Re: [ivtv-devel] msp3400

[Claude M. <cm...@ve...>]

Quoting PlasmaJohn <li...@pr...>:
> On Thu, 27 Mar 2003, d18c7db wrote:
> 
> > I've also been a tad too busy with real life, though I've had a brief
> > email exchange with Gerd Knorr about the msp3400 driver and he agreed
> > that probably the best way forward would be to add a new switch to
> > msp_command() that calls msp3400c_set_scart() and he suggested I send in
> > a patch.
> 
> Hrm.  In my mind, the problem is architectural.  I think what's needed here
> is
> a generalized interface (aka API) to assign msp inputs and outputs
> (scart1_in, scart2_out) to "names" ("tuner in", "radio in", "line out",
> "audio
> out").  "names" is a little inaccurate, perhaps "concept" or whatever.
> 
> Yeah, that sounds right.  The msp driver is missing a level of indirection
> here.  Am I going in the right direction?  If so, I'd like us to float that
> proposal/design back to Gerd with promises of producing a patch.
> 
> Pick one out of "m" space (AUDC_) and send it in via Gerd.  I'm thinking
> something like this: AUDC_MAP_SCART with the physical to logical mapping.
> 
> Once the mapping is in place the "normal" calls should just work.

It would be nice to have an abstraction. That said, though, here's two things to
consider:

 - Gerd is much more likely to accept a small patch at the moment until
   he gets to know more about us. Doubtful that he will go for a major
   change.. Right? Once we have that in, we can address a more long-term
   change in the msp driver if one is needed.

 - In the end, it's still the ivtv driver that needs to know about the
   mapping between input and output ports of the MSP chip. So you're
   not really gaining anything by adding that abstraction (as opposed
   to doing something like a #define XBAR_TUNER_INPUT xxx, or similar).
   The abstraction would only really help if the MSP driver could
   somehow know which input is tuner without the ivtv driver telling it
   first.

Or am I wrong?

-Claude

-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/