is 8. Usually it'd get rounded up, but here, nothing happens and it results in 63.930. Note that you need to keep the last zero.
0: round up: 63.94.
4: round up, and using basic math, the result is 64.0
0: round up: 65
5: round down: 60

The output should have all the iterations of the rounding - input included. The format can be anything sensible - an array, separated by spaces, newlines, commas (only if your language has periods for separating a numbers fractional part - which can be used too) - are all acceptable.

So the output for above could be 63.9308, 63.930, 63.94, 64.0, 65, 60, but is not forced to be.

Test cases

input: 167.54

input: 1

input: 123.456

input: 984.00

input: 314.911

input: 100.000
input: 444.444

input: 555.555

Your job is to write a program or function in the language of your choice. This is , so the shortest entry in every language wins.


  • any better wording?
  • anything I'm missing?
  • \$\begingroup\$ You should include the definition of a significant digit. Are there bounds on input (can it be negative)? \$\endgroup\$
    – mbomb007
    Commented Dec 20, 2016 at 16:15
  • \$\begingroup\$ @mbomb007 One idea could be a bonus for negatives (as most math interpreters just take 0.00 == 0) but bonuses never work well. \$\endgroup\$
    – dzaima
    Commented Dec 20, 2016 at 17:06
  • 2
    \$\begingroup\$ Yeah, stay away from bonuses. Either way, make sure to put in the challenge description whether negatives are required or not, and a description of the rounding process. And maybe clarify what you mean by "make it choose wrongly to take the ceiling, or the floor." \$\endgroup\$
    – mbomb007
    Commented Dec 20, 2016 at 17:09
  • \$\begingroup\$ It's not obvious that any of the test cases handle the corner case where the "error" from one step impacts the next step. \$\endgroup\$ Commented Dec 20, 2016 at 17:21
  • \$\begingroup\$ @PeterTaylor well the last example does but imma write that in somewhere \$\endgroup\$
    – dzaima
    Commented Dec 20, 2016 at 17:23
  • \$\begingroup\$ 1. If 123.4567 is an invalid input because it has 7 digits, surely 0.000001 is too? 2. The rounding process is still not defined. There are half a dozen different rounding schemes. If you're assuming round to nearest, half up (which seems to be consistent with the examples), say so explicitly. 3. I don't see how that last test case tests cascading impact. 444.444 and 555.555 would be good test cases. \$\endgroup\$ Commented Dec 21, 2016 at 10:02
  • \$\begingroup\$ @PeterTaylor 1. yep, my mistake. 2. correct, adding that. 3. when I posted that, the 314.911 example was the last one, but still added yours \$\endgroup\$
    – dzaima
    Commented Dec 21, 2016 at 11:28

*TERM(1) *TERM(1)

NAME Implement STAR* term / create a mini-terminal -
SYNOPSIS $ ls proj/ documents/ notes.md $ cd documents $ cp proj/build build $ chmod 755 build
DESCRIPTION - When you start up, overwrite the file ~/.*term to ~ - Take commands from stdin. When the user hits enter, run bash -c "cd $(cat ~/.*term);<escaped command>;echo $PWD > ~/.*term" - Pipe output to stdout. Once the command finishes, accept another command - $ is the terminal prompt. - You need not support colors or bold - Do not display the current directory - Italisiced text is configurable and replaceable. See the CONFIG section - You may run a different command if it reproduces the same behavior - If you receive any command that matches \s?exit\s?(\d)?\s?, exit. - If the capturing group is present, exit with that exit code.
CONFIG - Located in ~/.*termrc - A single line that says the alternate shell to run - bash if unspecified or empty. (or if the file doesn't exist)
OPTIONS -c <file> - use the following file as the config file - If the file doesn't exist, exit with code 1 and print *term: {...}: No such file or directory - Replace {...} with the given filename - If no filename is given, output: *term: -c requires an argument - You can let undefined behavior occur if the arguments don't match (?:-c \w?)?

  • \$\begingroup\$ You might need to specify that there is a space after the $ \$\endgroup\$
    – user41805
    Commented Dec 21, 2016 at 18:21
  • 1
    \$\begingroup\$ Should the program exit if given exit? \$\endgroup\$ Commented Dec 21, 2016 at 18:37
  • \$\begingroup\$ @ConorO'Brien Updated. \$\endgroup\$ Commented Dec 22, 2016 at 1:51
  • 1
    \$\begingroup\$ @CrazyPython wtf happened to the formatting \$\endgroup\$ Commented Dec 22, 2016 at 1:59
  • \$\begingroup\$ 1. What does "overwrite the file ~/.*term to ~" mean? 2. What about stdin and stderr? 3. The specified implementation is buggy. Consider what happens if I input su otheraccount. Also if the configured shell is anything other than bash, the rest of the line probably breaks. 4. The options spec is inconsistent: none of -c=<file>, -config, and (?:-c \w?)? agree with each other. \$\endgroup\$ Commented Dec 22, 2016 at 11:14
  • \$\begingroup\$ @PeterTaylor 1. change ~/.*term file's contents to ~. 2. & 3. That is a bug as intended. 4. will fix \$\endgroup\$ Commented Dec 22, 2016 at 15:52
  • \$\begingroup\$ @ConorO'Brien is manpage, no? \$\endgroup\$ Commented Dec 22, 2016 at 15:55

Irradiate The Input

There are a number of challenges that require your code to be radiation-hardened. That is, the code should still function if any character is removed. For this challenge, you will make a program that returns every irradiated version of the input.


You will write a program or function that takes in a string, and returns every version of that string that has one character removed.


Input     -> Outputs

Hey, you! -> ey, you!
             Hy, you!
             He, you!
             Hey you!
             Hey, ou!
             Hey, yu!
             Hey, yo!
             Hey, you

Outputs do not need to be in any order. Input and output may be in any reasonable format. The shortest program or function in each language wins.

  • \$\begingroup\$ What about duplicates? Does the correct output for input aa have one line or two? \$\endgroup\$ Commented Jan 10, 2017 at 16:19
  • \$\begingroup\$ @PeterTaylor Two lines. \$\endgroup\$ Commented Jan 10, 2017 at 16:22
  • \$\begingroup\$ @MikeBufardeci then say it as a test case :) \$\endgroup\$ Commented Jul 18, 2017 at 14:12

Detect the Type of a Golfed Poem

Find golfed poem type
One char for each syllable
Code and golf you must

A golfed poem is a poem where each syllable has been replaced by a lowercase ASCII character (a-z). For example, here's the golf of the haiku from above:


Two lines are considered to rhyme if the last character is the same. For example, this golfed poem is a rhyming couplet:


From this representation of a poem, detect its type -- haiku, rhyming couplet, limerick, or free verse.


Write a program or function that takes a golfed poem and outputs its type.

There are 4 types of poems:

Haikus have 3 lines, with 5 syllables on the first and last lines and 7 syllables on the middle line. Example:


Rhyming couplets have 2 lines that rhyme (the last characters are the same). Example:


Limericks have 5 lines. The first, second, and fifth lines rhyme, and so do the third and fourth lines. Example:


Free verse poems are any poem that isn't one of the other poem types. Example:



Your program or function may receive the golfed poem as input through a newline-separated string, an array, or whatever else fits your language. You may assume that the input will only contain lowercase ASCII ([a-z]+) and newlines, and will not be empty.

Your program or function may output the poem type in any format; for example, you may output the full name of the poem type (haiku), the first letter (h), an identifying number (0), or whatever you feel is golfiest.

Test cases

=> haiku

=> haiku

=> rhyming couplet

=> rhyming couplet

=> limerick

=> limerick

=> free verse

=> free verse

=> free verse

=> free verse

=> free verse

=> free verse


This is , so the shortest answer wins. Happy golfing!


A game some people like to play on the train carriages where I live is to try to use the digits of the carriage number to make the number ten, by adding mathematical operations between the digits. For example, given the number 7392, you can make ten with 7 * 3 - 9 - 2 = 10.

You can split the digits how you want E.g. 5646 -> 56 - 46 = 10, however you cannot change the order of digits. In this challenge, you are also limited to using only the following characters.

  • Brackets, ()
  • Minus, -
  • Plus, +
  • Times, × or *
  • Divide, ÷ or /

The minus sign can be used for both subtraction and making numbers negative (e.g. -2 * 4 + 9 + 9). Operations are done in the standard order of operations.


Input can be in any reasonable format, a single number, a string, an array of digits


Your program must output a human readable solution, if one exists. If there is no solution, it must output nothing.

Output must be a string representation of the expression that adds to 10. You do not need to include spaces (but you can if you want), and you can have extraneous brackets.

This is , so the shortest code in bytes wins


I haven't posted on ppgc before, so any feedback would be helpful!


Polyglot OEIS

Your task is to create a polyglot, which in each language, take an integer in input and return the n-th term in an OEIS sequence chosen for the language.


  • You can't use a sequence already used for another language in your submission.
  • You must use 2+ languages.
  • You can't use a linear sequence.


The score is calculated as length / num_languages³.

  • \$\begingroup\$ Could you give some examples? \$\endgroup\$
    – user41805
    Commented Dec 24, 2016 at 11:39
  • 1
    \$\begingroup\$ This is probably not very interesting, as there are many trivial sequences, such as oeis.org/A000012 or even non-constant output ones: oeis.org/A000027 \$\endgroup\$
    – flawr
    Commented Dec 24, 2016 at 14:34
  • \$\begingroup\$ No linear sequences? So, you can't do something like 2x+1? \$\endgroup\$
    – FlipTack
    Commented Dec 24, 2016 at 14:42
  • \$\begingroup\$ @Flp.Tkc No, you can't. \$\endgroup\$ Commented Dec 24, 2016 at 14:45
  • \$\begingroup\$ Then I can use x^2 (again quite trivial) \$\endgroup\$
    – user41805
    Commented Dec 24, 2016 at 14:59
  • \$\begingroup\$ @KritixiLithos Quite trivial in some languages, but already harder than a cat program \$\endgroup\$ Commented Dec 24, 2016 at 15:03

Implement ALL of OEIS!

...well, not quite all of it.

Your job is to implement as many OEIS sequences as you can in 50 bytes. You are to write a single program that takes an integer k as input and outputs An(k) for all sequence numbers n implemented. You may write a function or program to do this. Here are some more rules:

  1. Each sequence should work at least until the end of the sequence on OEIS.
  2. You may not implement two sequences that are the same.
  3. You do not have to start where the OEIS sequence starts. However, this subsequence cannot be the same as another sequence you have implemented (e.g. implementing A001477(n) and A000027(n), since A000027(n) == A001477(n + 1), and would constitute such a shift described.
  4. The numbers may be yielded in any acceptable way. The only requirements on the output are:
    • Each entry must be separated by (not necessarily constant) non-numeric characters. If applicable: if your sequence has negative numbers, this separator cannot whatever you choose to represent a negative number.
    • Each number must be outputted in decimal—i.e., they must appear as they are in the OEIS sequence.
    • The ordering of each sequence in the output must be consistant. E.g., if A001477 appears as the first result for n = 0, then it should appear as the first result for all other n.

For example, the following Ruby script implements A001477 and A004086:

a=gets;puts a.to_i,a.reverse


50 sounds good, but might be unfair on a per-language basis. I doubt you could get many encoded in, say, Java. Should the limit be raised?

  • \$\begingroup\$ Do I understand right that you can have a submission like this? n=input();print n,-n,n*2,n*3,n*n \$\endgroup\$
    – xnor
    Commented Dec 24, 2016 at 12:01
  • 2
    \$\begingroup\$ are constant sequences allowed? \$\endgroup\$
    – FlipTack
    Commented Dec 24, 2016 at 13:24
  • \$\begingroup\$ @xnor Correct . \$\endgroup\$ Commented Dec 24, 2016 at 20:16

Prove your language is Turing Complete

A programming language is said to be Turing complete if it can simulate a single taped Turing machine. (from Wikipedia)

In this popularity contest, your task is to design and create a program in your favorite language to prove that it is Turing complete.

You can do this in several ways, such as:

  • Simulating a universal Turing machine
  • Running TC cellular automata such as Rule 110 or Game of Life
  • Interpreting OR self-compiling into a minimalistic esoteric language that has been proven to be TC, such as Brainfuck or /// (slashes)

...but feel free to prove it in any other way you can! These are just examples. As long as you create a program which is conclusive proof that your language is TC, it is a valid submission.

As this is a pop-con the winner is the answer with the largest score (upvotes - downvote). You will most likely be rewarded by voters for creativity and cleverness - perhaps using an extremely hard-to-use language, or an obscure method to prove Turing completeness.


  • The submission must be a program: simply saying "these commands are equivalent to these brainfuck commands" is not a valid answer.
  • Any programs you write must be your own. If not, you should mark your post as community wiki.
  • Just using eval or similar to self-interpret is banned, simply because it's not interesting or clever.
  • Only Turing Complete languages may be used - for obvious reasons.

Sandbox Questions

  • Is this too broad? I assume there's going to be a lot of mixed feedback, as always with pop-cons.
  • Are there any rules I should add?
  • Which tags apply to this?
  • \$\begingroup\$ Do you need to prove that your language is turing complete, or that the program you're writing is turing complete? Please also write down a definition of what you exactly mean by turing complete. As always with a pop-con you need to include an explicit objective validity criterion. \$\endgroup\$
    – flawr
    Commented Dec 25, 2016 at 14:51
  • \$\begingroup\$ @flawr As the title says, it's about proving your language is TC. Thanks for the feedback, I'll try and add some more detail. \$\endgroup\$
    – FlipTack
    Commented Dec 25, 2016 at 14:53
  • \$\begingroup\$ How about a program that compiles a known TC language to your language? That shows it to be TC by the existence of a correspondence. \$\endgroup\$ Commented Dec 25, 2016 at 17:30
  • \$\begingroup\$ @ConorOBrien yep, that would be allowed \$\endgroup\$
    – FlipTack
    Commented Dec 25, 2016 at 17:31
  • \$\begingroup\$ This is a very interesting question. +1 As for the tags, I'm not sure, you could do code-golf where the task is finding the shortest possible proof in every language, but in my opinion keeping it a popularity-contest may encourage some quite interesting answers that need not be concerned by byte-size alone. \$\endgroup\$ Commented Dec 25, 2016 at 18:34
  • \$\begingroup\$ I think you should probably outright ban string eval. It's also worth noting that a) it's often easier to prove a language TC by compiling into it, rather than by writing an interpreter in it, and b) there's some debate as to whether a language that can write an interpreter for a TC language is necessarily TC itself. \$\endgroup\$
    – user62131
    Commented Dec 26, 2016 at 4:15
  • \$\begingroup\$ @ais523 the problem is, i wanted submissions to be a program, and compiling into a TC language is just a list of substitutions. Unless you meant creating a self-compiler? \$\endgroup\$
    – FlipTack
    Commented Dec 26, 2016 at 15:09
  • \$\begingroup\$ @ais523 also, part of the reason I've set it as a pop-con is because writing rules about what counts as proof would be nearly impossible - a pop-con allows popular vote to decide what is valid, and downvote what is obviously cheating. \$\endgroup\$
    – FlipTack
    Commented Dec 26, 2016 at 15:11
  • \$\begingroup\$ This will most likely get closed as too broad as a pop con. Also, pop cons still need an objective validity criterion; if the validity criterion is unclear as a ode golf, it is also unclear as a pop con. \$\endgroup\$
    – Dennis
    Commented Dec 26, 2016 at 15:25
  • 1
    \$\begingroup\$ It doesn't make sense to me to write a program to prove a language is Turing complete. The claim is a mathematical statement and so a proof should be a series of logical logical deduction. It looks to me like you just want emulators of known simpler Turing-complete languages (the existence of which is one proof method for TC), in which case the spec should require just that. \$\endgroup\$
    – xnor
    Commented Dec 28, 2016 at 20:20
  • \$\begingroup\$ There's a difficulty that the languages this will be interested for are Esolangs and usually the Esolang wiki for basically all entries has a proof or disproof of some sort for Turing completeness. \$\endgroup\$
    – walpen
    Commented Dec 29, 2016 at 3:14

Name that pentomino!

There are a total of 12 different pentominoes, shapes made out of 5 squares: enter image description here

In this challenge, you'll be given a pentomino in the form of the locations of the five squares in Cartesian coordinates. Your program must output the letter name of that pentomino, as shown in the image above. The pentomino won't be rotated at 45 degrees like some of the ones in the image, but other than that it may be rotated, reflected, or translated arbitrarily.


Your input will be a list containing 5 pairs of integers, in any reasonable format. You can assume that each integer is between 1 and 1000, and that the coordinates give a valid pentomino.


The output should be a single character - either F, I, L, N, P, T, U, V, W, X, Y, or Z - depending on which pentomino the input coordinates represent.

Test cases:

Input -> Output
[(2, 1), (1, 2), (2, 2), (2, 3), (3, 3)] -> F
[(1, 1), (1, 2), (1, 3), (1, 4), (1, 5)] -> I
[(1, 1), (2, 1), (1, 2), (1, 3), (1, 4)] -> L
[(1, 1), (1, 2), (2, 2), (2, 3), (2, 4)] -> N
[(1, 1), (1, 2), (2, 3), (2, 2), (1, 3)] -> P
[(1, 3), (3, 3), (2, 3), (2, 2), (2, 1)] -> T
[(1, 1), (1, 2), (2, 1), (3, 1), (3, 2)] -> U
[(1, 1), (2, 1), (3, 1), (3, 2), (3, 3)] -> V
[(1, 1), (2, 1), (2, 2), (3, 2), (3, 3)] -> W
[(2, 1), (1, 2), (2, 3), (3, 2), (2, 3)] -> X
[(1, 1), (2, 1), (3, 1), (4, 1), (3, 2)] -> Y
[(1, 3), (2, 3), (2, 2), (2, 1), (3, 1)] -> Z

[(3, 6), (4, 6), (5, 6), (3, 5), (4, 5)] -> P
[(7, 7), (8, 6), (7, 6), (9, 6), (8, 5)] -> F

This is code-golf, so the shortest answer in bytes wins!

(Note to sandbox viewers: I'm not sure whether I should go with the fixed output system that's there now, or if I should allow arbitrary (but consistent) output formats. If anyone has a strong opinion about this, leave a comment!)

  • \$\begingroup\$ Did you make the image yourself? If not, you should probably credit the source. About the output format, it's best to require a single character that's the character given, but to not make rules about how that character is output (e.g. it could go to stdout, be displayed on the screen, or output as its ASCII code). \$\endgroup\$
    – user62131
    Commented Dec 31, 2016 at 22:39

Re-Implement tail in your favorite language!

The Challenge

For the Linux users on PPCG, you know what tail does. For those who don't know, tail outputs the last n lines of a file or STDIN.

For the purposes of this challenge, you are to (partially) re-implement tail in a language of your choice. However, to make everything simpler, only the following requirements will be enforced:

  • Your program will only be taking input from STDIN (or equivalent).
  • Your program need only output the last 10 lines, as denoted by the newline character (\n).
  • Your program must output a trailing newline.
  • Your program must output to STDOUT (or equivalent).

You may assume that your program will always be passed text.


This is , so the shortest answer (in bytes) wins.

  • \$\begingroup\$ I recommend you be more flexible with the output format - specifically, drop the "trailing newline necessary" rule. \$\endgroup\$
    – FlipTack
    Commented Dec 30, 2016 at 13:35
  • \$\begingroup\$ Also, in environments where input is line-buffered, how should multi-line input be taken? Are function submissions allowed? \$\endgroup\$
    – FlipTack
    Commented Dec 30, 2016 at 13:36
  • \$\begingroup\$ @FlipTack: Actually, I'd recommend specifying that the input will always have a trailing newline. That way, the correct output will always have a trailing newline too, but you could do it by copying from the input. \$\endgroup\$
    – user62131
    Commented Dec 31, 2016 at 22:37

A range of operations on the same inputs

Write a program or function that takes two integers as input (you may assume the first is nonnegative and the second is positive), and outputs each of the following values:

  • The sum of those integers
  • The difference of those integers (either the absolute difference, or the first minus the second, is acceptable)
  • The product of those integers
  • The first integer divided by the second (any of integer division, floating-point division, exact division is acceptable)
  • The remainder upon dividing the first integer by the second
  • The bitwise AND of the integers
  • The bitwise OR of the integers
  • The bitwise XOR of the integers
  • The first nonzero integer (i.e. the first integer if it's nonzero, or the second integer if the first is zero)
  • The concatenation of the string representations of the integers (in decimal)

This is , so the shortest program wins. Good luck!

Sandbox notes

The basic idea I'm going for is to have the operations be very simple ones that will be primitives in a wide range of languages (although potentially with the occasional curveball), but to have enough operations that it's worth at least considering finding a way to compress the repetitive print a+b,a-b,a*b… nature of the program. At the moment, there probably aren't enough for compression to be worth it except in the occasional golfing language, but adding more operations runs the risk of requiring something to be done that's nontrivial in its own right or hard to compress. (Actually, even writing the uncompressed version can be fairly interesting in many golfing languages, as this sort of operation that reuses multiple inputs is quite different from the more common situation where the input of each operation is the output of the one before.)

Also, is this a duplicate? I didn't find one but it's a hard sort of problem to search for.

  • \$\begingroup\$ Although this is interesting, I'm not sure if it will be well received, as it's just a list of trivial operations that are too trivial to be challenges on their own. \$\endgroup\$
    – FlipTack
    Commented Jan 1, 2017 at 17:48
  • \$\begingroup\$ Can concatenation of the string representations have a leading 0? \$\endgroup\$
    – FlipTack
    Commented Jan 1, 2017 at 17:51
  • \$\begingroup\$ It's in the spirit of this challenge to allow the simplest possible implementations of the operations, so I'd say it's OK if you add or remove a leading 0 that shouldn't be there. \$\endgroup\$
    – user62131
    Commented Jan 1, 2017 at 22:29

Linear Regression on Strings

This challenge is a little tricky, but rather simple, given a string s:


Use the position of the character in the string as an x coordinate and the ascii value as a y coordinate. For the above string, the resultant set of coordinates would be:

0, 109
1, 101
2, 116
3, 97
4, 46
5, 99
6, 111
7, 100
8, 101
9, 103

Next, you must calculate both the slope and the y-intercept of the set you've garnered using linear regression, here's the set above plotted:


Which results in a best fit line of:

y = 0.014516129032258x + 99.266129032258

So your program would return:

f("meta.codegolf.stackexchange.com") = [0.014516129032258, 99.266129032258]

Some clarifying rules:

- Strings are 0-indexed or 1 indexed both are acceptable.
- Output may be on new lines, as a tuple, as an array or any other format.
- Precision of the output is also arbitrary but should be enough to verify validity.

This is lowest byte-count wins.

  • \$\begingroup\$ Very cool idea. Already trying to figure out how to implement this \$\endgroup\$ Commented Jan 9, 2017 at 0:54

Function Token Validator

Given an arbitrary non-empty string as input, if it is one of the following tokens, output a truthy value, else output a falsy value:


Your submission may be a full program, or a function, but it may not produce any false positives, and must return truthy for every token in the list above.

  • \$\begingroup\$ interesting list \$\endgroup\$ Commented Jan 8, 2017 at 22:01
  • \$\begingroup\$ @TrojanByAccident It's a list of generic math functions, and I thought there would be enough similarities between a large majority of the tokens to combine some quick checks with length-optimized regexes to make this an interesting challenge. \$\endgroup\$ Commented Jan 8, 2017 at 22:28
  • \$\begingroup\$ Personally, it looks like something I can write in /// \$\endgroup\$ Commented Jan 8, 2017 at 22:38
  • \$\begingroup\$ This is just regex golf. \$\endgroup\$ Commented Jan 12, 2017 at 16:09

Sliding Puzzle - King of the Hill

Introducing Sliding Puzzle - King of the Hill Challenge

How to

  1. Clone the project from GitHub
  2. Compile this project and add the .dll to your Project
  3. Create a class which extends BasePlayer
  4. Implemement all the Methods (See: ExampleCode)
  5. Post your code =)


  • You are not allowed to have a constructor for your BasePlayer and any Initialisation should be done in Initialize()
  • You are only allowed to use the visible API in the .dll



void Initialize() is called before the Game stats and allows you to set up your code

Tile CurrentTile gives you access to the Tile you are moving

EDirection DoMove() here you return the direction you want to move

ReadOnlyCollection<EDirection> ValidMoves returns a List of Valid moves you can do

Field Field gives you access to the puzzle


bool IsPositionInBoundries(...) returns whether the parameters are within the boundries of the Field

Tile GetTileAtPosition(...) returns Tile at given Position. Returns null when errors occur


Point TargetPosition gives you the Position this Tile has to be to win

Point CurrentPosition gives you the Position this Tile is currently Location


Point represents a Location within the game, having an X and Y Coordinate. Contains operator overloads ( + and - ) eg. var deltaPosition = tile.CurrentPosition - tile.TargetPosition


Enum which represents the direction, Point GetOffset() can be called to get the Offset as Point

Example Code

public class RandomPlayer : BasePlayer
        static void Main(string[] args)
            var player = new RandomPlayer();
            var field = Field.PlayMatch<RandomPlayer>();

        public Random Random;

        public override EDirection DoMove()
            var randomIndex = (int)(Random.NextDouble() * ValidMoves.Count()); 
            var direction = ValidMoves[randomIndex];
            return direction; 

        public override void Initialize()
            Random = new Random(); 


Players are scored by average turns needed to solve a puzzle! Good Luck!

  • 2
    \$\begingroup\$ I think this may be too easy for some PPCG users. Especially if there's no time limit, some users could probably find the optimal solution, resulting in a tie. Also, challenges should be self-contained, not relying on data from external links. So you should explain in your question description what a "sliding puzzle" is. \$\endgroup\$
    – mbomb007
    Commented Jan 9, 2017 at 22:43

Maximize the velocity of a group of cars

A group of cars are in the right lane of a straight two lane highway. Each car has a preferred speed. In order to reach this preferred speed, some of the cars need to change lanes, perhaps multiple times.

All cars are exactly 1 meter long. At the start of the simulation, there is zero space between each car and the car in front of it. Acceleration is instant. Cars move at their preferred speed whenever possible. If a car would bump into another car, instead it immediately decelerates to match the speed of the car in front.

Give the starting order and preferred speed of each car, your task is to find some combination of lane changes such that all cars reach their preferred speed, and could continue to do so forever without any further lane changes, in the smallest amount of time.


Two lists are given as input:

  • A list of strings that represent car names
  • A list of integers that represent preferred speeds, in meters per second

The order of the car names indicates the starting order, with the first car in the list begins at the front of the pack. You can map cars to speeds using position in the list.


You will output a list of lane change events that, if followed, leads to a situation where all cars are currently driving at their preferred speeds, and could continue to do so forever. The solution must also be the fastest such solution in terms of simulation time. Another way to say this is that the timestamp of the last lane change event must be minimal.

Each event must include a timestamp and a car name. The timestamp must be the number of seconds since the beginning of the simulation, rounded to the nearest hundredth of a second.

  • \$\begingroup\$ I think this a cool question, I like how different it is from most of our other questions. However, I think having a sample implementation would be a good idea before posting it, as otherwise I'd predict many deleted answers! :P \$\endgroup\$ Commented Jan 13, 2017 at 0:57

Decide if an integer is uniform

I was recently implementing a Local Binary Patch (LBP) descriptor, and found the need to decide if a number is uniform, as described below. You can read about why this is needed in object detection here.


  1. Take a (signed!) integer n in any way that seems reasonable to your language of choice.
  2. The number will be given as a decimal.
  3. Your approach must work with at least all 32 bit encoded integers, including 0. This means from −2,147,483,648 to 2,147,483,647, i.e. from −(2^31) to 2^31 − 1
  4. Negative numbers must be turned to binary with Two's-complement. (With One's complement the output is the same as for the unsigned version of the number every time).
  5. Only the minimal number of bits required to encode the given number matters.


A truthy/falsey value if the number is uniform.


An integer n is considered to be uniform, if the number of transitions from 1 to 0 or from 0 to 1 in the binary encoding of n ist less equal to 2.


Given the number 12, its binary encoding is 1100. There is 1 transition here:

1100 |

Thus, your output should be truthy.

Given the number 10, its binary encoding is 1010. There are 3 transitions in here:

1010 |||

Thus, your output should be falsey.

Given the number -12, its binary two's complement encoding is 0100. There are 2 transition here:

0100 ||

Thus, your output should be truthy.

Here is an example for which the two's complement matters: Given the number -125, its binary two's complement encoding is 10000011. There are 2 transitions here:

10000011 | |

Thus, your output should be truthy.

Side note: If we did the same with one's complement, the encoding of -125 would be 10000010, which has 3 transitions.

Shortest code in bytes wins.

Sandbox notes:

This is my first code golf question, feel free to yell at me. Tags:

  • 1
    \$\begingroup\$ Seeing as it's outputting truthy/falsy, decision-problem would be a great tag to use. Welcome to PPCG btw. \$\endgroup\$
    – FlipTack
    Commented Jan 15, 2017 at 15:20
  • 2
    \$\begingroup\$ Personally, I think that input should be guaranteed to be positive rather than using two's complement encoding. That allows more bit-shifting based algorithms. \$\endgroup\$
    – FlipTack
    Commented Jan 15, 2017 at 20:11

Make a shuffle quine!

Closely based off this sandbox post, however I deleted the old one and posted this new answer for new feedback.

An shuffle quine is defined as a quine, of which shuffled sets of "chunks" also form quines.

For example, pretend in my magical language the code 123 is a quine.

Let's split this into 3 chunks. 1, 2, and 3. In your answers, these chunks can be any lengths and there can be any number greater than 1 of chunks.

You get a better score for more possible shuffled quines you can make. For example, if 21, 23, 13, 312, and 31 are also quines, in addition to 123, you get (15/6 = 2.5) * <sum of chunk lengths> for your score.

Your score is calculated by the formula sum_of_chunk_lengths * (number of possible shuffled programs/number of quines). In this formula, least score is better.

This is code-golf, so the shortest code wins!

Final notes:

  • Each shuffled code is a quine itself, i.e. prints itself not the original program.
  • Programs have to be distinct, i.e. you can't count abb and abb as 2 because you swapped the bs.

Sandbox notes:

Yeah, this is really hard. I don't think it's impossible though.

  • \$\begingroup\$ I appreciate that the scoring makes it impossible to pad the program with junk in order to get an unlimited score. However, I suspect it'll be won by one of the two-byte quines with a score of 2 × 2 ÷ 1, as it'd only be possible to beat with a score of 2 × 2 ÷ 2, 3 × 6 ÷ 5, or better. \$\endgroup\$
    – user62131
    Commented Jan 16, 2017 at 22:18
  • \$\begingroup\$ @ais523 hm, true. I'll mess with the scoring to see what I can come up with. \$\endgroup\$
    – Riker
    Commented Jan 16, 2017 at 22:19
  • \$\begingroup\$ It strikes me that it might be interesting to require the program to be exactly a given number of bytes long. Then the challenge would be to maximise the number of permutations that are quines. Unfortunately, a large number of bytes would be hard to score, and a small number would exclude many languages. \$\endgroup\$
    – user62131
    Commented Jan 16, 2017 at 22:34
  • 1
    \$\begingroup\$ I don't see why it's supposed to be difficult. If each chunk is a quine (that doesn't screw up the global state somehow) then you can arrange the chunks in any order and it will still be a quine and you will score 1 for the fraction part. \$\endgroup\$
    – feersum
    Commented Dec 16, 2018 at 4:28

Prisoner's dilemma

Inspired by the "Mafia" proposal

Write a bot that plays the Prisoner's dilemma. King of the hill

Payoff matrix:

Hi Mom!


  • Your bot must be a full program, not a function.
  • Your bot must run when the server does ./run in it's folder.
  • No shenanigans.
  • If you want to remember something, save it to a file in your directory.


When the server does ./run:

from_server contains your opponent's last move - Either Cooperate or Defect. On the first round it contains Let's play!

When your program exits (60 sec max):

to_server contains your choice this round. Either Cooperate or Defect. Any other output is interpreted as forfeit; you got 0 points and your opponent gets 5. If both bots forfeit on any turn, nobody gets any points.

Provided bots

There are a few bots that are guaranteed to exist:

  • Always defect
  • Always cooperate
  • 75% cooperate / 25% defect
  • 25% cooperate / 75% cooperate
  • 50% cooperate / 50% defect


Your bot will be played against every other bot in a random number of rounds > 100.


Get the most points after you've played every other bot

Fiddly bits

The game is a file structure something like so:

  -> server.py
  -> yournameherebot/
     -> from_server
     -> to_server
     -> run
     -> *Any other files you want*
  -> someotherbot/
  -> anotherbot/
  -> titfortat/
  -> 50-50_random/


Please give your CC with your comment instead of a downvote. It's a sandbox for a reason :)

  • \$\begingroup\$ What's your win criterion? \$\endgroup\$
    – Riker
    Commented Jan 6, 2017 at 0:03
  • \$\begingroup\$ Most points after playing every other bot. I thought that was implied, just a sec :) \$\endgroup\$ Commented Jan 6, 2017 at 0:04
  • \$\begingroup\$ add a grim trigger? (cooperates, but as soon as enemy defects they always defect) \$\endgroup\$ Commented Jan 6, 2017 at 0:04
  • \$\begingroup\$ Define "point"? \$\endgroup\$
    – Riker
    Commented Jan 6, 2017 at 0:04
  • \$\begingroup\$ Is this a king of the hill challenge? (With bots competing against each other) in that case, it's probably a duplicate of this one: codegolf.stackexchange.com/q/2357/31716 \$\endgroup\$
    – DJMcMayhem
    Commented Jan 6, 2017 at 0:06
  • \$\begingroup\$ Easterly: Point = generic unit of win-ness | Watermelon: Well then, you can do that. I might remove TFT from the included bots. \$\endgroup\$ Commented Jan 6, 2017 at 0:06
  • \$\begingroup\$ @DJMcMayhem This is why we post in the sandbox, people. :( \$\endgroup\$ Commented Jan 6, 2017 at 0:08
  • \$\begingroup\$ I suppose it's not really a dupe -- We're playing a random number of rounds so you can't do things like defecting on the last round against TFT. \$\endgroup\$ Commented Jan 6, 2017 at 0:10
  • 2
    \$\begingroup\$ "75% cooperate / 75% defect" \$\endgroup\$
    – Pavel
    Commented Jan 6, 2017 at 0:16
  • \$\begingroup\$ also it doesn't make sense to only know the last turn \$\endgroup\$ Commented Jan 6, 2017 at 0:26
  • 1
    \$\begingroup\$ I agree that this is a duplicate, but I'll leave that to the voters. However, for your sanity, I really recommend you change how input/output works. Each time you talk to a bot you are going to have at least 2 file writes, 2 file reads, and starting up a new process. Depending on the number of submissions and iterations, this can take a long time. I recommend keeping the processes alive and communicating through stdin/stdout. \$\endgroup\$ Commented Jan 6, 2017 at 18:39
  • \$\begingroup\$ You only are told the last turn. if you want to remember further, you can write to a file. \$\endgroup\$ Commented Jan 17, 2017 at 14:33

Divide an array into left and right halves.

Input: An array of floating-point values, or whatever is a reasonably large numeric type in your language.

Output: Two arrays. The original array should be the concatenation of the output arrays. The sums of the elements in the two arrays should be as equal as possible.


[] -> [], []
[1] -> [1], [] or [], [1]
[1, 2, 1] -> [1, 2], [1] or [1], [2, 1]
[1, 2, 3] -> [1, 2], [3]
[1, -1] -> [1, -1], [] or [], [1, -1]

Functions or full programs please; no snippets. Normal rules and restrictions apply.

  • \$\begingroup\$ This challenge is similar to this recent sandbox post, though with a slightly different (more conventional) question format. \$\endgroup\$ Commented Jan 13, 2017 at 13:23
  • \$\begingroup\$ @VisualMelon You'll just have to take my word for it that I don't read the Sandbox myself, but I am surprised at the similarity. \$\endgroup\$
    – Neil
    Commented Jan 13, 2017 at 19:32
  • \$\begingroup\$ No problem ;) there is an awful lot to trawl through if you don't have time to check back regularly. \$\endgroup\$ Commented Jan 13, 2017 at 19:38
  • \$\begingroup\$ It's been done. \$\endgroup\$
    – xnor
    Commented Jan 19, 2017 at 2:54
  • \$\begingroup\$ @xnor That question doesn't have the concatenation requirement. (Not that I'm claiming that it's enough to make this question sufficiently distinctive.) \$\endgroup\$
    – Neil
    Commented Jan 19, 2017 at 8:37
  • \$\begingroup\$ My mistake, I somehow interpreted equality as being up to ordering. Maybe it would be more obvious if you said the array is split into a left segment and a right segment. I think that definitely makes this different. \$\endgroup\$
    – xnor
    Commented Jan 19, 2017 at 8:41

City Zoner KoTH

You are a city zoner, and your opponent want to zone dirty industrial. You need to stop him. The mayor has set the following rules.

  1. A random player is selected to zone the first location. They take turns zoning area. If one player has twice as much area zoned than the other player, that player is skipped (can happen repeatedly)
  2. We have a 20x20 square to zone (400 squares total). A player can choose how much they zone, but it must be a rectangle with an area of at least 5. A player cannot zone more than 200 squares total (across the entire game)
  3. If the game ends in a tie, the player who played last wins

Your goal is to ensure your opponent zones as few squares as possible.

  • 1
    \$\begingroup\$ huh, rule number 3 makes this interesting... \$\endgroup\$ Commented Jan 22, 2017 at 3:39
58 59
61 62

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .