Code Golf: Your own pet ASCII snake

So, I wrote myself a one-liner which printed out a snake on the console. It's a bit of fun, and I wondered how I might condense my code...

Here's a (short) example output:

+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+

Here's the specs:

• In each line, a single non-whitespace character (whichever you like) is printed to the console, initially with 29 to 31 spaces padding to the left of it.
• Each iteration, a random decision is made between these three actions
• The amount of padding decreases by 1
• The amount of padding remains the same
• The amount of padding increases by 1

Do this 30 times, to print 30-segment long a snake to the console.

The shortest answer in bytes wins.

• Can we return an array of lines? Are leading/trailing spaces or newlines allowed? Feb 20, 2018 at 12:07
• I assume trailing spaces on each line are allowed, right? Feb 20, 2018 at 13:20
• What does "random" mean? Uniformly random? Random from any distribution with positive support on each of the three outcomes? Random from any distribution whatsoever? Do the results have to be different on each run of the program? Feb 21, 2018 at 9:02
• Note that, by default, random does not mean uniformly so. For example, the second part of my brainfuck answer is (as of now) perfectly valid, despite it generating a straight line most of the time.
– Jo King
Feb 21, 2018 at 15:11
• Right now, there are two points that are unclear: 1) Does random mean uniformly random? (affects this answer) and 2) Must the output be different every time? (i.e., can the random number generator be unseeded? Affects this answer) Mar 2, 2018 at 20:41

05AB1E, 15 14 bytes

30DF2Ý<+ΩD0sú,

Try it online!

Uses 0.

Explanation

30DF2Ý<+ΩD0sú,
30D            # Push 30 to the stack (One for the amount of iterations we are going to perform and one for the initial padding)
F           # Pop one of the 30s and perform the following that many times...
2Ý          # Push [0,1,2] ...
<         # and create [-1,0,1] from that
+        # Add the last padding to every entry (e.g. 30 in the beginning resulting in [29,30,31]
Ω       # Pick one of the results at random ...
D      # and push it to the stack twice
0     # Push 0 (Any character will work fine here) ...
sú   # and pad it with the randomly chosen amount of spaces in the front
,  # Finally print the result with a trailing newline
• 05AB1E: 0-bytes, Try it online!... wait, no, almost though. Feb 20, 2018 at 14:36

Random Brainfuck, 123 122 121 bytes

+[--[<]>>+<-]>+[->+>+>+<<<]++++++++++>>++<[>>[->+<<.>]>[-<+>]>?>+++<[>->+<[>]>[<+>-]<<[<]>-]>-->,<[-<<<+>>>]<<<<+.-<<.>-]

Try it online!

Random Brainfuck is an extension of brainfuck, with the helpful addition of the ? command, which sets the current cell to a random byte. This prints a snake made of !s, which looks more like footsteps than a snake funnily enough.

How It Works:

+[--[<]>>+<-]>+ Create the value 30
[->+>+>+<<<]    Copy it three times
++++++++++      Create a newline cell
>>++<            Adds 2 to the second copy to make it a space and move to the counter
[ While counter
>>[->+<<.>]>[-<+>] Print out the padding cell number of spaces
?>+++<[>->+<[>]>[<+>-]<<[<]>-] Get 3-(random byte%3)
>-->,<[-<<<+>>>]   Add (result-2) to the padding cell
<<<<+.-<           Print an exclamation mark
<<.                Print a newline
>-                 Decrement counter
] end loop

Another solution that sticks to the letter of the question, rather than the spirit.

87 bytes

+[--[<]>>+<-]>+[->+>+>+<<<]++++++++++>++>[>[->+<<<.>>]>[-<+>]?[,<+>]?[,<->]<<<+.-<.>>-]

Try it online!

This one is heavily biased towards leaving the padding alone, but increasing or decreasing the padding are both equally possible. Each one has a slightly less than 1 in 256 chance to happen.

• Very nice! I was unaware of the ? command. +1 Feb 23, 2018 at 16:39
• @GrantMiller ? is only available in Random Brainfuck, not classic brainfuck
– Jo King
Feb 23, 2018 at 23:01

C (gcc), 6158 56 bytes

Answer edited to reflect rules changes...

i;f(s){for(s=i=31;--i;printf("%*d\n",s+=1-rand()%3,8));}

Try it online!

• save a byte by moving s+=1-rand()%3 to the printf function. i;f(s){for(s=i=31;--i;)printf("%*c\n",s+=1-rand()%3,43);} Feb 21, 2018 at 2:37
• @Vaelus That makes the first line have a varying amount of spaces instead of the 30 specified in the question. Feb 21, 2018 at 7:52
• @Steadybox Which is allowed in the comments apparently.. Feb 21, 2018 at 8:03

Retina, 24 bytes

30* +
-29{¶<^

S,2@1

Try it online!

Explanation

30* +

Initialise the working string to the first line, i.e. 30 spaces and a +.

-29{¶<^

There's a space on the second line. -29{ wraps the remainder of the program in a loop, which is run 29 times. ¶< prints the working string at the beginning of each loop iteration with a trailing linefeed. The atomic stage itself inserts a space at the beginning of the string (the basic idea is to insert one space, and then randomly remove 0–2 spaces, because that's a byte shorter than randomly choosing between deletion, inserting and no-op).

S,2@1

This matches the empty regex against the input, which gives us every position between characters (and the start and end of the string). Then ,2 keeps only the first three matches, i.e. the matches after zero, one and two spaces. @ selects a random one of those three matches. Then the split stage (S) splits the input around that match. And the 1 tells it to keep only the second part of the split. In other words, we discard everything up to our random match.

The 30th line, which is the result of the final loop iteration, is printed implicitly at the end of the program.

• You know what's even shorter than removing 0--2 spaces? Shuffling spaces around!. By the way, the new Retina is great :D
– Leo
Feb 22, 2018 at 5:48

JavaScript (ES8), 6362 60 bytes

Includes a trailing newline. *2-1 could be replaced with -.5 for a 1 byte saving but the chances of each line being the same length as the previous line would be greatly increased. Of course, as "random" isn't defined in the challenge, the RNG could be replaced with new Date%3-1 for a total byte count of 55.

f=(x=y=30)=>x?.padEnd(y+=Math.random()*2-1)++
+f(--x):

Saved a byte thanks to someone who deleted their comment before I could catch the name. I'd actually tried it this way with repeat and padStart but didn't think to try padEnd - don't know why!

o.innerText=(
f=(x=y=30)=>x?.padEnd(y+=Math.random()*2-1)++
+f(--x):
)()
<pre id=o>

Bonus

For the same number of bytes, here's a version that takes the number of starting spaces & iterations as input.

f=(x,y=x)=>x?.padEnd(y)++
+f(--x,y+Math.random()*2-1):

o.innerText=(
f=(x,y=x)=>x?.padEnd(y)++
+f(--x,y+Math.random()*2-1):
)(i.value=30);oninput=_=>o.innerText=f(+i.value)
<input id=i type=number><pre id=o>

• f=(x=y=30)=>x?+\n.padStart(y+=Math.random()*2-1)+f(--x): is one byte shorter. (Note: Since SO doesn't allow line breaks in comments, I had to type \n instead of actually using a line break.) Feb 20, 2018 at 18:52
• Thanks, @Stefnotch. Unfortunately, even with an initial value of y=31, there would be a possibility of the first line being too short. tio.run/##BcFLDsIgEADQvSeZkUCs7kzQE7hyqSYzKfRjKBCYGHp6fO/… Feb 20, 2018 at 19:01
• Correct me if I'm wrong, but doesn't your current solution also suffer from that issue? Feb 21, 2018 at 20:55
• The first line should have exactly 30 spaces, according to the challenge. Feb 23, 2018 at 13:34
• @Nit, the OP has confirmed that the first line may contain 29-31 spaces but has yet to edit that fact into the challenge, despite repeated requests to do so. Feb 23, 2018 at 13:45

VBA, 6059 49 Bytes

For l=1To 30:?Spc(30+i)"+":i=i+Sgn(Rnd()-.5):Next

Paste it in the Immediate window and hit enter. (Make sure explicit declaration is turned off!)

Far more likely to move than to stay in a line (i.e. actions are not equally weighted) but that was not a specified requirement (Fortunately!)

{EDIT} Saved 1 byte by removing the space between =1 and To

{EDIT2} Saved 10 bytes thanks to remoel's comment

Old Versions:

'V1
i=30:For l=1 To 30:?String(i," ")&"+":i=i+Sgn(Rnd()-.5):Next
'V2
i=30:For l=1To 30:?String(i," ")&"+":i=i+Sgn(Rnd()-.5):Next
• Welcome to PPCG! Feb 20, 2018 at 14:53
• You can save a few bytes by replacing String(i," ") to Spc(30+i) then remove i=30: . Or -1 byte by removing & . :) Feb 21, 2018 at 5:28

C# (.NET Core), 11211010610099 98 bytes

v=>{var r="";for(int t=30,i=t;i-->0;r+="+\n".PadLeft(t+=new System.Random().Next(3)-1));return r;}

-1 byte thanks to @raznagul.
-1 byte thanks to @auhmaan.

Explanation:

Try it online.

v=>{                      // Method with empty unused parameter and no return-type
var r="";               //  Result-string, starting empty
for(int t=30,           //  Temp-integer, starting at 30
i=t;i-->0;          //  Loop 30 times
r+=                   //   Append the result-String with:
"+\n"              //    The character and a new-line,
.PadLeft(     //    left-padded with t spaces,
t+=new System.Random().Next(3)-1));
//    after t first has been changed with -1, 0, or 1 randomly
return r;}              //  Return the result-string
• One strange thing: If I call new Random().Next() multiple times locally (.net Framework Version 4.6.1) I always get the same result. I need to add a Thread.Sleep(10) between the call to reliably get different results. With a sleep time of less than 10ms I sometimes still get the same result. So .net-Framework and TIO (.net-Core) have different PRNGs or at least use different seeds. If I switch your program in TIO to C#-mono I get the same behavior as I get locally in .net-Framework. Feb 20, 2018 at 13:26
• @raznagul Hmm, that output with C# (Mono C# compiler) is indeed weird.. Feb 20, 2018 at 13:30
• @raznagul new Random() uses the time as the seed so in a tight loop the time is the same and so the result is the same. Feb 20, 2018 at 15:28
• @TheLethalCoder: Yes, that is what I (more or less) expected. The strange things are that 1) .net-Core (at least on TIO) behaves differently. 2) That I need Thread.Sleep(10) to reliably get different results and Thread.Sleep(1) or even 9ms is not enough. Feb 20, 2018 at 15:33
• @raznagul Well sleeping for 1, 9 or 10ms should all sleep for the same time depending on other processes running... odd indeed. Feb 20, 2018 at 15:37

C, 56 bytes

n;f(p){n>29?n=0:f(printf("%*d\n",n++?p-rand()%3:31,0));}

Try it online!

Explanation:

n; // As a global variable, n is initialized to zero.
f(p)
{
// Call the function recursively until n > 29.
n > 29

// At the end, set n back to zero.
? n=0

// On the first iteration, n == 0 and p has an indeterminate value.
// 'n++ ? p-rand()%3 : 31' returns 31 (without reading p), and thus
// 30 spaces get printed. printf() returns the number of characters
// printed, 32 (30 spaces + '0' + '\n').
//    On subsequent iterations, p has the value the previous printf
// call returned, which is the padding on last iteration + 2. Also,
// n > 0, so the same expression now returns p-rand()%3, and thus
// the padding changes either by -1, 0, or 1 spaces. The function
// is again called with the value of the current line's padding + 2.
: f(printf("%*d\n", n++ ? p-rand()%3 : 31, 0));
}

C (gcc), 55 bytes

n;f(p){n=n<30&&f(printf("%*d\n",n++?p-rand()%3:31,0));}

Depends on f "returning" the value assigned to n in the function, which is undefined behaviour, but works consistently with gcc when no optimizations are enabled.

Try it online!

Java 8, 89 87 bytes

First golf, I'm sure it could be much better..

Edit: Fixed first line thanks to Steadybox.

l->{for(int a=31,b=a;--a>0;){System.out.printf("%"+b+"c\n",'+');b+=2-Math.random()*3;}}

Try it online!

l->{                                           //Begin lambda
for(int a=31,b=a;--a>0;)                  //Initialise vars, loop through 30 lines
{
System.out.printf("%"+b+"c\n",'+');     //Print result
b+=2-Math.random()*3;                   //Change padding by -1, 0, or 1
}
• Welcome to PPCG! :) Feb 20, 2018 at 22:27
• 83 bytes Feb 20, 2018 at 22:43

Python 2 , 8365 64 bytes

Straightforward approach:

import os
k=30
exec"print' '*k+'+';k+=ord(os.urandom(1))%3-1;"*k

Try it online!

Thanks to @Rod for saving some bytes! Thanks to @ovs for -1 byte!

Edit: changed variable name and output string to the letter 's'

More snake-like output for 88 bytes:

from random import*
s=[30,0]
exec"print' '*sum(s)+'(S)'[s[-1]+1];s+=[randint(-1,1)];"*30
• I like this. Really neat solution. Feb 20, 2018 at 14:01

APL (Dyalog), 20 bytes

1 byte saved thanks to ngn

↑'+'↑⍨¨-+\30,2-?29⍴3

Try it online!

• ¯2+ ---> 2-
– ngn
Feb 20, 2018 at 12:23

Charcoal, 14 11 bytes

× ³⁰Ｆⅈ✳～‽³+

Try it online! Link is to verbose version of code. Explanation:

× ³⁰        Print 30 spaces (forces the desired indentation)
Ｆⅈ      Repeat the same number of times
✳～‽³  Move a random downward direction after printing
+ Print a +

Would be only 10 8 bytes if there was no initial indentation requirement.

• @KevinCruijssen Hmm, that's awkward, because Charcoal likes to trim by default.
– Neil
Feb 20, 2018 at 12:26

PHP, 61 bytes

for($p=32;$i<30;$i++)echo str_pad("+ ",$p+=rand(-1,1),' ',0);

Try it online!

• $i<30;$i++ can be $i++<30; to save 2 bytes. Feb 20, 2018 at 13:44 • 55 bytes: for($p=30;$i++<30;$p+=rand(-1,1))printf("%{$p}s\n",'+'); (the \n is counted as 1 char, and should be replaced by a real newline) Feb 23, 2018 at 15:37 • Alternative 55-byte long, without warnings: for($i=$p=30;$i--;$p+=rand(-1,1))printf("%{$p}s\n",'+'); Feb 23, 2018 at 15:41

Perl, 36 bytes

perl -E '$#a=29;map{$#a+=rand(3)-say"@a -"}@a'
• So good. I always forget you can set an array length like that... and using say for subtraction. Am I right in thinking it doesn't change the number of runs when $#a is incremented because it's not a reference? Feb 20, 2018 at 16:38 • @DomHastings: It's because I use map which seems to first lay out the elements on the stack. for does not and would have had an unpredictable loop length Feb 20, 2018 at 18:11 • That's really useful to know. Thanks! Feb 20, 2018 at 19:54 • Nice golfing. I tried to shave off another byte with perl -E 'map{$#a+=rand(3)-say"@a -"}@a=1..30', but that sometimes (not every time) resulted in segmentation fault. Could it be a bug in perl v5.22.1 and v5.16.3? Feb 21, 2018 at 16:49
• @Kjetil Yes, that is a well known problem that probably will never be fixed. When looping an array the entries don't get an extra refcount, so if you delete them they are freed by the time the loop reaches them and you hit freed memory. Fortunately you never do this in real programs. Feb 21, 2018 at 18:20

Java 8, 131129127126119108 101 bytes

v->{String r="";for(int i=30,j,t=i;i-->0;r+="+\n")for(j=t+=Math.random()*3-1;j-->0;r+=" ");return r;}

Explanation:

Try it online.

v->{                     // Method with empty unused parameter and String return-type
String r="";           //  Result-String, starting empty
for(int i=30,j,t=i;    //  Two index integers, and a temp integer (starting at 30)
i-->0;             //  Loop 30 times:
r+="+\n")          //    After every iteration: Append the character and a new-line
for(j=t+=Math.random()*3-1;
//   Change t with -1, 0, or 1 randomly
j-->0;r+=" ");   //    And append that many spaces to the result-String
return r;}             //  Return the result-String

Old 119 byte answer:

v->{String s="",r=s;int i=90,t=30;for(;i-->t;s+=" ");for(;i-->0;t+=Math.random()*3-1)r+=s.substring(t)+"+\n";return r;}

Explanation:

Try it online.

v->{                      // Method with empty unused parameter and String return-type
String s="",            //  Temp-String, starting empty
r=s;             //  Result-String, starting empty
int i=90,t=30;          //  Temp integer, starting at 30
for(;i-->t;s+=" ");     //  Fill the temp String with 60 spaces
for(;i-->0;             //  Loop 30 times:
t+=Math.random()*3-1//    After every iteration: Change t with -1, 0, or 1 randomly
r+=s.substring(t)     //   Append the result with 60-t amount of spaces
+"+\n";            //   + the character and a new-line
return r;}              //  Return the result-String

Vyxal, 12 bytes

30:(kɽ℅+:×꘍,

Try it Online!

30:          # Push 30 twice
(         # Iterate over one copy (30 times)
+     # Increment the number on the stack (initially 30) by...
℅      # A random choice from
kɽ       # [-1, 0, 1]
:    # Duplicate
×꘍  # Prepend that many spaces to an asterisk
, # And output, leaving the number on the stack

An alternative

Japt, 13 12 bytes

Outputs an array of lines.

The first line can be removed, saving 3 bytes, to allow the number of starting spaces & iterations to be taken as input.

30
°ÆQùU±Ó3ö
30\n°ÆQùU±Ó3ö
30\n              :Assign 30 to variable U
°             :Postfix increment U
Æ            :Map the range [0,30)
Q           :  Quotation mark
ù          :  Left pad with space to length
U±        :    Increment U by
Ó       :      Bitwise NOT of the negation of (i.e., decrement)
3ö     :      Random integer in the range [0,3)
• Wow, can't believe we're winning at the moment! I think you can save a byte with this (on mobile so I can't really verify) Feb 20, 2018 at 13:52
• @ETHproductions: Ha! That's identical to what I started out with. The problem is, though, if -1 is returned by the RNG on the first iteration, we'll end up with a total line length of 29 when it should be 30, 31 or 32. Feb 20, 2018 at 13:55
• Hmm, I'm confused about how your version gets around that then... Also, I believe OP has clarified in the comments that they don't care if an extra space is removed or added in the initial iteration. Feb 20, 2018 at 13:59
• Actually I'm confused by "when it should be 30, 31, or 32"—where is that required? Feb 20, 2018 at 14:01
• @ETHproductions: We start with 30 and then add -1, 0 or 1 giving us 29, 30 or 31 - add the " and that gives us a total length of 30, 31 or 32 for the first line. Feb 20, 2018 at 14:04

R, 7269 67 bytes

cat(sprintf(paste0("% ",cumsum(c(30,sample(3,29,T)-2)),"s"),"+\n"))

Thanks to Zahiro Mor for 2 extra bytes!

Try it online!

• Switching from sample(3,29,T)-2 to runif(29,-1,1) will reduce byte count by 2 but moves are not equally likely anymore. And could you also switch to paste("%" instead of paste0("% " or am I missing something here?
– Rift
Feb 20, 2018 at 16:45
• @Rift if i was using paste the resulting strings would be of the form % 30 s instead of % 30s. As you said runif would mess up the probabiities. Feb 20, 2018 at 17:46
• Locally sprintf("%30s"), sprintf("% 30s") and sprintf("% 30 s") return the same results for me. But on TIO only the first two have identical results, so paste0("%" should save a byte. And there is no requirement that every move has the same probability.
– Rift
Feb 21, 2018 at 12:41

Swift, 101 bytes

import UIKit
var g=29;for _ in 0...g{print((0..<g).map{_ in" "}.joined(),0);g+=Int(arc4random()%3)-1}

Explanation

A full program. This uses a rather odd trick: arc4random() is a member of the Darwin module, but UIKit also comes with this function installed, so it saves a byte :) Also uses one of my Swift golfing tips for repeating strings an arbitrary number of times.

import UIKit        // Imports the UIKit module, necessary for the RNG.
var g=29;           // Declares an integer variable g by assigning it to 30.
for _ in 0 ... g {  // Execute the code block 30 times (for each integer in [0; g]):
print(             // Output the following:
(0..<g).map       // For each integer in [0; g)...
{_ in" "}        // ... return a literal space character.
.joined()       // ... And join the result to a single string.
,0     // Also print a "0" preceded by a single space (g starts from 29).
);
g+=                 // Increment the variable g by...
arc4random()%3   // ... A random integer, modulo 3...
Int(...)-1       // ... Casted to an integer (yes, this is needed!) and decremented.
}
• Doesn't for _ in 0 ... g execute the code block 29 times now instead of 30 (loop from 0 to 29 (exclusive))? Feb 20, 2018 at 16:21
• @KevinCruijssen No, 0...g generates all the integers in [0; g]. My bad, fixed the explanation. 0..<g would generate the integers in [0; g) :P Feb 20, 2018 at 16:22
• Ah, the [0; g) you've edited to [0; g] indeed confused me. :) Hmm, but isn't it possible to start at g=30 and loop [1; g] in that case? Feb 20, 2018 at 17:55
• @KevinCruijssen Looping over either [0; g) or [1; g] would definitely be possible if I choose g=30 instead, but then print(...,0) needs to be changed to print(...+"0"), because an additional (extraneous) space would be prepended before the 0 otherwise. Either way, the byte count remains the same. Feb 20, 2018 at 17:58

Jelly, 18 bytes

1ŒRX+
30Ç’Ð¡⁶ẋ;€0Y

Try it online!

The chosen character is 0. If returning a list of list of characters is allowed, then the Y can be dropped and the submission can be turned into a niladic chain for 17 bytes. Alternative.

How it works

30Ç’Ð¡⁶ẋ;€0Y | Niladic main link.
30           | Starting from 30...
Ç’Ð¡       | ... Repeat the helper link 29 times and collect the results in a list.
| (This list includes the first 30, so there are actually 30 numbers).
⁶ẋ     | Repeat a space that many times, for each item in the list.
;€0  | Append a 0 to each.
Y | And join by newlines.
-------------+
1ŒRX+        | Monadic helper link. Alternatively, you can use µ1ŒRX+µ instead of the Ç.
1            | The literal one.
ŒR          | Symmetric range from –1 to 1.
X+        | Choose a random number therein and add it to the argument.

Jelly, 16 bytes

Combining mine, Erik’s and Jonathan’s solutions, we can golf this down to 16 bytes. The chosen character is 1.

’r‘X
30Ç’Ð¡Ṭ€o⁶Y

Try it online!

Thanks to Jonathan Allan for the heads-up (on Ṭ€o⁶).

• You could use Ṭ€o⁶ in place of ⁶ẋ;€0 like my 18 byter does and get to down to 17. Feb 21, 2018 at 1:00
• @JonathanAllan Thank you! Combining the three Jelly answers, I actually ended up having 16 bytes. But I posted that as a secondary solution because it is not entirely mine. :-) Feb 21, 2018 at 5:16

R, 54 53 bytes

cat(sprintf('
%*s',cumsum(c(30,sample(3,29,T)-2)),0))

Similar idea as above, but with shortened sprintf code and a shorter character string literal. Instead of \n (two bytes) I’m using a literal line break (one byte).

Try it online!

• A field width or precision (but not both) may be indicated by an asterisk *: in this case an argument specifies the desired number. I've been using sprintf for years and somehow always missed that part... Thanks for the reminder! Feb 20, 2018 at 18:51
• The second line is sometimes indented two spaces instead of one. Feb 21, 2018 at 5:18
• @ScottMilner Found some time to fix it. Feb 21, 2018 at 14:54

Ruby, 45 39 bytes

x=30
x.times{puts' '*(x+=rand(3)-1)+?S}

Try it online!

Modifying x during the loop does not affect the loop counter. I chose S as a particularly snakelike output character.

-6 bytes: Use rand(3)-1 instead of [-1,0,1].sample. Thanks, Eric Duminil!

• You can save two bytes with x.map instead of x.times (equivalent since you don't use the return value) Feb 21, 2018 at 2:11
• Oops, you're right, ignore me! Feb 21, 2018 at 3:34
• OK I've got a better one: rand -1..1 is five bytes shorter than [-1,0,1].sample Feb 21, 2018 at 4:36
• @RJHunter: Or rand(3)-1 for 6 bytes less. Feb 21, 2018 at 7:52
• (x=30).times{puts' '*x+?+;x+=rand(3)-1} (same size) will print exactly 30 spaces for the head of the snake as requested by the challenge Feb 28, 2018 at 9:54

SenseTalk, 237 198 Bytes

This is a language that I came to know and love about a decade ago. It's the scripting language that drives the automated testing tool Eggplant Functional. I was an avid user of the tool for many years before joining the company for a while. It's not the most golf-capable language, but I find it very enjoyable to write in. Golfing in it is actually quite challenging as the language is meant to be verbose and English-like... took me quite a while to get it down to 237 bytes.

set s to "                              +"&lf
set p to s
repeat 30
set a to random(0,2)
if a equals 0
delete first char of p
else if a equals 1
put " " before p
end if
put p after s
end repeat
put s

Ungolfed/Explanation

set the_snake to "                              +"&lf #assign the first line of the snake
set previous_line to the_snake                        #set up for the loop

repeat 30 times                                       #loop 30x
set action to random(0,2)                         #random add/subtract/stay the same

if action equals 0
delete the first character of previous_line   #SenseTalk really shines at string manipulation
else if action equals 1
put " " before previous_line                  #insert a character at the beginning
end if

put previous_line after the_snake                 #plop the new segment into the string
end repeat                                            #close the loop

put the_snake                                         #print to standard out

Edit: Saved 36 bytes thanks to @mustachemoses

• Is the whitespace necessary? Feb 27, 2018 at 5:52
• I've a lot of respect for this. Particularly against the background of 'golfing languages' which seem to exist to be utterly unreadable, but short. It's nice to have an example that's just showing what you can do with your language. Feb 27, 2018 at 9:11
• Good call @MustacheMoses! Updated. Feb 28, 2018 at 4:17
• I count 198 bytes (well, the userscript counts that many) Feb 28, 2018 at 5:16
• @AllenFisher Do you have a standalone interpreter or compiler for this language that I can use without getting a Eggplant demo? Feb 28, 2018 at 18:15

Try it online!

Loops from 1 to $l=30. Each iteration we put$l spaces plus an x onto the pipeline as a string, then += either of -1, 0, 1 based on Get-Random into $l for the next loop. Those strings are gathered from the pipeline and an implicit Write-Output gives us a newline-separated list for free. Bash, 53 • 3 bytes saved thanks to @Dennis for((i=p=30;i--;p+=RANDOM%3-1));{ printf %${p}s+\\n;}
• p+=RANDOM%3-1 works too. Feb 20, 2018 at 18:47
• @Dennis Yes of course! Thanks! Feb 20, 2018 at 19:35

Octave, 53 51 50 49 bytes

printf('%*d\n',[a=31+cumsum(randi(3,1,30)-2);~a])

Try it online!

Saved 1 byte by no longer doing any looping. Saved another as Octave has printf as well as fprintf.

This new code creates an array of 30 random integers in the range -1:1. It then cumulatively sums the array and adds 30, which gives the desired sequence.

The result is printed using fprintf with a format that says "A decimal number, padded to a specified width, followed by a new line. The width will be the first value input, and the decimal number will be the second value input. If the number of values input is more than this, Octave will keep repeating the print automatically to get the desired output.

To achieve the looping then, we need only interleave zeros between the sequence array so the fprintf function uses each value in the sequence as a width, and each zero as the digit to be printed.

Prints an output like:

0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

The above code doesn't always print exactly 30 spaces on the first line. It will be either 29, 30, or 31. To correct that, you would use this 53 byte version:

x=31;for i=2:x;fprintf('%*d\n',x,0);x+=randi(3)-2;end
• You can save two bytes: x=31;for i=2:x;fprintf('%*d\n',x+=randi(3)-2,0);end Feb 21, 2018 at 9:11
• @KevinCruijssen I thought about that, but it doesn't work the same. It would result in the first line starting with either 29, 30, or 31 spaces. Feb 21, 2018 at 9:13
• @KevinCruijssen in fact, never mind. I've just noticed the OPs comments saying that is allowed. Feb 21, 2018 at 9:14
• Yep. It's a bit annoying a rule is in the comment (especially since it contradicts with the current challenge description..). I've asked OP to edit the challenge to reflect you can start with 29, 30 or 31, since he seems to allow it in the comments. Feb 21, 2018 at 9:19

Lua, 81 75 bytes

n=30;for i=1,n do print(("%-"..n.."s+"):format(" "))n=n-2+math.random(3)end

In for i=1,n ... the to_exp n is evaluated only once before entering the loop, saving one byte.

-6 thanks to @user202729

Try it online!

• Welcome to PPCG! You may want to add a TIO link to your post so people can test your program more easily. Feb 26, 2018 at 15:53
• You can golf your answer down to 76 bytes. Besides the page Tips for golfing in Lua may be useful. Feb 26, 2018 at 15:54
• Feb 26, 2018 at 15:58

Python 3.6, 8473 69 bytes

from random import*
x=30
exec("print(' '*x+'+');x+=randint(-1,1);"*x)

Thanks to @WheatWizard for -11 bytes. Thanks to @JoKing for -4 bytes.

• Since you don't end up using i you can use for i in[1]*30 instead to save bytes. Feb 23, 2018 at 6:31
• You can also from random import* so that you don't need the random. later on. And you can remove the newline space after your :. Feb 23, 2018 at 6:32
• And if you change 30 to 29 the "+".rjust(x) can be replaced with " "*x+"+". Feb 23, 2018 at 6:35
• Actually regarding my last comment it should be 30 spaces not 29. Your current answer only puts 29 spaces and thus fails to meet the spec. This can be fixed by changing 30 to 31. Feb 23, 2018 at 6:59
• @WheatWizard Thanks! I've added your changes and credited you as well. I changed [1]*30 to [1]*x because it's one byte shorter. Feb 23, 2018 at 19:32

ES5, 9795 81 bytes

for(p=i=30;i--;)console.log(Array(p).join(" ",r=Math.random()*3|0,p+=r>1?-1:r)+0)

ES5, 112 98 bytes if function format is needed:

function a(){for(p=i=30;i--;)console.log(Array(p).join(" ",r=Math.random()*3|0,p+=r>1?-1:r)+0)}a()

• Welcome to PPCG! I think your variables have to be declared inside of a function - _=>{p=30;for(i=0;i<p;i++){console.log(Array(p).join(" ")+"+\n");r=~~(Math.random()*3);p+=r==2?-1:r}} Feb 23, 2018 at 16:37
• That'll kill the bytes, I'll tell you that. ES5 doesn't include the arrow functions by default. I'll update Feb 23, 2018 at 16:46
• @Oliver, function format added :) Feb 23, 2018 at 16:52
• Nice! I'm not sure when it was introduced, but I think you can replace join(" ") with join  Feb 23, 2018 at 16:54
• @Oliver I'm not sure you could do that in ES5, could you? I thought that was a template literal added in ES6? Feb 23, 2018 at 16:56