Re: [LAD] Library for accurate time stamping using a DLL

Stefan Kost wrote: Thanks for this remark. I did not know clock_gettime(). However, the real question may not be which clock source is the most reliable, but which one is used as the master clock. For example, in FFmpeg, each audio and video packets are tagged using a PTS which is computed using av_gettime() which calls gettimeofday(). This is the master timing source to synchronize the audio and video input streams which may well be generated according to their own clocks. This is why I thought about adding a new function to the API, pendule_set_master_time_reader(double (* func) (void)) or similar, that would allow to set which ever function you like to retrieve the master time. This function should returned the current time as a double representing the number of seconds since the Epoch. pendule_cycle() must be called before pendule_gettime() or pendule_stats(). That's the way all fields get initialized. Otherwise the results are undetermined. I suppose stating this in the (to be written) docs should be sufficient. Initializing values in pendule_new() is irrelevant, since this can only be done in real-time, which pendule_cycle() is for. For example, setting current_time to zero (1/1/1970 at midnight) in pendule_new() isn't more relevant than any other random value to me. I'll try to add doc blocks to pendule.h soon. -- Olivier