Hi,

I ran your code and it does run well. However fails for large cases such as pascal(200000).

You mentioned you worked for Google and Amazon, how can you modify the code to get an answer for pascal(200000)

Yeah, for large cases the long will overflow (also it will do a lot of realloc()). Your two main options for this are:

  1. use double to get approximately-correct answers
  2. use an abitrary-precision library like GMP

What if I wanted an exact precision. Could I use a custom data type to get better precision? If I used a custom data type, how could I make the code work for a distributed system to get parallel behavior?

upvoted for realloc trick

This inspired me to write one that is even shorter and doesn't heap-allocate any memory at all. What can I say, I'm a sucker for minimal C.

  #include <stdio.h>
  #include <stdlib.h>

  int main(int argc, char *argv[]) {
    int row = atoi(argv[1]);
    long data[row];
    for (int i = 0; i < row; i++) {
      data[i] = 1;
      for (int j = i - 1; j > 0; j--) data[j] += data[j - 1];
    }
    for (int i = 0; i < row; i++) printf("%ld ", data[i]);
    printf("\n");
    return 0;
  }

When did C start accepting non-constants for array size declarations?

I was wondering the same thing. But I can confirm it compiles and runs, though I had to supply a -std=c99 flag to gcc.

Ahhh, useful tip. I've just been using alloca instead...

Variable-length arrays were added in C99.

Best solution I've seen here so far.

If row is too big you can bust your stack frame when you define your data array.

True, though in this example integer overflow would happen first.

I think your code is off by 1.