shell sort example in "Advanced C"

[sshader]

Hi, I'm starting to get into C and am reading "Advanced C" by Herbert Schildt. He uses this example of the shell sort (K&R style?):

void shell(item, count)
char *item;
int count;
{
register int i, j, k, s, w;
char x, a[5];

a[0]=9; a[1]=5; a[2]=3; a[3]=2; a[4]=1;

for(w=0; w<5; w++)  {
k = a[w]; s = -k;
for(i=k; i<count; ++i)  {
x = item[i];
j = i-k;
if(s==0) {  s = -k;
s++;
item[s] = x;
}
while(x<item[j] && j>=0 && j <=count)  {
item[j+k] = item[j];
j = j-k;
}
item[j+k] = x;
}
}
}

so my question is what is 's' for? why would it ever be 0? thanks

[phantomotap]

O_O

Wow.

You are in desperate need of a different book.

Soma

[anduril462]

To further explain Soma's comment a bit, the author Schildt sucks so badly, they made up a word for it: bullschildt.

[rcgldr]

There are better sort methods like quick sort or merge sort, but here is an example of a shell sort. This example is either from an old book or from some web article:

void shell_sort(int *a, int n)
{
int h, i, j, k;     /* h is "gap" size */

    for (h = n; h /= 2;){
        for (i = h; i < n; i++){
            k = a[i];
            for (j = i; j >= h && k < a[j - h]; j -= h){
                a[j] = a[j - h];
            }
            a[j] = k;
        }
    }
}

int main(int argc, char *argv[])
{
int a[] = {41, 53, 26, 0, 67, 98, 39, 84, 72, 15};
int n = sizeof(a) / sizeof(a[0]);

    shell_sort(a, n);

    return 0;
}

Wiki article about shell sort, including variations on the gap size (which is what "h" is used for in the above example):

http://en.wikipedia.org/wiki/Shell_sort

[sshader]

I'd heard of schildts reputation but I suppose I was just curious of his thought process. guess it's undecipherable here. Thanks again

[oogabooga]

It's so badly written that it's literally painful to read.

To answer your original question: s doesn't seem to be doing anything and I can't imagine how it would ever become 0.

[laserlight]

It's so badly written that it's literally painful to read.

I had the impression that they are usually stylistically well written and hence a pleasure to read... if you enjoy fiction.

[phantomotap]

I had the impression that they are usually stylistically well written and hence a pleasure to read... if you enjoy fiction.

^_^v

Delicious!

Soma

[MutantJohn]

When I read this question, I expected more Bourne-syntax and stuff lol.

I've actually never once heard of shell sorting before this.

[Adak]

When I read this question, I expected more Bourne-syntax and stuff lol.

I've actually never once heard of shell sorting before this.

When T-Rex ruled the food chain, Shell sort was the top dog in-place comparison sorter. This was before Quicksort became well known, and some quirks in picking the pivot were worked out.

Shell sort uses Insertion sort, but the changing gap helps it to move many values a larger distance, with far fewer comparisons. As cache sizes have increased, Shell sort performance has improved. Until you get past 50,000 numbers, it's hard to beat Shell sort.

[Cat]

I had the impression that they are usually stylistically well written and hence a pleasure to read... if you enjoy fiction.

I tend to be a bit more generous to Schildt than others. I would compare his writing style to that of a fourth-grade science book - oversimplified and imprecise, but with the intent to deliver some key ideas without being bogged down in details. In that kind of niche, his work has a use for someone still learning the very basics. Once you've learned the fundamentals, you will have little use for his work. It's about as useful as a reference guide for a serious programmer as One Fish, Two Fish is for an ichthyologist.

[whiteflags]

You don't have to dumb down science for 4th graders, you just have to be Bill Nye.

[Malcolm McLean]

I tend to be a bit more generous to Schildt than others. I would compare his writing style to that of a fourth-grade science book - oversimplified and imprecise, but with the intent to deliver some key ideas without being bogged down in details. In that kind of niche, his work has a use for someone still learning the very basics. Once you've learned the fundamentals, you will have little use for his work. It's about as useful as a reference guide for a serious programmer as One Fish, Two Fish is for an ichthyologist.

As the author of a programming book myself, I've got quite a bit of sympathy for Schildt.

Schildt writes a beginner's guide to C, and he's quite readable. That's why the books are so popular. A lot of the time, you actually make a point harder to understand by insisting on complete and total accuracy. Some of the mistakes are just mistakes, but even here, arguably they make the book better. "The textbook is always right" is a great enemy of education.

[phantomotap]

A lot of the time, you actually make a point harder to understand by insisting on complete and total accuracy.

O_O

THIS IS PROGRAMMING!

The compiler and environment will insist on complete and total accuracy.

Some of the mistakes are just mistakes, but even here, arguably they make the book better.

No.

Mistakes are only arguably forgivable, but mistakes absolutely never make a programming book better.

That's why good writers and publishers produce freely available errata long after the book is no longer sold.

[Edit]
Oh, and before anyone reaches for "What about examples that are supposed to have syntax errors or similar?", imagine how frustrating it would be to have a "Correct this example code." style exercises which actually have no errors yet have corrections listed in supplemental materials. No, that does not help; it spreads a disease of incompetence among the new individuals which is precisely what good instruction texts alleviates.
[/Edit]

Soma

[Adak]

Programming books don't have to include every possible detail in their discussions, but the examples must compile, run, and produce accurate results.

Mistakes in a programming example, do NOT make a programming book, better.