> > card will not pass-through any more (unless it is one of the inputs with > > amux == LINE1). > > That is fine (and expected by the user) ! > When the application close()s we pass-through again. Sorry, no. close() will never ever switch inputs (or otherwise change the status of the card). That would break stuff like "v4lctl setstation foo; v4lctl volume mute off" to turn on TV sound for station foo. It would also break the radio device quite badly. Some radio apps open the device only to change settings. > > Users will have to learn that the pass-through works only with the > > correct input anyway (unless they never actually use the driver, but why > > you would buy a card then in the first place?). > > By policy of least-surprise, the user doesn't need to learn special rules. We can't avoid that without breaking the API, and I'm not going to do that. > > LINE1 is the power-on default setting of the saa7134 chip, thus the > > cards can't show different behaviour here ... > > They can, when the manufacturer implmented audio routing by a 4052. Like the FlyVideo 2000? The LINE1-passthrough-without-driver-loaded doesn't work anyway for this card, thus it doesn't really matter ... Gerd -- You can't please everybody. And usually if you _try_ to please everybody, the end result is one big mess. -- Linus Torvalds, 2002-04-20