C:UNIX-systems programmingnewbie question- copying file using System call

[creeping death]

#include<unistd.h>
#include<sys/types.h>
#include<sys/stat.h>
#include<fcntl.h>
#include<stdlib.h>

int main( int argc,char *argv[] )
{
    char buf;
    int sourcefile,destfile,n;
    if(argc!=3)
    {
        write(STDOUT_FILENO,"prgm1 <sourcefile> <destination file>\n",50);   
        exit(1);
    }
    else
    {
         sourcefile=open(argv[1],O_RDONLY);
         if(sourcefile==-1)
         {
            perror("SOURCE FILE ERROR");
            exit(0);
         }
         else
         {
            destfile=open(argv[2],O_WRONLY | O_CREAT , 0641);
            if(destfile==-1)
            {
                perror("DESTINATION FILE ERROR");
                exit(0);
            }
            else
            {
                while((n=read(sourcefile,&buf,1)) != -1)
                {
                    write( destfile, &buf, 1 );
                }
                write(STDOUT_FILENO, "FILES COPIED\n" , 15);    
                close(sourcefile);
                close(destfile);
            }
        }
    }
    return 0;
}

its says in the textbook that read() will return -1 on error(like not able to read from the sourcefile in this case)
but the value returned by read() never becomes -1 ? why?

[matsp]

Because reaching the end of file is not an error in the respect that "you have passed in a handle that isn't valid" or "you don't really want to do that". It is just "there was nothing more to read", which results in a zero-length. Since zero is not -1, you never get out of your loop.

It is perfectly valid to read from a file, get zero bytes, then wait for, say, 10 seconds, and read again - if that gave back -1 as an error, how do you know the difference between that, and "someone removed the disk that the data is on" - the latter should probably make the program exit, or at least close the file and try to open in it again, whilst "nothing more to read" should not cause the program to exit, just do whatever it does when there was nothing else to read. A command that does something like this is "tail -f somefile" - if you send some (slow) output from a file (say a loop that prints a number every second), then you can see the output in another window using "tail -f ..."

--
Mats

[creeping death]

Because reaching the end of file is not an error in the respect that "you have passed in a handle that isn't valid" or "you don't really want to do that". It is just "there was nothing more to read", which results in a zero-length. Since zero is not -1, you never get out of your loop.

It is perfectly valid to read from a file, get zero bytes, then wait for, say, 10 seconds, and read again - if that gave back -1 as an error, how do you know the difference between that, and "someone removed the disk that the data is on" - the latter should probably make the program exit, or at least close the file and try to open in it again, whilst "nothing more to read" should not cause the program to exit, just do whatever it does when there was nothing else to read. A command that does something like this is "tail -f somefile" - if you send some (slow) output from a file (say a loop that prints a number every second), then you can see the output in another window using "tail -f ..."

--
Mats

my textbook says

if successful, read() returns the number of bytes that it read, otherwise , it returns -1

i misunderstood this...
thanks,problem solved

[creeping death]

ok, i have another doubt on open()
i change my c-code
from

sourcefile=open(argv[1],O_RDONLY);

to

sourcefile=open(argv[1],O_RDONLY,0664);

shouldn't open() be able to override the default permissions already set?
i tried and apparently it does not. is there any reason, besides enforcing security(thats the only thing that i could think of)? and is there any way around this? (besides giving the file permissions using the system("chmod "); function?)

[matsp]

I guess you could use the chmod() system call instead of calling system("chmod ...").

Setting the mode of the file is non-trivial, as there are a number of things that affect the mode. So you may want to look at the documentation for chmod() to see what it says [I think it explains it] - note, it's "man 3 chmod" or some such, as "man chmod" will give you the command, not the function.

--
Mats

[creeping death]

yeah.
thanks again

its
man 3p chmod for fedora....POSIX specs of chmod system call...