c pointer and array question

[Bo Sun]

hi:

please take a look at the following code fragment:

char* hello = "Hello World\n";

hello[5] = 's';

why I cannot modify hello[5]?

thanks,

bo

 

[Christopher Benson-Manica]

char* hello = "Hello World\n";
hello[5] = 's';

why I cannot modify hello[5]?

Short answer: Because the Standard says so.

hello points to a string literal. The Standard states that attempting
to modify a string literal (as you are trying to do) results in
undefined behavior. Some implementations may let you get away with it
anyway; however, in many (most?) cases, string literals are stored in
read-only memory, and attempting to modify the contents of such memory
is a sure way to cause problems for yourself.

 

[Alan Balmer]

hi:

please take a look at the following code fragment:

char* hello = "Hello World\n";

hello[5] = 's';

why I cannot modify hello[5]?

thanks,

bo

Christopher told you why. What you probably want to do is:

char hello[] = "Hello World\n";

Try that, then think about the difference.

 

[Kevin Goodsell]

hi:

please take a look at the following code fragment:

char* hello = "Hello World\n";

Never do this. You can't modify a string literal, so make your intention
to NOT modify it clear by making the pointer const:

const char *hello = "Hello World\n";

I would strongly recommend enabling the option to have your compiler
warn you about this, if you compiler has such an option. In gcc, the
option is -Wwrite-strings.

hello[5] = 's';

After the change above, this will cause your compiler to issue a
diagnostic. This is good, since modifying a string literal (or
attempting to) gives undefined behavior, and will probably crash your
program or worse.

why I cannot modify hello[5]?

You cannot modify a string literal.

-Kevin

 

[Jared Dykstra]

char* hello = "Hello World\n";
hello[5] = 's';

why I cannot modify hello[5]?

Short answer: Because the Standard says so.

hello points to a string literal. The Standard states that attempting
to modify a string literal (as you are trying to do) results in
undefined behavior. Some implementations may let you get away with it
anyway; however, in many (most?) cases, string literals are stored in
read-only memory, and attempting to modify the contents of such memory
is a sure way to cause problems for yourself.

If you are intentionally using string literals--and I recognize you
are not in this case--you should use the const keyword. Then if you
did try to change it, the compiler would complain in most cases.
Note, however, that const can always be circumvented.

 

[Chris Torek]

char* hello = "Hello World\n";
hello[5] = 's';

[does not work; why not?]

Others have already posted correct answers (which boil down to
"because the C standard says it does not have to work", and in your
case your C compiler makes sure it does not work), but I want to
clarify one bit. The thing with the double quotes above *is* a
"string literal", in the Standard's terminology, but the array to
which "hello" points is not actually a "string literal".

String literals are purely source-code constructs, rather like
keywords. If you write "int x, y;" then x and y have type "int",
but there is no requirement -- and often no actual -- run-time
information lying around saying "this memory over here is what was
originally named x, and has type int". One can reverse-engineer
object code and often figure out that "there is an int or long in
this memory over here", but perhaps not which one it is (int or
long), and in most cases, there is no way at all to recover the
name of the variable.

In any case, the anonymous array that the string literal produces
in this context has type "array 12 of char", rather than the more
logical "array 12 of const char". If you apply The Rule about
arrays and pointers in C (see http://web.torek.net/torek/c/pa.html),
you will find that "array 12 of char" is usable wherever a value
of type "char *" is required. Thus, this is a valid initializer
for the variable "hello" -- but the array's elements are allowed
to be, and your compiler makes them, read-only.

Note that when a string literal is used as an initializer for an
object of type "array N of char", the string literal does not (or
at least need not) produce any array of its own, and the characters
inside the string literal are used instead to initialize the object.
The "const"-ness (i.e., readonly-ness) of the object is then
controlled by the object's type: if it is "array 12 of char" it is
read/write; if it is "array 12 of const char" it is read-only.
I wrote a long article about this just a week or so ago.

 

[Keith Thompson]

char* hello = "Hello World\n";
hello[5] = 's';

why I cannot modify hello[5]?

Short answer: Because the Standard says so.

hello points to a string literal. The Standard states that attempting
to modify a string literal (as you are trying to do) results in
undefined behavior. Some implementations may let you get away with it
anyway; however, in many (most?) cases, string literals are stored in
read-only memory, and attempting to modify the contents of such memory
is a sure way to cause problems for yourself.

A small (and mostly irrelevant) clarification:

In most systems, string literals are not actually stored in ROM (i.e.,
in memory chips whose contents physically cannot be modified by
software). They're usually stored in the same kind of RAM
(random-access read/write memory) as all the other code and data in
your operating system and programs. (I'll ignore virtual memory.)

When I ran a program that tried to modify a string literal, I got a
segmentation fault -- not because the string literal was stored in
ROM, but because the linker and operating system arranged for the
memory to be read-only for my program. It was written by the OS when
it loaded my program, and again when my program terminated.

In an embedded system, however, string literals, along with your
program's code, may be stored in some kind of ROM. In that case,
you'd have to burn your program to a ROM chip (or EEPROM, or whatever)
before you could run it.

The standard doesn't distinguish between these two cases. Trying to
modify a string literal results in undefined behavior, whether it's
because the memory is marked read-only by the OS, or because it's
burned into ROM, or because it's engraved on non-erasable stone
tablets. It could even succeed in modifying it; that's just one of
the infinitely many possible consequences of undefined behavior.

(Note that the phrase "modifying a string literal" is a shorthand. A
string literal exists only in a program's source code. What you're
trying to modify is the anonymous static array that's initialized from
the string literal.)

 

[Christopher Benson-Manica]

A small (and mostly irrelevant) clarification:
(string literal clarification)

Irrelevant? Hardly! I appreciate knowing all the details - thanks.