What may the C program output if we assign a character without ending with a null character in different declaring ways?

Question

I have been reading C Primer Plus (5th edition) by Stephen Prata. The Program listing 11.9 is a program to show what may the puts() function do if we declare a string and assign a value without an ending null character. (Of course this is viewed as a grammar mistake, but I want to know what error will this bug cause.) I experimented on my computer (with CodeBlocks my ide) a similar program, and tried different ways to declare the string.

For the program

#include <stdio.h>

int main()
{
     char a[]={'a','b','c','d'};
     char b[100]="Stack Overflow";
     puts(a);
     return 0;
}

in which I used a[] to declare the string, the program indeed output some thing other than abcd . I tried several times, and it always output abcd< and an empty line:

output:

abcd<

If I declare the string by a[100] , i.e. the program

#include <stdio.h>

int main()
{
     char a[100]={'a','b','c','d'};
     char b[100]="Stack Overflow";
     puts(a);
     return 0;
}

Then it seems quite normal. The program always output abcd and an empty line after I have repeated several times:

output:

abcd

If I use the pointer to declare the string:

#include <stdio.h>

int main()
{
     char *a={'a','b','c','d'};
     char b[100]="Stack Overflow";
     puts(a);
     return 0;
}

Then it only output an empty line:

output:

Why different ways of declaring the string, which is assigned a value without an ending '\0' yields different outputs?

*I want to know what error will this bug cause.* Undefined behavior is **undefined**. Anything can happen - including appearing to work in some way. — Andrew Henle, Apr 08 '23 at 13:48
Does this answer your question? [Undefined, unspecified and implementation-defined behavior](https://stackoverflow.com/questions/2397984/undefined-unspecified-and-implementation-defined-behavior) — Karl Knechtel, Apr 08 '23 at 13:51
@AndrewHenle: No, the meaning of the term “undefined behavior” is limited **only** to what the C standard says or does not say about it. It **only** means the **C standard** does not impose any requirements on the behavior. It does not mean there is no other information about the behavior. — Eric Postpischil, Apr 08 '23 at 13:57
You cannot "declare a string". You can declare an array of char. If that array contains a certain set of values, then it is a string. If that char array does not contain the proper values, then it is not a string. — William Pursell, Apr 08 '23 at 14:10
@KarlKnechtel This reference is very helpful. But it seems that this post did not discuss this specific situation of undefined behaviour. — Asigan, Apr 08 '23 at 14:45
@EricPostpischil *It does not mean there is no other information about the behavior.* Whatever information that can be gleaned from the results of undefined behavior will be unreliable while being particular to the compiler, the compiler options used, the platform, the phase of the moon, the signs of the zodiac, chicken bones, and the current migrations of Norwegian lemmings. Even on the same platform and in the exact same environment, it can not be used to reliably predict future behavior. On another platform or with a different compiler, the information is effectively useless. — Andrew Henle, Apr 08 '23 at 15:04
When _I_ ran this, I got: `abcd╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠╠Stack Overflow` as output. So it kept printing whatever was in memory until it encroached on `b` which _did_ have a null termination. — Nick is tired, Apr 08 '23 at 15:42
@AndrewHenle: No, it is not unreliable, nor does the fact that it may be particular to a compiler, a platform, or whatever mean it is not useful and worth knowing. Being familiar with what behaviors various kinds of bugs typically cause helps diagnose software problems. It saves time tracking down the wrong cause. It is useful for planning mitigations and hardening software against exploits. It is useful for attackers (good or bad) seeking exploits. It is useful for understanding how hardware and software work below the abstract level of a programming language… — Eric Postpischil, Apr 08 '23 at 16:03
… Dismissing it as “Undefined behavior is undefined” promotes ignorance. It is harmful to education. — Eric Postpischil, Apr 08 '23 at 16:04

score 3 · Accepted Answer · edited Apr 08 '23 at 14:10

char a[100]={'a','b','c','d',};

If an aggregate (an array or structure) is explicitly initialized but not completely, the elements or members without initializers are implicitly initialized the same way as objects with static storage duration (per C 2018 6.7.9 21). For char elements, that means they are initialized to zero (per 6.7.9 10). So, after a[0] through a[3] are initialized with the given character codes, a[4] is initialized to zero. So the string beginning at the start of a is null terminated, by the null character in a[4], and passing that string to puts prints a string in the ordinary way.

char *a={'a','b','c','d',};

This is not a proper way to initialize a. {'a','b','c','d',} does not represent an array; it is a list of initial values. Since a is declared as a pointer, it should be initialized with a single value that is a pointer.

When you compiled this, the compiler warned you that 'a' was the wrong type to initialize a, and it warned you there were excess initializers in the list.

When the compiler proceeded after this warning, it converted 'a' to a pointer, which typically yields an invalid address. It is surprising that puts(a) did not cause the program to crash in attempting to access that address. (Perhaps you are running in a GUI, saw an empty output window, and did not notice a separate message about the program crashing?) If the program did not crash, there could have been a zero byte at the address a was initialized with, so passing a to puts printed nothing.

Thanks, this answer is detailed, and very helpful! – Asigan Apr 08 '23 at 14:43 — Asigan, Apr 08 '23 at 14:43

What may the C program output if we assign a character without ending with a null character in different declaring ways?

1 Answers1