i can call the function “cbfunc(myst *mt)” directly.
then why i am assigning “cbfunc(myst *mt)” function to “callback” function pointer & then calling it.

okki

http://www.ibm.com/developerworks/library/l-osmig3.html#N101FA
http://linux.die.net/man/3/sem_wait


void GuaranteedSleep(uint32_t msec)
{
struct timespec timeout0;
struct timespec timeout1;
struct timespec* tmp;
struct timespec* t0 = &timeout0;
struct timespec* t1 = &timeout1;

while ((nanosleep(t0, t1) == (-1)) && (errno == EINTR))
{
tmp = t0;
t0 = t1;
t1 = tmp;
}
}


When I actually constructed this into my program, I ran across a couple problems. (I’m on a mac, using standard C++)

First is, my compiler complained it did not know what rl_abort was. I searched the web, couldn’t find anything, checked out the readline doc, and they recommend: rl_insert. Tried that, fixed the compile issue.

After I got the basic program working with autocomplete I integrated this into my application and got a segfault whenever I hit tab.

This took a while for me to set up gdb and walk through this but I traced it back to the array:
char* cmd [] ={ “hello”, “world”, “hell” ,”word”, “quit”, ” ” };

and

while ((name = cmd[list_index])

I’m not sure if this was something wrong with my implementation (I copied your code verbatim, even the cmd string), but the while loop overruns the array. I added a return character to the array, identifying this as the end of the array, and checking for this character before continuing. Ex:

char* cmd [] ={ “hello”, “world”, “hell” ,”word”, “quit”, ” “, “\r” }; and

while ((name = cmd[list_index]) && strncmp(cmd[list_index],”\r”, 1) != 0)

Thanks again, everything is working great now.

If one needs background processing while waiting, then one would do much better by using a proper multi-event capable select() mechanism in the app.
If one needs such handling, then this is the way to go.
If one for some reason doesn’t want to use a select()-based mechanism, then launch one or more worker threads.
But simply burning CPU for CPU upgrading’s sake (the infamous Wintel behaviour) is a no-no.

void msleep(int ms) {
struct timespec time;
time.tv_sec = ms / 1000;
time.tv_nsec = (ms % 1000) * (1000 * 1000);
nanosleep(&time,NULL);
}