Default Rules: Code Golf I/O
What input/output methods do you think should be allowed/disallowed while code golfing on this site? What guidelines for input/output methods do you think should be in place?
One method per answer, please. Vote up/down according to if you believe the given method should be allowed.
Functions may take input via a …
3y ago
Functions may use arguments or …
3y ago
You may output delimiter-separ …
3y ago
Functions may take multiple ar …
3y ago
Programs may take input from p …
3y ago
Regexes may output via the lis …
3y ago
In languages without STDIN (eg …
3y ago
You may modify the input in-pl …
3y ago
Turing machines supporting mul …
3y ago
Command-line arguments may be …
3y ago
Unless explicitly asking for e …
3y ago
Numerical I/O may be given as …
3y ago
Programs may output via exit c …
3y ago
Turing machines may use the co …
3y ago
Image output may be a pixel sh …
3y ago
Programs/functions may use the …
3y ago
Functions may return bools via …
3y ago
A program may use IOs for inpu …
1y ago
A program may output a boolean …
1y ago
19 answers
You are accessing this answer with a direct link, so it's being shown above all other answers regardless of its score. You can return to the normal view.
A program may use IOs for input and output.
A input can be read via IO, which can be a port with (multiple) IOs each IO reads a symbol (a bit or more when you have a ADC). Or it can come from frames, such as UART frames, I2C, SPI, Ethernet, .... (any bus you like).
And Output can be done the same but reversed.
0 comment threads
Functions may take input via arguments and output via return value
Functions may also use STDIN/STDOUT as they wish.
You may output delimiter-separated values instead of a list
For example, the output [1,2,3] could be represented as 1 2 3 or 1|2|3.
Functions may take multiple arguments with currying
For some languages like Haskell this is almost a necessity, as only one-argument functions exist and multi-argument functions are implemented with currying. (You might take a list or tuple of the values but this isn't the natural way to do multi-arg functions in Haskell.)
The same thing should be allowed for languages which do have proper multi-argument functions. In JavaScript, one might want to use f=a=>b=>... and call f(a)(b) instead of using f=(a,b)=>... and calling f(a, b).
1 comment thread
Programs may take input from prompts from the GUI
For Mathematica, JS, Matlab, et. al. this is the closest thing they have to STDIN.
0 comment threads
In languages without STDIN (eg ///) programs may input through insertion into the source code
Also applies to cellular automata, in which the most natural way of taking input is specifying some space for a user-created structure.
0 comment threads
Regexes may output via the list of matched strings (capture group 0)
0 comment threads
Turing machines supporting multiple halt states may output with the state they halt on
Similar to the exit code submission.
You may modify the input in-place
If a function takes an input a, it is acceptable that a contains the intended output after executing instead of other forms of output. Note that this means a must be mutable to begin with and the function must actually mutate a.
Command-line arguments may be used as input instead of stdin
Languages/systems that support reading input from command-line arguments may use those as input instead of reading from stdin.
1 comment thread
Unless explicitly asking for exactly two values, you may use any truthy/falsey values in decision problems
For instance, considering the hypothetical challenge "Determine if a number is non-divisible by 3", n=>n%3 would be a valid solution, as it outputs a truthy (non-zero) value when it is non-divisible and a falsey (zero) value when it is.
0 comment threads
Numerical I/O may be given as a character code
Input 64 may be given as @ instead. This mostly exists for languages like Brainfuck that only take input through character codes.
0 comment threads
Programs may output via exit code
Exit codes are to programs what return values are to functions, so it makes sense that exit codes would be allowed.
Turing machines may use the contents of their tape pre-execution as their input
Image output may be a pixel shader
A pixel shader inputs x,y coordinates of a pixel and prints the color of a pixel (scalar for grayscale, tuple for full color, bool for binary...). Relevant for Shadertoy/GLSL answers and graphical output challenges.
0 comment threads
Functions may return bools via the presence/absence of an error
Crashing to mean false and not crashing to return true. Another branch of the exit code answer.
0 comment threads
Programs/functions may use the presence/absence of output as a bool
0 comment threads
A program may output a boolean value by using different amount of time or instructions till it returns.
If a program needs to determine a boolean value, it may return in less than N second (or instructions) when the value is false but take longer than N + ϵ second (or instructions) when the value is true or vice versa.
The difference (ϵ) has to be large enough to be detectable without much effort. I would suggest there is a factor of at least 16 between true and false (i.e. $ϵ > 15 \cdot N$) . A infinite ϵ should be allowed (if it doesn't return after N second, kill it and consider the result as true). The poster has to give an estimate of N.
0 comment threads