>but i cannot make good quick sort algorithm
There are three immediate improvements to the basic algorithm. First, you want to come up with a partitioning scheme that finds a pivot point as close to the median as possible so that the recursive branching is balanced. Second, you want to avoid recursing for small sets but adding a cutoff when the partitions are small. Lastly, you can partition three or more ways and then work out an elegant handling of duplicate values.

The first improvement is usually made by either choosing a random pivot as cscgal did, or by choosing three values in the partition and using the median of those three as the pivot. You could use more than three, but that means extra work and extra time, just like calling a random number generator. Both are undesirable in an efficient sorting routine. The second improvement is usually made by setting a cutoff in the recursive path and then using insertion sort to finalize the routine. Because insertion sort is zippy on almost sorted files, this increases quicksort's speed.

The last improvement is complicated and I don't see it much, so unless large amounts of duplicate values are a problem for you, you can ignore it.

The most obvious problem with your code aside from the syntax errors in spilt is that you mix prototypes with old style function definitions. This is a huge no-no. Let K&R C die and use proper ISO C function definitions please. For help on the actual algorithm, start by throwing away that awful book you're using. The code rarely works without a great deal of tweaking. :rolleyes: The text is wonderful, but the code just sucks.