[mythtv] re: Broken DEC2000-T filtering. (was generic digest subject)

[Nathaniel Daw]

On Wed, 20 Apr 2005, Taylor Jacob wrote:

> The implimentation of the dvb drivers in Myth is as it should be per the
> linuxtv-dvb api.

And I am suggesting a trivial patch in which they will still be as they
should be (in fact exactly equivalent), per the api, and thus will work
exactly the same with the majority of clients but in practice will work
better with one real-world driver on most real-world networks.

I don't understand what is the problem with that? It's not some ugly hack,
and it's not stopping you from picking up 0x46 where you can. (In fact,
0.18 already contains special-purpose hacks to support this hardware.)

> Tracking/handling the 0x46 as well as 0x42 tables is important, but not
> yet not implimented in myth, so even if by some miracle the 0x46 swapped
> with 0x42 networks fixed their crap I still would have to filter the
> same PID twice with your suggestion..  I think you really need to get
> onto linux-dvb and try and get the driver fixed or just get a new piece
> of hardware..

Yes, to perfectly support this driver (in some future world when you are
actually reading 0x46 tables) you would have to filter twice. That is a
red herring. In particular, it is not a reason not to make a trivial
change that will allow the driver to work adequately now while leaving the
code correct and making no difference for other drivers. As it stands,
you're going to get either 0x42 or 0x46 on this driver and both on
everything else -- why not make the choice that works on most networks?

Sure, I should also try to get the driver fixed, assuming that is
possible. If I could do so, the small change to mythtv I am proposing
would literally make no difference.  But it would in the meantime make
mythtv work adequately for most people who own this hardware and haven't
updated.

Best wishes,

Nathaniel