You, of course, already knew this. But I learn from your comments, so please leave some. Thanks!
This may be the next “old dog, new trick” I adapt too.
When I started learning C in the late 1980s, I had a compiler manual (not very useful for learning the language) and a Pocket C Reference book — both for pre-ANSI K&R C. I may have had another “big” C book, but I mostly remember using the Pocket C Book.
Looking back at some of my early code, I find I was declaring “fixed” strings like this:
And this shows us:
char version[5]="0.00"; /* Version number... */Odd. Did I really count the bytes (plus 0 at the end) for every string like that? Not always. I found this one:
char filename[28]="cocofest3.map";…but I think I remember why 28. In the OS-9/6809 operating system, directory entries were 32 bytes. The first 28 were the filename (yep, back in the 80s there were operating systems with filenames longer than FILENAME.EXT), and then three at the end were the LSN (logical sector number) where the File ID sector was. (More or less accurate.)
I also found arrays:
int *days[] = { "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat" };But why is that an int??? That must have been a typo/bug. Other places, I did it more correctly:
char *items[] = { /* employee info prompt thingies */
"Employee :",
"Min/Week :",
"Max/Week :",
"Min/Shift:",
"Max/Shift:"
};At some point in my C history, I just started using pointers to strings like this:
char *version = "1.0.42b-delta";I guess I got tired of [brackets]. I mean, “they work the same way”, don’t they?
void function (char *line)
{
if (NULL != line)
{
printf ("Line: '%s'\n", line);
}
}…and…
void function (char line[])
{
if (NULL != line)
{
printf ("Line: '%s'\n", line);
}
}…both end up with line pointing to the start of wherever the bytes to that string are in memory. I’ve seen main() done the same ways:
int main (int arc, char *argv[] )…and…
int main ( int argc, char **argv )For years, I’ve been doing it the second way, but all my early code was *argv[] so I suspect that is how I learned it from my early K&R C books.
I have no idea why I changed, or when, but probably in the mid-to-late 1990s. I started working for Microware Systems Corporation in Des Moines, Iowa in 1995. This was the first place I used an ANSI-C compiler. In code samples from the training courses I taught, some used “*argv[]” but ones I wrote used “**argv”.
Does it matter?
Not really. But let’s talk about it anyway…
There was a comment left on one my articles last year that pointed out something different I had no considered: sizeof
If you have “char *line” you cannot use sizeof() to give you anything but the size of the pointer (“sizeof(line)”) or the size of a character (or whatever data type used) that it points to (“sizeof(*line)”).
If you have “char line[]”, you can get the size of the array (number of characters, in this case) or the size of one of the elements in it:
#include <stdio.h>
#include <stdlib.h> // for EXIT_SUCCESS
int main(void)
{
char *line1 = "1234567890";
char line2[] = "1234567890";
printf ("sizeof(line1) = %zu\n", sizeof(line1));
printf ("sizeof(*line1) = %zu\n", sizeof(*line1));
printf ("\n");
printf ("sizeof(line2) = %zu\n", sizeof(line2));
printf ("sizeof(*line2) = %zu\n", sizeof(*line2));
return EXIT_SUCCESS;
}This produces:
sizeof(line1) = 8 <- size of a 64-bit pointer
sizeof(*line1) = 1 <- size of a char
sizeof(line2) = 11 <- size of the character array
sizeof(*line2) = 1 <- size of a char
I cannot remember ever using sizeof() on a string constant. You may recall I was surprised it worked when I learned about it a few months ago.
But, now that I am aware, I think I may start moving myself back to where I started and using the [brackets] when I have constant strings. Using sizeof() in the program just embeds a constant value, while strlen() is a function that walks through each byte looking for the end zero, thus adding more code space and more execution time.
If I wanted to copy some constant string into a buffer, I could try these two approaches:
// Copy message into buffer.
char *line1 = "This is a message.";
strncpy (buffer, line1, strlen(line1)); // strlen
printf ("Buffer: '%s'\n", buffer);
char line2[] = "This is a message.";
strncpy (buffer, line2, sizeof(*line2)); // sizeof
printf ("Buffer: '%s'\n", buffer);And the results are the same:
Buffer: 'This is a message.'
Buffer: 'This is a message.'
I would use the second version since using sizeof(*line2) avoids the overhead of strlen() scanning through each byte in the string looking for the end zero.
NOTE: As was pointed out in the comments, strlen() returns the number of characters up to the zero. “Hello” is a strlen() of 5. But sizeof() is the full array or characters including the 0 at the end so “Hello” would have a sizeof() of 6.
char line[] = "1234567890";
printf ("strlen(line) = %u\n", strlen(line));
printf ("sizeof(line) = %u\n", sizeof(line));strlen(line) = 10
sizeof(line) = 11
If you wanted them to be the same, it would be “sizeof(line)-1”.
It’s all fun and games until you pass a parameter…
This “benefit” of sizeof() is not useful if you are passing the string in to a function. It just ends up like a pointer to wherever the string is stored:
#include <stdio.h>
#include <stdlib.h> // for EXIT_SUCCESS
#include <string.h>
void function1 (char *line)
{
printf ("function1():\n");
printf ("sizeof(line) = %zu\n", sizeof(line));
printf ("sizeof(*line) = %zu\n", sizeof(*line));
printf ("strlen(line) = %zu\n", strlen(line));
}
void function2 (char line[])
{
printf ("function2():\n");
printf ("sizeof(line) = %zu\n", sizeof(line));
printf ("sizeof(*line) = %zu\n", sizeof(*line));
printf ("strlen(line) = %zu\n", strlen(line));
}
int main(void)
{
char *line1 = "1234567890";
printf ("Line 1: '%s'\n", line1);
function1 (line1);
function2 (line1);
printf ("\n");
char line2[] = "1234567890";
printf ("Line 2: '%s'\n", line2);
function1 (line2);
function2 (line2);
return EXIT_SUCCESS;
}Above, I create a “*line” pointer to a string then pass it in to two functions. The first expects a *line as the parameter, and the second expects a line[].
Then I do a “line[]” array and pass it to the same two functions.
The results are the same:
Line 1: '1234567890'
function1():
sizeof(line) = 8
sizeof(*line) = 1
strlen(line) = 10
function2():
sizeof(line) = 8
sizeof(*line) = 1
strlen(line) = 10
Line 2: '1234567890'
function1():
sizeof(line) = 8
sizeof(*line) = 1
strlen(line) = 10
function2():
sizeof(line) = 8
sizeof(*line) = 1
strlen(line) = 10
And, if you use a “good compiler,” you may get a warning about doing sizeof() like this:
main.c: In function ‘function2’:
main.c:16:44: warning: ‘sizeof’ on array function parameter ‘line’ will return size of ‘char *’ [-Wsizeof-array-argument]
16 | printf ("sizeof(line) = %zu\n", sizeof(line));Notice that warning was from function2(), and not from function1(). This is one difference in using “*line” versus “line[]” in the functions. For function1(), no warning is given:
void function1 (char *line)
{
printf ("function1():\n");
printf ("sizeof(line) = %zu\n", sizeof(line));
printf ("sizeof(*line) = %zu\n", sizeof(*line));
printf ("strlen(line) = %zu\n", strlen(line));
}Since the function takes a “pointer to one or more chars”, doing sizeof() what that pointer points to makes sense. It is what you asked for. The “C gibberish” website says:
declare line as pointer to char
– https://cdecl.org/?q=char+*line
But for the second one…
void function2 (char line[])
{
printf ("function2():\n");
printf ("sizeof(line) = %zu\n", sizeof(line));
printf ("sizeof(*line) = %zu\n", sizeof(*line));
printf ("strlen(line) = %zu\n", strlen(line));
}…a warning is given about the “sizeof(line)” because it cannot tell us the size of a line[] array — it became a pointer to the character memory when it went into the function. But because the function parameter was “line[]”.
declare line as array of char
– https://cdecl.org/?q=char+line%5B%5D
Doing sizeof() an “array of char” is valid. But it was passed into the function, even though the parameter was a line[] it is passed as a pointer to the data. I guess this is one of those “I’m sorry, Dave. I’m afraid I can’t do that” moments ;-)
Is this useful? It certainly will let you use sizeof() instead of strlen() on a string if you have direct access to the string variable. But passing strings into functions? Not so much. (Or am I mistaken?)
But I do think I am going to try to go back to using “line[]” for my string declarations. I like retro.
Until next time…
