From Matějů Miroslav, Ing. in 2016 - correspondence summarised below I'd be pleased if you referred me. I have a website at http://mirekm.wz.cz/ but I've not updated it for ages. I must put relevant information there and create English version soon. Best regards, Miroslav > -----Original Message----- > From: Andrew Wood > Sent: Monday, May 16, 2016 12:57 AM > To: Matějů Miroslav, Ing. > Subject: Re: ETA calculation in PV > > Thank you for your email. I will look into using clock_gettime(), as > it sounds like it will work much better. > > Do you have a web site I can link to in the contributors section of the > documentation? > > > On Mon, May 09, 2016 at 08:16:45AM +0000, Matějů Miroslav, Ing. wrote: > > Thank you for making PV, a great tool to monitor a long-lasting pipe > > process. I have one instance being run for 97 hours already! > > > > However, I experienced a problem in the calculation of ETA. The above > > mentioned process happens in a virtual machine that I suspend regularly. > > The ETA calculation seems to include the time when the machine was > > suspended into the elapsed time and the reported ETA always increases > > abruptly after waking the machine. > > > > I looked into the PV source and discovered that you use the function > > gettimeofday() for getting elapsed time. Its manual page says that it is > > affected by discontinuous jumps in the system time like time-zone change. > > Moreover, it is marked as obsolete in POSIX.1-2008. It would be probably > > better to use clock_gettime() with CLOCK_MONOTONIC or CLOCK_MONOTONIC_RAW > > which would provide correct results.