[linux-dvb] Re: blocking FE_SET_FRONTEND ioctl in non-blocking mode

[Andrew de Quincey <adq_dvb@lidskialf.net>]

On Wednesday 12 May 2004 03:33, you wrote:
> On Wednesday 12 May 2004 02:13, Andrew de Quincey wrote:
> > We could remove it if people want - but it'll break anything which is
> > sensitive to the hw_sections=0 problem.
> >
> > BTW: I only have budget cards, and they had the same issue, which is why
> > I added it.
> >
> > So: remove+fix the userspace software to have a flag to ignore data until
> > tuning is completed (or something), or leave+have a long delay... or some
> > other solution. Suggestions?
>
> Hmm, MythTV (and VDR I belive) handles this just case with strange data
> from channel changes just fine, so I guess my oppinion is drop the delay,
> because it's the applications domain to do filtering. Would not this
> overlap of data occur on regular channel changes to? I'm pretty sure I came
> across this when developing the MythTV DVB code, so what is the point of
> doing filtering on open, but not general channel changes?

Hmm, the code I'm talking about would occur on all calls to FE_SET_FRONTEND, 
not just open(). 

Another thing - the original poster (Franck Arnaud) didn't specify which 
version of the drivers he was using. There was a bug in older dvb-kernels 
which would cause a long-ish multi-second delay on every tune attempt. This 
is fixed in recent versions.

I'll do some tests tonight to check.


-- 
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe linux-dvb" as subject.