[mythtv] NTP config

Daniel Kristjansson danielk at cuymedia.net
Thu Oct 20 14:20:37 UTC 2005

On Thu, 2005-10-20 at 09:57 -0400, Steve Adeff wrote:
> On Thursday 20 October 2005 09:25, Ivor Hewitt wrote:
> > Daniel Kristjansson wrote:
> > > DVB/ATSC time can never be as accurate as ntpd time because the
> > > transmission, buffering, and capture delays can not be measured.
> > > But it should always be within a few seconds of the correct time,
> > > so it is more than good enough for a MythTV backend.
> >
> > Actually you might say for mythbackend purposes it might be more
> > accurate, since it's the current time that should match the start times
> > of programs accurately.... even if it is delayed slightly.
> is this why so many shows don't start "on time"? If the TV stations follow 
> their own time code then it would make more sense for mythbackend to run on 
> this time. I guess the only issue is if two stations are running on their 
> own, differing time...?

In the US companies using the public airwaves are supposed to use GPS
to set their clocks, so a program should always start a little late,
and never early. This regulation doesn't apply to cable, and I don't
know what the laws are in DVB countries. But this is what pre-roll
and post-roll are for. If you add 5 seconds on each end you should get
the entirety of any program when you don't have a recording scheduled
before or after the program, at that point MythTV ignores the pre-roll
and post-roll settings on the assumption that you would rather get the
other program than apply this extra safety buffer.

There is some controversy over whether MythTV should have a pre-roll
and post-roll which is never ignored. But this is mostly due to
channels in Australia which don't adhere to their published schedule
and air programs up to 20 minutes later than scheduled. This would
be illegal in the other countries unless they transmitted updated
EIT data or the government activated some sort of emergency broadcast
override, say in the case of global thermonuclear war.

-- Daniel

More information about the mythtv-dev mailing list