function with no return type

hello all.

just observe this piece of code

Code:

---


fun()
{
    printf("Hello world");
}

int main()
{
      int c = 5;

      printf("c = %d\n",c);
      c=fun();
      printf("c = %d\n",c);
     
      return 0;
}

---

on compilation and execution usgin gcc, i get the answer as
c = 5
c = 11 (-----> the no of chracters printed by printf())
but since there is no return statement in fun(), the second value of c should contain a junk value.

according to k & r if a function doesnt have a return statement, its return valu is junk value.
but why does c contain the no of characters printed by printf ???

Let's look at it on the assembly level. eax is always the register of return value.
printf() sets eax to the no of characters it printed and eax is not changed anymore. So eax is still the number of characters printed by printf() when fun() is returned.

wow!!!! thanks.can you please tell me which book should i refer to for learning such concepts.
and also is the ANSI draft good enough to improve your knowledge of the C language or is it just to learn the standards

What makes you think 11 isn't a junk value?

The fact that it seems consistent doesn't make it any less junk.
The fact that it seems to be the number of characters printf printed doesn't make it any less junk.

> but why does c contain the no of characters printed by printf ???
Because that's what printf() returns, the number of characters it printed (not that anybody ever checks this).
And so by being the last thing delivering an answer inside fun(), by "proxy" as it were, it also becomes the result of fun(), because that's the only 'answer' going on at the moment.
But that's only one possible scenario.



This is why we have warnings enabled to spot dumb stuff
$ gcc -W -Wall -ansi -pedantic -O2 new.c
new.c:3: warning: return type defaults to `int'
new.c: In function `fun':
new.c:5: warning: control reaches end of non-void function

Best soluation would be to get a decent book on C programming. The detiel series is an excellent one as everything it teaches you is told in exuisit detail. It is also crammed with exercises to do after each chapter too. I use the C++ fifth edition advanced personally, but they do sell a C programming beginners book. Look on amazon or ebay.
I highly encourage you to check it out.

> can you please tell me which book should i refer to for learning such concepts.
Would be a complete waste of your time.

maxorator gave a plausable explanation of how it works for you AT THE MOMENT.
Knowing such detail of how it's done now will do you no good at all in the future. You can't use that information in any meaningful way, so there's no point in learning it.

Enable the compiler warnings, and pay attention to writing correct code which will work everywhere.

Salem why you don`t like the man to learn how to do it ?!!
is it illegal !

who say it`s useless to learn this stuff .. ?

Hackers who interest with those kinds of problems and flaws.

i think learning assembly and how compilers works is the way to learn this things

i recommanded to read this book : http://download.savannah.nongnu.org/...0-booksize.pdf

Because I've seen far too many programmers wander in with some crappy code saying "But this used to work with my old compiler".

The usual reason being they've made some vast assumption about how C works where in fact what they've really done is figured out what their old compiler did (and which their new compiler doesn't).

Now since C compilers only refer to the C standard (and not each other), figuring out the minutia of your current compiler isn't going to do you any good for your future compiler.

I'm not going to tell you want to learn and what not to learn.

But I can sure tell you that 'knowing' that fun() will return what printf() returns in the example poorly written program isn't worth knowing, because the answer is unusable in any other C program.

Quote:

---

Knowing such detail of how it's done now will do you no good at all in the future. You can't use that information in any meaningful way, so there's no point in learning it.

---

Knowing such details comes with learning assembly. You can't only learn the details :D

Why are you even talking about assembler - this is C.

More traps and pitfalls, just in case you think that getting too "chummy" with the compiler by peering at the generated assembler will be useful.
http://cboard.cprogramming.com/showt...ight=undefined
http://cboard.cprogramming.com/showt...ight=undefined
http://cboard.cprogramming.com/showt...ight=undefined
The board is full of this type of stuff.

Case in point
> printf() sets eax to the no of characters it printed and eax is not changed anymore
My processor is a MIPS, what on earth are you talking about?

Quote:

---

My processor is a MIPS, what on earth are you talking about?

---

In that case the return value would be stored in $v0 :)

the declaration of the printf is
int printf(
const char *format [,
argument]...
);

it returns the number of characters printed, or a negative value if an error occurs

you can download the MSDN labrary,which will be helpful to you

1) The code under discussion cannot be compiled if you make the warning level higher.
2) Compiler warnings are good, because they can significantly reduce the time spent on debugging. Let the compiler tell you if it sees something suspicious. Sooner or later one of those bad practices will strike back otherwise.
3) Is it really so hard to write it properly?

Code:

---

int fun()
{
    return printf("Hello world\n");
}

---

There's no point in learning to write bad code...