OMGWTF programming contest

[plams]

The Olympiad of Misguided Geeks contest at Worse Than Failure.

Basically you can win a MacBook Pro if you write a 4-function calculator. However, you have to impress the judges by writing truly awful code. If you can make somebody go 'WTF?!', and even laugh, when they look at your code you're probably on right track.

One of my ideas that I'm probably not going to use since it's too obvious is to pipe your input through Google Calculator and print the result.

[EvanED]

What? Why are you announcing this? Don't you know that the more people enter the lower our chances will be?

[plams]

But the more hillarious the winning entry will be!

Actually.. I think the worst part about this thread is that no one will reveal their genius ideas in fear of IP theft. Maybe we can share our ideas after the deadline at may 14.

[evilbeanfiend]

Code: Select all


int doAdd(int lhs, int rhs)
{
  int result=lhs;
  for(int i=0;i<rhs;++i)
  {
    ++result;
  }
  return result
}

int DoMult(int lhs, int rhs)
{
  int result=lhs
  for(int i=0;i<rhs;++i)
  {
    result = DoAdd(result,lhs);
  }
  return result
}



as a start. but im sure many layers of wtf will be required.

[hotaru]

i can do better than that...

Code: Select all

int doAdd(int lhs,int rhs){
 for(;rhs>0;rhs=~-rhs)lhs=-~lhs;
 for(;rhs<0;rhs=-~rhs)lhs=~-lhs;
 return lhs;
}

[Yakk]

Code: Select all

template<int n, int m, int base>
struct add {
  enum {
    result = (n+m)%base,
    carry = ((n+m)/base)?1:0
  };   
};

template<int base>
struct add_work {
  template<int n>
  struct arg_1 {
    template<int m>
    struct arg_2 {
      enum {
        upper = add_work<base>::arg_1<n/base>::arg_2<m/base>::result,
        carry = add<n%base, m%base, base>::carry,
        lower = add<n%base, m%base, base>::result,
        result = base*(upper+carry)+lower,
      };
    }
    template<>
    struct arg_2<0> {
      enum {result = n};
    }
  }
  template<>
  struct arg_1<0> {
    template<typename m>
    struct arg_2 {
       enum {result = m};
    };
  };
};

template<int base, int n, int m>
struct adder {
  enum{result=add_work<base>::arg_1<n>::arg_2<m>::result};
};


Write up the calculator using silly template metaprogramming.

Have your calculator take the expression, then use preprocessor macros to compile the template metaprogramming programs with -define flags.

Extract out from the resulting binary the value of the result without running it, as it will be a compile-time constant.

Return the constant.

[plams]

I'll raise you a

Code: Select all


int doAdd(int lhs, int rhs)
{
  int result=lhs;
  while(result<(lhs+rhs))
  {
    ++result;
  }
  return result;
}
int DoMult(int lhs, int rhs)
{
  if(rhs > lhs)
  {
    // exploit commutative property
    return DoMult(rhs,lhs);
  }
  else
  {
    return lhs*rhs;
  }
}


[HiEv]

If it were legal, I'd be tempted to write C++ code that outputs, compiles, and runs Pascal code to find the answer, and then returns that.

[evilbeanfiend]

exploiting the commutativity is nice but your function looks like it is going to end up being more efficient then the previous ones. ideally we want addition and multiplication to be worse than O(n) though which seems hard without putting in gratuitously useless steps (bogo sort to get the answer maybe?).

[Yakk]

exploiting the commutativity is nice but your function looks like it is going to end up being more efficient then the previous ones. ideally we want addition and multiplication to be worse than O(n) though which seems hard without putting in gratuitously useless steps (bogo sort to get the answer maybe?).

O(n lg n): Increment bitwise manually.

O(n e^n): Randomly search integers until you find the one that is 1 greater than your target number.

O(e^e^n): Randomly search integers until you find the one that is 1 greater than your target number. Restart entire calculation if you make any error anywhere.

[evilbeanfiend]

exploiting the commutativity is nice but your function looks like it is going to end up being more efficient then the previous ones. ideally we want addition and multiplication to be worse than O(n) though which seems hard without putting in gratuitously useless steps (bogo sort to get the answer maybe?).

O(n lg n): Increment bitwise manually.

O(n e^n): Randomly search integers until you find the one that is 1 greater than your target number.

O(e^e^n): Randomly search integers until you find the one that is 1 greater than your target number. Restart entire calculation if you make any error anywhere.

the second two was pretty much what i was thinking when i said bogo sort - extra trickness should be in the target matching methnks

[hotaru]

Code: Select all

#define _ rhs--
int DoAdd(int lhs,int rhs){
 if(!lhs)return rhs;
 if(rhs&(1<<31))return -DoAdd(-lhs,-rhs);
 for(;_;)lhs*=-(float)~lhs/lhs;
 return lhs;
}
int DoMult(int lhs,int rhs){
 int result=0;
 for(;_;)result=DoAdd(result,lhs);
 return result;
}

[EvanED]

Personally, I like this solution:

Code: Select all

const int DIM = 600000;
const int FIELD_WIDTH = 6;
char buf[FIELD_WIDTH + 1] = { 0 };

fd = fopen("add.dat");
fseek(fd, FIELD_WIDTH * (DIM * x + y) );
fread(fd, buf, FIELD_WIDTH);
ans = atoi(buf);

The problem is that this would have to come with an 8 TB database of answers, as it needs to work with numbers up to 500,000-some in both positions.

[hotaru]

The problem is that this would have to come with an 8 TB database of answers, as it needs to work with numbers up to 500,000-some in both positions.

actually, it only needs to work with the test cases listed here.
and only in that order... so actually you could just have it display those answers in that order, one each time "=" is pressed, and it would be correct for that contest. it's only required to display the results on the screen (you don't have to display the numbers as they're entered) and those are the only test cases that will be used, so you can just ignore all the other buttons...

[EvanED]

The problem is that this would have to come with an 8 TB database of answers, as it needs to work with numbers up to 500,000-some in both positions.

actually, it only needs to work with the test cases listed here.
and only in that order... so actually you could just have it display those answers in that order, one each time "=" is pressed, and it would be correct for that contest. it's only required to display the results on the screen (you don't have to display the numbers as they're entered) and those are the only test cases that will be used, so you can just ignore all the other buttons...

Yes, but that destroys the elegance of my solution.

[Andrew]

actually, it only needs to work with the test cases listed here.
and only in that order... so actually you could just have it display those answers in that order, one each time "=" is pressed, and it would be correct for that contest.

Yeah, I think that kind of detracts from the fun of it. It seems like it'd be much better to produce a working calculator that operated on an insane system than produce something that'd throw up a set list of numbers by any system. I hope the winner is a proper calculator.

[evilbeanfiend]

yes i agree, i also think that adding any operation that a modern compiler can just elide should not be included in the winner. we want wtfery that cant be fixed by simply selecting and deleting some preamble.

[Yakk]

We could view the + and * operations as unknown functions, and do analysis to determine what the correct result would be.

Heck, we could solve the problems in the frequency domain.

[evilbeanfiend]

and store all the numbers internally in a strange base.

[Andrew]

If it integer only was acceptable I'd love to store all numbers as arrays of prime factors. That'd give great multiplication and division routines, and I'd be interested to know if addition could be done without conversion back to numbers (which would ideally only be done by the display routine). I suppose you could count "-1" as prime for the sake of negative numbers.

But I don't think you could shoehorn decimals into there, so that'd kill it.

[evildave]

Format the arguments into the command line for the 'infamous sed calculator', then invoke sed and wait for the result, then take that result and make a number out of it again.

http://www.npcguild.org/~ksb/hack/math.sed

[Yakk]

Just keep track of where the decimal point is is.

You can do finite-decimal math using integers pretty damn well.

[EvanED]

I can't think of any way to do addition with sets of factors though, besides either converting back to integers or special casing it.

[tendays]

Something like this would be a start:

Code: Select all

int mult(int a,int b) {
int r=0;
for (int i=0;i<a;i++) {
r += b;
}
return r;
}


(is it supposed to work with negative numbers, too?)
That was the easy part. Then:

Code: Select all

int sum(int a,int b) {
int c=mult(1<<a,1<<b);
for (int r=0;c;r++) c>>1;
return r;
}


(That's the general idea - I tried to put myself in the mind of a wtf-generating person)
(would probably have to generalise the multiplication operator to work on large numbers)

[Scarblac]

(That's the general idea - I tried to put myself in the mind of a wtf-generating person)
(would probably have to generalise the multiplication operator to work on large numbers)

No no no, that would be improving your code

If it fails some tests, you can always add special cases for each. Or something.

[Andrew]

Something like this would be a start:

Code: Select all

int mult(int a,int b) {
int r=0;
for (int i=0;i<a;i++) {
r += b;
}
return r;
}


(is it supposed to work with negative numbers, too?)

More to the point, that doesn't work on non-integers.

Imagine how beautiful the code would be with interpolation.

Even in

Code: Select all

int add(int a,int b) {
int r=0;
for (int i=0;i<a;i++) {
r ++;
}
for (int i=0;i<b;i++) {
r ++;
}
return r;
}


...you do it twice and take a weighted average. And the results would be bang on, too.

[Hamorad]

This looks fun

Code: Select all

int doAdd(int lhs, int rhs)
{

bool a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,flase,ture;int x,y,z; for(z=0;x<89430;y++--)a=false;b=true;c=true;d=a;a=c; d=b;b=k;n=o;o= a;flase=true;ture=i; i=l;e=f;l=s;q=r;r=q;s=n;o =o;q=r;d=e; flase= ture;i=q;n=o; j=s;h=j;j=i;if (r==q)z=12; r=s; x += z+1020;j =k;n=q;x=y; ture=g;h=flase;int result=lhs;x=y;z=q;l=m;e=d;h=i;ture =a;a=true;ture=a;b= flase;while(result<(lhs+rhs))++result;flase=true;ture=i; i=l;e=f;l=s;q=r;r=q;s=n;o=o;!r;q=r;x=900494;/*return result;*/q=r;r=q;s=n;o =o;q=r;d=e;d=b;b=k;n=o;o= a;flase=true;ture=i; i=l;e=f;l=s;!n;q=r;r=q;s=n;o =o;q=r;d=e; flase= ture;i=q;n=o; j=s;h=j;j=i;if (r==q)z=12; r=s; x += z+1020;j =k;n=q;x=y; ture=g;h=flase;/*icecreamgoeshere=s*/e=f;l=s;q=r;r=q;s=n;o i=l;e=f;l=s;q=r;r=q;s=n;o =o;q=r;d=e; flase= ture;i=q;n=o; j=s;h=j;j=i;if (r==q)z=12;=o;q=r;d=e; flase= ture; for(double kfol = 3.84590; x< 0289;x--+++++++-);flase=true;ture=i; i=l;e=f;l=s;q=r;r=q;s=n;o =o;q=r;x=900494;return result;q=r;r=q;s=n;o =o;q=r;q=s;r=p;n=o;d=e;d=b;b=k;n=o;o= a;flase=true;ture=i; i=l;e=f;z=0;!q;l=s;q=r;r=q;s=n;o =o;q=r;

return result;
}

There are some errors in there - I'm aware of that... but it's late, I don't feel like looking for them, and ugh. Pah.

[Hex]

I hope to enter when I get an idea clever enough .

[Taejo]

I hope to enter when I get an idea clever enough .

It's probably best to get started while you're still clueless.

[umbrae]

One of my ideas that I'm probably not going to use since it's too obvious is to pipe your input through Google Calculator and print the result.

Plams, an idea for you:

Pipe your input through all three of the major search engines: google, yahoo, and live. (They all do calculator stuff).

If they don't end up having the same answer, return "I don't know"

Edit: I might end up doing this myself.

[Strilanc]

The winning code should definitely have "estimate" somewhere in the comments, preferably in the integer addition part.

[parallax]

If it integer only was acceptable I'd love to store all numbers as arrays of prime factors. That'd give great multiplication and division routines, and I'd be interested to know if addition could be done without conversion back to numbers (which would ideally only be done by the display routine). I suppose you could count "-1" as prime for the sake of negative numbers.

But I don't think you could shoehorn decimals into there, so that'd kill it.

Allow your prime factors to have negative exponents. Then you can handle all rational numbers.

I don't think there are any good addition routines, however.

[adlaiff6]

This looks fun

Code: Select all

int doAdd(int lhs, int rhs)
{

bool a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,flase,ture;int x,y,z; for(z=0;x<89430;y++--)a=false;b=true;c=true;d=a;a=c; d=b;b=k;n=o;o= a;flase=true;ture=i; i=l;e=f;l=s;q=r;r=q;s=n;o =o;q=r;d=e; flase= ture;i=q;n=o; j=s;h=j;j=i;if (r==q)z=12; r=s; x += z+1020;j =k;n=q;x=y; ture=g;h=flase;int result=lhs;x=y;z=q;l=m;e=d;h=i;ture =a;a=true;ture=a;b= flase;while(result<(lhs+rhs))++result;flase=true;ture=i; i=l;e=f;l=s;q=r;r=q;s=n;o=o;!r;q=r;x=900494;/*return result;*/q=r;r=q;s=n;o =o;q=r;d=e;d=b;b=k;n=o;o= a;flase=true;ture=i; i=l;e=f;l=s;!n;q=r;r=q;s=n;o =o;q=r;d=e; flase= ture;i=q;n=o; j=s;h=j;j=i;if (r==q)z=12; r=s; x += z+1020;j =k;n=q;x=y; ture=g;h=flase;/*icecreamgoeshere=s*/e=f;l=s;q=r;r=q;s=n;o i=l;e=f;l=s;q=r;r=q;s=n;o =o;q=r;d=e; flase= ture;i=q;n=o; j=s;h=j;j=i;if (r==q)z=12;=o;q=r;d=e; flase= ture; for(double kfol = 3.84590; x< 0289;x--+++++++-);flase=true;ture=i; i=l;e=f;l=s;q=r;r=q;s=n;o =o;q=r;x=900494;return result;q=r;r=q;s=n;o =o;q=r;q=s;r=p;n=o;d=e;d=b;b=k;n=o;o= a;flase=true;ture=i; i=l;e=f;z=0;!q;l=s;q=r;r=q;s=n;o =o;q=r;

return result;
}

There are some errors in there - I'm aware of that... but it's late, I don't feel like looking for them, and ugh. Pah.

If only you used all those variables.

[evilbeanfiend]

This looks fun

Code: Select all

int doAdd(int lhs, int rhs)
{

bool a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,flase,ture;int x,y,z; for(z=0;x<89430;y++--)a=false;b=true;c=true;d=a;a=c; d=b;b=k;n=o;o= a;flase=true;ture=i; i=l;e=f;l=s;q=r;r=q;s=n;o =o;q=r;d=e; flase= ture;i=q;n=o; j=s;h=j;j=i;if (r==q)z=12; r=s; x += z+1020;j =k;n=q;x=y; ture=g;h=flase;int result=lhs;x=y;z=q;l=m;e=d;h=i;ture =a;a=true;ture=a;b= flase;while(result<(lhs+rhs))++result;flase=true;ture=i; i=l;e=f;l=s;q=r;r=q;s=n;o=o;!r;q=r;x=900494;/*return result;*/q=r;r=q;s=n;o =o;q=r;d=e;d=b;b=k;n=o;o= a;flase=true;ture=i; i=l;e=f;l=s;!n;q=r;r=q;s=n;o =o;q=r;d=e; flase= ture;i=q;n=o; j=s;h=j;j=i;if (r==q)z=12; r=s; x += z+1020;j =k;n=q;x=y; ture=g;h=flase;/*icecreamgoeshere=s*/e=f;l=s;q=r;r=q;s=n;o i=l;e=f;l=s;q=r;r=q;s=n;o =o;q=r;d=e; flase= ture;i=q;n=o; j=s;h=j;j=i;if (r==q)z=12;=o;q=r;d=e; flase= ture; for(double kfol = 3.84590; x< 0289;x--+++++++-);flase=true;ture=i; i=l;e=f;l=s;q=r;r=q;s=n;o =o;q=r;x=900494;return result;q=r;r=q;s=n;o =o;q=r;q=s;r=p;n=o;d=e;d=b;b=k;n=o;o= a;flase=true;ture=i; i=l;e=f;z=0;!q;l=s;q=r;r=q;s=n;o =o;q=r;

return result;
}

There are some errors in there - I'm aware of that... but it's late, I don't feel like looking for them, and ugh. Pah.

doesn't that come under the heading of obfuscation rather than wtfery?

[adlaiff6]

My plan so far is to implem^H^H^H^H^H^Hsimulate a Turing Machine with two counters, and use that to make a 4-function calculator.

I expect failure.

[evilbeanfiend]

looking at the rules i think the obfuscated one is indeed not allowable

Since the fine folks over at the IOCCC have already compiled some of the most beautiful Ugly Code ever created, and since looking at Ugly Code is more or less what many of us do each day at work, I’ve decided to exclude Ugly Code as an option altogether. In other words, entries for this contest will need to focus on clean and human-readable code that is, first and foremost, Clever and, if desired, Buggy.

[Yakk]

My plan so far is to implem^H^H^H^H^H^Hsimulate a Turing Machine with two counters, and use that to make a 4-function calculator.

I expect failure.

Have the turing machine simulate a turning machine, that simulates a simple VN processor, that simulates a turing machine, that implements the simple calculator.

And have a bug in one of the turing machines.

[adlaiff6]

I don't think I'll be able to add anything more than 3+3 with my current setup; I don't want to make it any harder.

[Amnesiasoft]

I was tempted to enter, but the Dream Build Play contest has much better prizes.

I did like my idea of emulating the universe, emulating a quantum computer, emulating a pentium with the FDIV error. But then decided it would be too slow.

[bavardage]

I think it would be rather funky to do a wiki style calculator - connect it to an online database, and if you type in a calculation that had not already been entered you get a "This calculation has no answer currently. If you would like to give a value for this calculation click here"

Just make sure that the required calculations are in the database upon entry, and also do something that puts all calculations into a common form: i.e. typing 3+4 would query the database for the same thing as 4+3.

If I had C++ GUI skillz I would be tempted to try, but I could only really ever create that kindof thing in java.

No stealing of teh idea, or I will have to find out where you live and give you a hug (or force you to find bugs in my badly written SourceModding C++.)