[linux-dvb] Re: What is happening on Astra 11837 h, PID 18?

[Johannes Stezenbach <js@linuxtv.org>]

Josef Wolf wrote:
> On Tue, Feb 08, 2005 at 01:44:10AM +0100, Johannes Stezenbach wrote:
> > IIRC either ETR-211 or EN-300-468 state that it is segmented by time,
> > i.e. a segment of 16 sections covers three hours. The intention is
> > that low-end boxes without enough memory can set the section filters so
> > they just filter out the data e.g. for the next day.
> 
> While I can see the reason to segment the data, I fail to see why they
> leave holes in the list of section.  Why not just send empty sections?

Waste bandwidth so implementors don't need to strain their brains?

> > The repetition rate of the sections can be different, e.g. the data
> > for the current and next day can be sent more often than the data for
> > the rest of the week.
> 
> Hmmm, then it would be bad to wait for all required (that is, modulo the
> marked holes) sections of a table to arrive because it can take a long
> time?

IIRC ETR-211 lists worst-case repetition rates. ARD is an exception
because they started to repeat EPG data very often when they started
to broadcast an MHP-Xlet that makes use of that data...

> > You can also process each EIT section independently, there is no need
> > to collect all sections before processing.
> 
> Oh, I'd prefer to do it the other way around because in general, I don't
> like special-handling.  But when some segments are broadcasted more often
> tha other, not special-handling the EIT could result in slowing down the
> update mechanism :-(.  Is this assumption correct or am I still missing
> something?

EIT is special.

Johannes