Fizzbuzz!

[yy2bggggs]

Code: Select all

@echo off
for /l %%x in (1,1,100)do call :a %%x
goto :eof
:a
set /a f=%1%%3
set /a b=%1%%5
set r=
if %f%==0 set r=Fizz
if %b%==0 set r=%r%Buzz
if %r%.==. (echo %1) else (echo %r%)

179 char batch file (using single char line feeds, omit the final one)

[Drostie]

EvanED, here's the trick:

In Java, nearly everything is an Object -- which includes integer primitives at some times. While you might not think of the primitives as objects, a recent feature called autoboxing allows primitives (byte, char, int, etc.) to be cast to their corresponding objects (Byte, Character, Integer, etc.)

As such, the following code compiles validly in Java:

Code: Select all

class B {
    public static void print(Object obj){
        System.out.println(obj);
    }
    public static void main(String[] args){
        Object o = 1;
        print(1);
        print( (Object) 1); // redundant.
        print(true?o:2);
    }
}

Now, autoboxing is not universal -- that is, a primitive is not always considered equivalent to its object. An example is the code (1).reverse(); if you try it, you'll get an error that the int cannot be dereferenced.

But autoboxing does work on things like class-casts and function calls.

So, to answer your question, it's not an overloading issue; the type of (A ? i : "stuff") is simply Object, since that's the lowest class which has both of those as subclasses; and it works because 1 can be autoboxed to new Integer(1).

So, the code works because PrintStream.println(Object o) is defined. As mentioned in the API, that particular function just calls String.valueOf(o) and prints it out the same way it prints out anything else.

If you change the above code to say print(String s) instead of print(Object o), you'll see some pretty angry compiler messages.

[EvanED]

Ahhhhhh, gotcha. Makes sense. Thanks.

And that explains why it started working in 1.5, 'cause I knew that they added autoboxing then.

[warriorness]

If you change the above code to say print(String s) instead of print(Object o), you'll see some pretty angry compiler messages.

I tried it, and it's not that bad. But it's absolutely hideous if you conflict generics.

Code: Select all

import java.util.*;

public class Test
{
   public static void main(String[] args)
   {
      boolean x = false;
      
      LinkedList<String> y = new LinkedList<String>();
      LinkedList<Integer> z = new LinkedList<Integer>();
      
      y.add("Hello!");
      z.add(1337);
      
      print( x?y:z );
   }
   private static void print(LinkedList<String> o)
   {
      System.out.println(o.toString());
   }
}

Try compiling it, and you get:

Code: Select all

Test.java:15: print(java.util.LinkedList<java.lang.String>) in Test cannot be applied to (java.util.LinkedList<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<?>>>&java.util.AbstractSequentialList<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<?>>>&java.util.AbstractList<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<?>>>&java.util.List<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<?>>>&java.util.AbstractCollection<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<?>>>&java.util.Queue<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<?>>>&java.util.Collection<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<?>>>&java.lang.Iterable<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<? extends java.lang.Object&java.io.Serializable&java.lang.Comparable<?>>>)
                print( x?y:z );
                ^
1 error

However, simply remove the generic type in the parameter of print(...) so it is now "print(LinkedList o)", it compiles fine! (and the output is "[1337]").

[Drostie]

Yeah, that's nasty, but none of that is caused by the actual autoboxing. (You get the same error if you replace 1337 with new Integer(1337), where you do the boxing yourself.)

Two things are happening in your code, both of them related to this "cast-up" effect of the ternary operator (the question-mark (?) operator thing).

I earlier said that if you try to do A? B : C, then it looks for the lowest mutual superclass it can do. That's only a part of the truth -- it then *extends* this superclass, with a derived one that implements any mutual interfaces that both subclasses have, but the superclass does not.

When you try to use a ternary operator on an Integer and a String, instead of just getting an Object, you get a:

Code: Select all

? extends java.lang.Object
    &java.io.Serializable
    &java.lang.Comparable<
        ? extends java.lang.Object
            &java.io.Serializable
            &java.lang.Comparable<?>>

Now, one of the things that's killing your code is just the redundancy of the generic in the Comparable class. The object that the ternary operator creates is, as far as I can tell, only Comparable to itself -- you cannot, as far as I know, create *any* another object of the same anonymous class to compare it with. And so the above specification states that outright. It just means many more characters to deal with.

Okay, so let's replace that whole long "? extends Object" thing in the code block, above, with the letter A. If we call this anonymous class A, what's happening in your particular compiler error?

Well, first off, the LinkedList doesn't get cast to a different class. It doesn't need to be; both objects are LinkedLists. So their superclass is a LinkedList, too.

But aha, these aren't *quite* the same class. One is a subclass called LinkedList<Integer> and the other is a subclass called LinkedList<String>. And, though you don't need to send the LinkedList to a superclass, you do need to send the generics there. So we're casting to a LinkedList<A>.

And that's not quite enough, either. LinkedList<String> also implemented some of its interfaces with generics -- for example, it implements Iterable<String>. Any genericized interfaces *also* need to be respecified.

If I just rewrite your error message with my A convention, and with some whitespace tossed in, it says:

Code: Select all

print(java.util.LinkedList<java.lang.String>) in Test cannot be applied to (
java.util.LinkedList<A>
    &java.util.AbstractSequentialList<A>
    &java.util.AbstractList<A>
    &java.util.List<A>
    &java.util.AbstractCollection<A>
    &java.util.Queue<A>
    &java.util.Collection<A>
    &java.lang.Iterable<A>)

So, all of the genericized LinkedList interfaces also got respecified.

And that's how your compiler message from hell was created -- generics + ternary operators + two classes which were both Comparable and Serializable, but could not be traced to a superclass.

[rulerofutumno]

After a lot of struggling, I managed another solution of 106 bytes. Now it's less readable than Perl.

Code: Select all

b s d n|mod n d==0=s|True="";m""x=show x;m y _=y;main=putStr$do n<-[1..100];m(b"Jizz"3n++b"Buzz"5n)n++"\n"

You could always replace "True" with "1>0" or something along those lines and save one byte.

Haskell (105 bytes):

Code: Select all

main=putStr$unlines$map h [1..100];h a|mod a 15<1="FizzBuzz"|mod a 3<1="Fizz"|mod a 5<1="Buzz"|3>2=show a

C# (124 bytes):

Code: Select all

class P{static void Main(){for(int i=0;i++<100;)System.Console.WriteLine(i%5<1?i%3<1?"FizzBuzz":"Buzz":i%3<1?"Fizz":""+i);}}

</first post>

[william]

Python's method-based REs are readable and nice, but the lack of a global variable ($_, in Perl) makes them harder to use. If you want to print a text file on a line-by-line basis using regexes, Perl is easier.

But that might just be because the Python regex tutorial I found was less comprehensible than the Perl one.

Perl and Python use the same "guts" in their regex libraries, IIRC. It's just that Perl has built-in syntax for it, which helps if you use regeces but just adds to the bloat of the language if you don't. (This wasn't always the case though. Once python had Emacs-style regeces which aren't as good as Perl-style regeces)

[Torn Apart By Dingos]

After a lot of struggling, I managed another solution of 106 bytes. Now it's less readable than Perl.

Code: Select all

b s d n|mod n d==0=s|True="";m""x=show x;m y _=y;main=putStr$do n<-[1..100];m(b"Jizz"3n++b"Buzz"5n)n++"\n"

You could always replace "True" with "1>0" or something along those lines and save one byte.

Haskell (105 bytes):

Code: Select all

main=putStr$unlines$map h [1..100];h a|mod a 15<1="FizzBuzz"|mod a 3<1="Fizz"|mod a 5<1="Buzz"|3>2=show a

Right. And using the trick that ==0 is the same as <1, I can trim my code to 104 bytes. I could improve your code:
Haskell (98 bytes):

Code: Select all

main=putStr$unlines$map h[1..100];h a|a%15<1=h 3++h 5|a%3<1="Fizz"|a%5<1="Buzz"|3>2=show a;(%)=mod

I'd like to see the 85 byte solution for Haskell.

EDIT:
An improvement! Down to 95 bytes.

Code: Select all

main=mapM_(putStrLn.h)[1..100];h a|a%15<1=h 3++h 5|a%3<1="Fizz"|a%5<1="Buzz"|3>2=show a;(%)=mod

[musicinmybrain]

At least for Python, it will let you shave off a byte by submitting a file without a newline at the end. (Your text editor may add the newline automatically, as text files are really supposed to end with newlines.) I don't know if it will throw errors when you do that for other languages.

Here's my best effort so far in Python (76 bytes):

Code: Select all

for x in range(1,101):print str(x)*(x%3*x%5>0)+"Fizz"*(x%3<1)+"Buzz"*(x%5<1)

I tried weaseling the size down by using urllib to retrieve the answer from a remote location, but the server didn't like that. Can't say I'm surprised.

I'm amazed people have done it in 58 bytes.

[hotaru]

i've managed to get it down to 50 bytes in perl, a few different ways:

Code: Select all

print"Fizz"x!($_%3).Buzz x!($_%5)||$_,$/for 1..100

Code: Select all

print"Fizz"x!($_%3).Buzz x/[05]$/||$_,$/for 1..100

Code: Select all

print+Fizz x!($_%3).Buzz x!($_%5)||$_,$/for 1..100

Code: Select all

print+Fizz x!($_%3).Buzz x/[05]$/||$_,$/for 1..100

[rulerofutumno]

I could improve your code:
Haskell (98 bytes):

Code: Select all

main=putStr$unlines$map h[1..100];h a|a%15<1=h 3++h 5|a%3<1="Fizz"|a%5<1="Buzz"|3>2=show a;(%)=mod

I'd like to see the 85 byte solution for Haskell.

EDIT:
An improvement! Down to 95 bytes.

Code: Select all

main=mapM_(putStrLn.h)[1..100];h a|a%15<1=h 3++h 5|a%3<1="Fizz"|a%5<1="Buzz"|3>2=show a;(%)=mod

Not bad, not bad... some things there that I don't know about yet - I only started Haskelling a few days ago.

Seeing that 85-byte program would be pretty awesome, but more than that I want to see the winning C# entry - it's one byte smaller than mine and I have no idea where that byte could come from.

[Torn Apart By Dingos]

Better still in Haskell (91 bytes):

Code: Select all

main=mapM_(\a->putStrLn$[show a,"Buzz","Jizz","JizzBuzz"]!!(0^mod a 5+2*0^mod a 3))[1..100]

Here's another 95 byte solution:

Code: Select all

main=putStr$[1..100]>>=(++"\n").(zipWith(++)(cycle["Jizz","",""])(cycle["Buzz","","","",""])!!)

It's really frustrating that I can't reach 85 bytes.

[rulerofutumno]

It's really frustrating that I can't reach 85 bytes.

Getting closer every day though..

[zybler]

I got 81 characters in C. The previous poster who got 79 characters had some error.

Code: Select all

main(i){while(printf(i%5?i%3?"%d\n":"Fizz\n":"FizzBuzz\n"+(i%3?4:0),i),i++<100);}


[hotaru]

121 bytes in factor:

Code: Select all

: d dupd mod 0 = ; 100 [ 1+ 15 d [ drop "FizzBuzz" ] [ 3 d [ drop "Fizz" ] [ 5 d [ drop "Buzz" ] when ] if ] if . ] each

[Earlz]

there should really be a show-off type forum where you are limited to code of like 200 bytes or so and just so people can make random stuff...maybe make it a competition...

edit:
what about in assembly? because it is basically not abstract, could it be that the actual compiled code count for the size?

[torne]

what about in assembly? because it is basically not abstract, could it be that the actual compiled code count for the size?

I'd do that now if I wasn't at work x86 DOS COM executables can be written entirely using printable ASCII characters, too, for maximum amusement (thus meaning you could still trivially paste it as a forum quote).

[mrkite]

what about in assembly? because it is basically not abstract, could it be that the actual compiled code count for the size?

Great, now I feel old.

Reading that reminded me that I had once participated in a 256 byte fire competition. I did a search to see if I could find it.

It turns out the competition was a #coders competition in 1996.

[yy2bggggs]

I got 81 characters. The previous poster who got 79 characters had some error.

Code: Select all

main(i){while(printf(i%5?i%3?"%d\n":"Fizz\n":"FizzBuzz\n"+(i%3?4:0),i),i++<100);}


Which was?

[zybler]

I got 81 characters. The previous poster who got 79 characters had some error.

Code: Select all

main(i){while(printf(i%5?i%3?"%d\n":"Fizz\n":"FizzBuzz\n"+(i%3?4:0),i),i++<100);}


Which was?

Which was:

main(i){while(printf(i%5?i%3?"%d\n":"Fizz\n":"FizzBuzz\n"+i%3*4%8,i),i++<100);}
The bolded was not mathematically equivalent to (i%3?4:0). If you go and submit it to http://golf.shinh.org/p.rb?FizzBuzz. It would failed.

[yy2bggggs]

The bolded was not mathematically equivalent to (i%3?4:0).

Good catch.

If you go and submit it to

...but why would you do that?

[Victorkm]

I only know Visual basic for Excel but without the workbook attached its only 489 bytes. With the workbook its 22k

Code: Select all

Sub count()
For butt = 1 To 100
Worksheets("Sheet1").Cells(butt, 1).Value = butt
check = 0
check2 = 0
For butt2 = 1 To 33
If butt / butt2 = 5 Then
Worksheets("Sheet1").Cells(butt, 1).Value = "Buzz"
check = 1
End If
If butt / butt2 = 3 Then
    Worksheets("Sheet1").Cells(butt, 1).Value = "Fizz"
    check1 = 1
    If check1 + check = 2 Then
        Worksheets("Sheet1").Cells(butt, 1).Value = "FizzBuzz"
        Exit For
    End If
End If
Next butt2
Next butt
End Sub


[hotaru]

74 bytes in false.

Code: Select all

1[$101\>][$$[$2>][3-]#\[$4>][5-]#$@$@*[@$.@@]?~["Fizz"]?~["Buzz"]?"
"1+]#

[Defenestrator]

56 bytes in Python:

Code: Select all

i=1;exec"print'FizzBuzz'[i%-3&4:12&8-i%5]or i;i+=1;"*100

Multiplying a string by an integer will return that string repeated that many times, so "foo" * 3 == "foofoofoo". Using this together with exec is shorter than any other type of loop in this case. The loop body prints a cleverly calculated substring (or the value of i, if the substring is empty).

For the first substring index, i%-3 returns 0, -2 or -1. Bitwise and-ing with 4 returns 0, 4 or 4 respectively.
For the second substring index, i%5 returns 0, 1, 2, 3 or 4, then 8-i%5 returns 8, 7, 6, 5 or 4, and 12&8-i%5 returns 8, 4, 4, 4 or 4 respectively.

[Hench]

I've got 15th place in Java at 134 bytes for

Code: Select all

class a {
    public static void main(String[]b) {
        for(int i=0; ++i<101; System.out.println(i%3<1 ? i%5<1?"FizzBuzz":"Fizz" : i%5<1?"Buzz":""+i));
    }
}

without the un-required whitespaces and line returns, of course.

[Nexuapex]

As small as I can figure out how to get it in Lua:

Code: Select all

for i=1,100 do s=(i%3<1 and"Fizz"or"")..(i%5<1 and"Buzz"or"")print(#s<1 and i or s)end

[adlaiff6]

56 bytes in Python:

Code: Select all

i=1;exec"print'FizzBuzz'[i%-3&4:12&8-i%5]or i;i+=1;"*100

Very congratulations!

[Porges]

the type of (A ? i : "stuff") is simply Object, since that's the lowest class which has both of those as subclasses

If you want to get technical, it's the join, or least upper bound. Since Java doesn't have a particularly rich type system this is always Object :p

[EndangeredMassa]

How about Javascript?

Code: Select all

for(i=1;i<101;i++)document.writeln(i%3*i%5?i:i%5?"Fizz":!i%3?"FizzBuzz":"Buzz")

That's a mere 80 characters.

[else]

My solution. 80 characters in C.

Code: Select all

main(i){char*a="FizzBuzz";while(i++<=99)printf(i%3?i%5?"":&a[4]:i%5?"Fizz":a);}


Edit:

Actually that char array does nothing for me. 72.

Code: Select all

main(i){while(i++<=99)printf(i%3?i%5?"":"Buzz":i%5?"Fizz":"FizzBuzz");}


[zybler]

My solution. 80 characters in C.

Code: Select all

main(i){char*a="FizzBuzz";while(i++<=99)printf(i%3?i%5?"":&a[4]:i%5?"Fizz":a);}


Edit:

Actually that char array does nothing for me. 72.

Code: Select all

main(i){while(i++<=99)printf(i%3?i%5?"":"Buzz":i%5?"Fizz":"FizzBuzz");}


I don't think your solution is correct. You're supposed to print the number out as well if the number is not a factor of 3 or 5 or both. Not to mention the questions requires a newline after each number.

[Random832]

I will admit, I know of two ways where Perl is better than Python:

1. Regular expressions.

2. Making hilariously indecipherable programs, which is really a bad sign for heavy-duty programming languages but makes for great fun.

One of my personal favorites is this one:

Code: Select all

 not exp log srand xor s qq qx xor
 s x x length uc ord and print chr
 ord for qw q join use sub tied qx
 xor eval xor print qq q q xor int
 eval lc q m cos and print chr ord
 for qw y abs ne open tied hex exp
 ref y m xor scalar srand print qq
 q q xor int eval lc qq y sqrt cos
 and print chr ord for qw x printf
 each return local x y or print qq
 s s and eval q s undef or oct xor
 time xor ref print chr int ord lc
 foreach qw y hex alarm chdir kill
 exec return y s gt sin sort split

No literals. At all.

I'm pretty sure the use of the 'q' and 'qq' operators count as literals. What i'm not sure is whether the space can act as a delimiter, or if the next letter after the space is what does it.

[aldimond]

I don't know how some dude got down to 43 in bash... I'm hovering around 101 with

Code: Select all

a()(o=
((i%3))||o+=Fizz
((i%5))||o+=Buzz
(((i%3)*(i%5)))&&o+=$i
echo $o
((i++))
((i-101))&&a)
i=1
a


I'm definitely not an expert on crazy bash syntax... so if there are cheaper ways to do math and conditionals I may be missing out.

EDIT: Down to 94 by consolidating i++ with i-101. Obviously there is something fundamentally better to be done. Also for some reason early on I thought that the recursive call was smaller than a normal "for i in {1..100}" loop, but it winds up being exactly the same.

[mrkite]

I currently hold the record for TCL.. at 85 bytes.

Code: Select all

set i 0;while {[incr i]<101} {puts [expr $i%3?$i%5?$i:"Buzz":$i%5?"Fizz":"FizzBuzz"]}


[aldimond]

Woo, have come up with a much better way in bash, currently at 51 bytes. I'll spoilerize this, 'cause it doesn't suck as much as my last attempt.

Spoiler:

Code: Select all

seq 100|sed -e15~15cFizzBuzz -e3~3cFizz -e5~5cBuzz

Yes, this is not so much bash as sed. Sue me. Maybe I'll do one fully in sed.

Edit: It's at 46 like so:

Spoiler:

Code: Select all

seq 100|sed '0~15cFizzBuzz
0~3cFizz
0~5cBuzz'

Note that sed is really strange... after you "c" a line it's no longer available to future "c" commands.

[Yakk]

I'm pretty sure the use of the 'q' and 'qq' operators count as literals. What i'm not sure is whether the space can act as a delimiter, or if the next letter after the space is what does it.

It depends on the version of perl. Whitespace being used as a delimiter was nerfed in perl5, I think.

[aldimond]

I decided to get addicted to code golf for a while, and I did the little-endian calculator one in C. And I'm currently the low man for C, with these 184 characters of noise:

Spoiler:

Code: Select all

main(i){char*f,a[99],b,*c;while(gets(a)){b=*(strchr(a,32)+4)-'d';i=strtoul(a,&f,b?10:16);for(f-=c=a;f;i>>=8,f=b?i:f-2,c+=2)sprintf(c,"%02hhx\n",i);printf(b?a:"%u\n",strtol(a,0,16));}}

For correctness, the last strtol should be a strtoul. None of the test cases on the site excercise it (they will break it if the first strtoul is a strtol), so I guess I'm cheating .

[julesh]

How about Javascript?

Code: Select all

for(i=1;i<101;i++)document.writeln(i%3*i%5?i:i%5?"Fizz":!i%3?"FizzBuzz":"Buzz")

That's a mere 80 characters.

Three incremental improvements:

Code: Select all

for(i=0;++i<101;)document.write(i%3*i%5?i:i%5?"Fizz":!i%3?"FizzBuzz":"Buzz")

Just trimming unnecessary characters from your version, down to 76, excluding newline.

Code: Select all

s="";for(i=0;++i<101;)s+=i%15?i%5?i%3?i:"Fizz":"Buzz":"FizzBuzz";alert(s)

Rewriting the conditional expression. Plus, just for a laugh, nobody specified where it has to be output to, so put it in a dialog box instead.

Code: Select all

for(i=0;++i<101;)document.write(i%15?i%5?i%3?i:"Fizz":"Buzz":"FizzBuzz")

The best I can do. 72 chars, excluding newline.

[Notch]

It's a shame the java upload won't run static initializers

Code: Select all

class a{public static void main(String[]a){}}
// Is much larger than
class a{static{System.exit(0);}}

On my regular jvm, that loads the class, runs the static initializer, then exits the jvm before looking for the main method, and I think it's even JVM specification compliant.

[Robin S]

there should really be a show-off type forum where you are limited to code of like 200 bytes or so and just so people can make random stuff...maybe make it a competition...

Late reply, but you may be interested in .kkrieger. That article also mentions

a fly-through of the first level of Descent in a mere 4,095 bytes.

That can be found here.