K&R2, exercise 5.3

[arnuld]

PURPOSE: write a pointer version of strcat function.

WHAT I DID: I have 2 implementations of this function. Both compile
and run but 1st one gives some strange results, 2nd is ok. I am unable
to figure out why the 1st implementation gives strange results:
/* A rudimentary strcat function
*
*/
#include <stdio.h>
#include <stdlib.h>
enum MAXSIZE { ARR_SIZE = 1000 };
void my_strcat( char s[], char t[] );
int main()
{
char s[1000] = "Saurabh";
char t[1000] = "Nirkhey\n" ;

printf("\ns --%s\nt --%s\n\n", s, t );
printf("--- concatenating strings -----\n\n");
my_strcat( s, t );
printf("\ns --%s\nt --%s\n\n", s, t );

return EXIT_SUCCESS;
}
void my_strcat( char s[], char t[] )
{
char *ps, *pt;

ps = s;
pt = t;

while( *ps++ != '\0' )
{
printf("*ps = %c\n", *ps );
}

while( *pt != '\0' )
{
*ps++ = *pt++;
}

}

=========== OUTPUT ===============

/home/arnuld/programs/C $ gcc -ansi -pedantic -Wall -Wextra 5-3.c
/home/arnuld/programs/C $ ./a.out

s --Saurabh
t --Nirkhey
--- concatenating strings -----

*ps = a
*ps = u
*ps = r
*ps = a
*ps = b
*ps = h
*ps = ^@

==2nd version's strcat function:

void my_strcat( char s[], char t[] )
{
char *ps, *pt;

ps = s;
pt = t;

while( *ps != '\0' )
{
printf("*ps = %c\n", *ps );
ps++;
}

while( *pt != '\0' )
{
*ps++ = *pt++;
}

}

============ OUPUT ============
/home/arnuld/programs/C $ gcc -ansi -pedantic -Wall -Wextra 5-3.c
/home/arnuld/programs/C $ ./a.out

s --Saurabh
t --Nirkhey
--- concatenating strings -----

*ps = S
*ps = a
*ps = u
*ps = r
*ps = a
*ps = b
*ps = h

s --SaurabhNirkhey

t --Nirkhey
/home/arnuld/programs/C $


-- http://lispmachine.wordpress.com/

Please remove capital 'V's when you reply to me via e-mail.

[Irrwahn Grausewitz]

arnuld schrieb:

PURPOSE: write a pointer version of strcat function.

WHAT I DID: I have 2 implementations of this function. Both compile
and run but 1st one gives some strange results, 2nd is ok. I am unable
to figure out why the 1st implementation gives strange results:

[SNIP]

void my_strcat( char s[], char t[] )
{
char *ps, *pt;

ps = s;
pt = t;

Here the pointer ps gets incremented, regardless of the result of
the comparison. That way, in the loop you print always the character
*following* the one ps pointed to when the loop condition was last
checked; the output of your debug print statements should have been
a hint...

while( *ps++ != '\0' )
{
printf("*ps = %c\n", *ps );
}

At this point, ps points one character *beyond* the terminating null
character in s. Effectively, you are now going to copy the contents
of t to "dead"[1] space in the array s:

while( *pt != '\0' )
{
*ps++ = *pt++;
}

[SNIP]

==2nd version's strcat function:

[SNIP]

This time you got it right: increment ps only if the loop condition is
true (and debug-print the character ps points to *before* incrementing
the pointer).

while( *ps != '\0' )
{
printf("*ps = %c\n", *ps );
ps++;
}

[SNIP]

Alternatively, you could have used a for loop, of course:

for ( ps = s; *ps != '\0'; ++ps )
/* skip, or whatever */;

[1] In the sense of: "will be ignored, when s is treated as a string".
Best regards
--
Irrwahn Grausewitz [ir*******@freen et.de]

[Philip Potter]

arnuld wrote:
[problem with strcat implementation]
[first version, buggy:]

void my_strcat( char s[], char t[] )
{
char *ps, *pt;

ps = s;
pt = t;

while( *ps++ != '\0' )
{
printf("*ps = %c\n", *ps );
}

while( *pt != '\0' )
{
*ps++ = *pt++;
}

}

==2nd version's strcat function:

[working]

>
void my_strcat( char s[], char t[] )
{
char *ps, *pt;

ps = s;
pt = t;

while( *ps != '\0' )
{
printf("*ps = %c\n", *ps );
ps++;
}

while( *pt != '\0' )
{
*ps++ = *pt++;
}

}

You have correctly identified the location of the problem. Specifically,
the problem is the following while loop in the first version:

while (*ps++ != '\0')

This says "Check whether *ps equals '\0', and afterwards, increment ps".
Note that ps is incremented whether or not the equality test succeeds.

Look at the printf() output. By the time printf() prints *ps, ps has
already been incremented. This is why you never output the first
character of the string.

Similarly, when you detect a '\0' character, you then increment ps past
that character. As a result, the '\0' is never overwritten with a
character from pt; the contents of the char array after the call are as
follows:

"Saurabh\0Nirkh ey\n"

The '\0' part-way through a char array means "end of string", so the
printf() call in main() won't print anything beyond that point.

I hope this helps you understand why your first version doesn't work but
your second version does.

[Peter 'Shaggy' Haywood]

Groovy hepcat arnuld was jivin' in comp.lang.c on Tue, 8 Apr 2008 8:04
pm. It's a cool scene! Dig it.

PURPOSE: write a pointer version of strcat function.

WHAT I DID: I have 2 implementations of this function. Both compile
and run but 1st one gives some strange results, 2nd is ok. I am unable
to figure out why the 1st implementation gives strange results:

In addition to the problems pointed out by other people, you have
another serious bug in both versions of this.

void my_strcat( char s[], char t[] )
{
char *ps, *pt;

ps = s;
pt = t;

Why create more variables? s and t are both pointers. I know they look
like arrays, somewhat, but they're pointers; and they're local to the
function. You don't need to keep their initial values for anything, so
you can use these instead of ps and pt.
But that's just a matter of style more than anything else. It's not
the source of the problem I mentioned.

while( *ps++ != '\0' )

Problem here already mentioned by others.

{
printf("*ps = %c\n", *ps );
}

while( *pt != '\0' )
{
*ps++ = *pt++;
}

Here's the other problem. You copy every character pointed at by pt to
the corresponding location indicated by ps. However, you don't
terminate the resulting string with a null character. This is a serious
problem that just happens to have eluded you by sheer luck. Whether
that luck is good or bad depends on your outlook. Do you consider it
good luck to have a serious bug lurking in code, unbeknownst to the
coder, waiting to pounce and bring down a billion dollar client's
entire network when he least expects it, costing billions of dollars to
fix? (I know, I know. This code is just a small program whose purpose
is to educate you on the ways of C. But in future, when you use C in
the workplace, errors like this are easy to miss. So you need to learn
about these things now, so you'll know what to look for then.)
Actually, the reason you've not seen this problem is that your arrays
in main() were initialised, which sets the entire contents of the
array. Array elements beyond the length of the initialiser are filled
with the value 0. So your concatenated string just happens to be null
terminated. But as a general replacement for strcat(), this function is
a disaster waiting to happen.
You need to explicitly terminate the string, which is as simple as
this:

*ps = '0';

}

The other version of your function has the same problem as I have
mentioned here.

--
Dig the sig!

----------- Peter 'Shaggy' Haywood ------------
Ain't I'm a dawg!!