Hi Peter,
I appologize, situation has been different.
Please forget my "good ideas".
I have not seen this part of code since 2002 :-)
Am Wed, 11 Jul 2018 09:55:01 +0200
schrieb Jens Thoms Toerring <
address@hidden>:
Hi,
On Wed, Jul 11, 2018 at 05:31:11AM +0200, Werner Heisch wrote:
As far as I remember, the problem has been a task change beween
setitimer() and signal(). setitimer has triggered the signal before
the signal has been defined.
Couldn't that have been fixed by simply first installing
the signal handler and only then calling setitimer() (which
the example code probably also should do!)?
You are right, it was the opposite sequence :
signal()
setitimer()
pause()
What I needed was only a usleep().
setitimer is fireing only one time.
Problem was: If pause() has not been completely finished while the
signal initiated by setitimer has fired, the program never woke up.
The original program came from Coherent and only had to sleep for some
milliseconds. On Coherent, this was nap (ms).
In '95 ported to Linux: nap() did not exist, usleep() would be the
equivalent but also does not exist in those days.
It also can be built by select() .. I know today.
cheers
Werner