[linux-dvb] Re: wanting to contribute dvb-t tuning information but small problem with scan

[Johannes Stezenbach <js@convergence.de>]

Christoph Burger-Scheidlin wrote:
> 
> On Wed, 2004-07-14 at 15:23, Johannes Stezenbach wrote:
> > Should be fixed in CVS. Please test.
> 
> I just tried the CVS. I do not get the infinite loop any longer, but
> still experience some strange behaviour.
> 
> The tuning still is trying to access the wrong frequency once.

That is expected and unavoidable. We cannot know which frequency is
the correct one before we've tried it. Actually, it might even be
possible that more than one frequency works and we must chose the one
which works best. But I cannot test this here in Berlin (yet).

> This
> causes  some of the channels to be discovered twice. Scan however stops
> after discovering those. As you can see, scan discovers a total of 19
> channels. In fact I can only see 15 channels. 4 channels are discovered
> twice with slightly different settings in channels.conf, differing in
> the second FEC value.

The ones with FEC_NONE are wrong.

It seems that those FEC_NONE values crept in through the initial
data files. I added code in scan.c to use FEC_AUTO instead.

(AFAIK hierarchical coding is not yet used in Germany, so the
code_rate_LP is irrelevant, but FE drivers reject FEC_NONE
because it is invalid for DVB-T).

I fixed a bug where the same frequency would be scanned twice
in CVS. Please test again.


Johannes