Hi,

I would like to convert two integer variables into string (contiguously)...So I did

char strA[1], strB[1];
sprintf(strA, "%u", intA);
sprintf(strB, "%u", intB);

printf("StrA = %s", strA); //debug
printf("StrB = %s", strB); //debug

Notice that value of strA becomes nothing, but strB has the correct value(intB). How should I overcome this problem?

Thanks

Cheers,
GH.

You need a longer character array. A C-style string of length n is n characters terminated by the NULL character (total n+1 characters); so allocating just 1 character isn't enough.

(nevermind)

Hmmm.... I solved the problem by creating a func , refer from http://www.eskimo.com/~scs/cclass/int/sx5.html

char *itoa(int n)
{
char *retbuf = malloc(25);
if(retbuf == NULL)
return NULL;
sprintf(retbuf, "%d", n);
return retbuf;
}


Thanks for your prompt reply!

OK -

1. You realize the original problem was that you weren't declaring a buffer large enough to hold the correct value, don't you?

2. It's worth pointing out that the reason you saw strB (but not strA), was that strB was *corrupting* strA because of the buffer overflow. The write to strA was undoubtedly corrupting something, too - you just didn't happen to see it. Both instances were "bad".

3. The cleanest solution is arguably to do exactly what spooon and freegianghu suggested - simply declare a larger buffer. If you declare them as local variables ("e.g. char strA[5], strB[5]"), then the size is right there in front of you, and cleanup is automatic.

4. Is "strA[5]" big enough? Only if you're sure the number isn't going to exceed four digits (0..9999).

5. What's wrong with your "*itoa()"?

It works (you can't argue with success ;-)).

But I would object to three things:
a) Why "malloc(25)"? What 32-bit "int" value is going to take 24 digits?

b) If you "malloc()" the data, you need to be sure that *somebody* does
a corresponding "free()". Otherwise you've got a memory leak.

That's one of the benefits C++ classes have over C functions: you can
encapsulate that kind of "protocol" entirely within the class itself.

c) Finally, the system already *has* a function "itoa()".
If it's got the same semantics as your function, why not just use the library?
If it's got different semantics, why don't you name it something else?

IMHO .. PSM

good points.
advice:
don't mess about with malloc until you really need to, or know what you are doing.
why do people allocate such small buffers?

if you are playing with this sort of stuff don't mess about
why not char buffer[1024]?
how many MB have you got?
it saves all sorts of grief.

Where is it (what header)?

Because it doesn't make sense? If you need an int-array with 20 elements, do you write:
int a[ 1024 ];
(I hope you don't!)
Even an unsigned 64-bit integer needs at most 20 decimal digits (plus sign, if desired), so
char buffer[ 22 ];
is enough, even on systems that treat int as 64-bit (Linux and Windows don't).

Does it hurt?
How many MB on the average PC?
Are you writing a kernel module or just messing about playing with char buffers?
It helps save beginner programmers continually core dumping.

I does hurt if you do that kind of stuff in a recursive function; it's simply a senseless waste of memory, no matter how much memory you've got.
Further, if someone reads your code, he will think: "Why does he reserve so much memory? There is probably a reason for this, so did I miss something?"

Even beginners can use their minds; they know that a 32-bit number doesn't need 1023 decimal digits. Teaching them to reserve "more memory than needed, just to be sure" does not lead to safer programs. Instead, it leads to a "don't think, use your intuition"-attitude. I suppose you agree that this is very bad.