I've almost finished a little app in KDevelop; on selecting certain parameters I have created a string which is something that you would type at a bash prompt in the following basic format:

echo 58000 > /sys/class/pvrusb2/sn-8563645/ctl_volume/cur_val

Does anyone know if it's possible to execute a command from within an application? I am sure it must be achievable somehow. I have done the same thing with a Python script which was pretty straightforward.

Yes, you can do it with exec(). In your case it will probably be fork() to create two processes and exec() later.

Seeing only one example, I'd say it's easier to make open(), write() and then close().

For completeness, I mention system(), which does essentially what fork() + exec() + wait() do, all rolled into one call (but has some caveats wrt suid root programs), as well as popen(), which gives you the option of feeding data to the launched program or reading its standard output data into your program.
--- rod.

While searching on Google for how to use exec, I found system(). I knoecked up a quick program which on a button press invoked this line:

system("kwrite");

Sure enough that launched kwrite, so I assume it will execute other command line statements. I need to hook up the Hauppaugge box and see if it works now.

Just one more point: system() loads a shell, which will interpet your commandline. If your argument list is already defined, then it will be more efficient (speed, memory) to use either popen() or fork() + exec(). It is easier to monitor the state of your child process if you use exec(), by calling wait(). This can be useful if you want to kill the child, if you want to limit the number of children, etc.
Also, if any of your commandline arguments come from user input, you must be careful about what arguments you send to the system() call. This is a security issue.
--- rod.

Just to clarify, if you had this:

char buf[1024];
sprintf(buf, "kwrite \'%s\'", userarg); //userarg has already been checked for length
system(buf);

The user could inject code into the server by making their argument something like this:
' $(rm -rf your_files) '

Which would thus make the system call this:

kwrite '' $(rm -rf your_files) ''

Now I have a question: How would one execute a command (e.g. kwrite) without invoking a shell while having the ability to read the stdout from that command?

popen()
--- rod.

Unfortunately, popen() itself makes use of the shell as well (and thus carries any similar “security risk”). You can duplicate all the functionality of popen() (except the shell part) by calling pipe() and fork(). In the child you should do a dup2() on the relevant file descriptors, followed by execlp(). In the parent, you can then read() or write() (whichever you decided per the dup2()) the other end of the pipe. If you want a true “FILE *” interface, just use fdopen().