cc [ flag ... ] file ... -lgen [ library ... ]
#include <libgen.h>

int p2open(const char *cmd, FILE *fp[2]);

int p2close(FILE *fp[2]);

The p2open()gfunction forks and execs a shell running the command line pointed to by cmd. On return, fp[0] points to a FILE pointer to write the command's standard input and fp[1] points to a FILE pointer to read from the command's standard output. In this way the program has control over the input and output of the command.

The function returns 0 if successful; otherwise, it returns -1.

The p2close() function is used to close the file pointers that p2open() opened. It waits for the process to terminate and returns the process status. It returns 0 if successful; otherwise, it returns -1.

A common problem is having too few file descriptors. The p2close() function returns -1 if the two file pointers are not from the same p2open().

Example 1 Example of file descriptors.

#include <stdio.h>
#include <libgen.h>

main(argc,argv)
int argc;
char **argv;
{
 FILE *fp[2];
 pid_t pid;
 char buf[16];

 pid=p2open("/usr/bin/cat", fp);
 if ( pid == -1 ) {
 fprintf(stderr, "p2open failed\en");
 exit(1);
 }
 write(fileno(fp[0]),"This is a test\en", 16);
 if(read(fileno(fp[1]), buf, 16) <=0)
 fprintf(stderr, "p2open failed\en");
 else
 write(1, buf, 16);
 (void)p2close(fp);
}

See attributes(7) for descriptions of the following attributes:

fclose (3C), popen (3C), setbuf (3C), attributes (7)

Buffered writes on fp[0] can make it appear that the command is not listening. Judiciously placed fflush() calls or unbuffering fp[0] can be a big help; see fclose(3C).

Many commands use buffered output when connected to a pipe. That, too, can make it appear as if things are not working.

Usage is not the same as for popen(), although it is closely related.