[linux-dvb] Re: backwards compatability -was- actual cvs broken?

[Philip Prindeville]

Manu Abraham wrote:

> Edgar Toernig wrote:
>
>> It should be possible to create a dummy bttv driver that only
>> supplies GPIO and I2C access.  This dummy driver could even be
>> part of the dvb-bt878 driver.  (I may be wrong but wasn't the
>> first version of the dvb-bt878 driver like that?)  But you
>> have to find someone who's willing spends some hours of work
>> to do that :-)  I wouldn't - it's unsatisfying to invest a
>> lot of work just to "remove" functionality.
>>
>>  
>>
>
> Yeah, i can understand your feelings. It is kind of very frustraing 
> when working with a card than spans subsystems, especially when you 
> have a layered module. I have been subjected to the extreme 
> frustration a while back, that bttv in V4L used to get screwed up, 
> resulting in breakages elsewhere. The clueless users will keep 
> complaining that other modules are not functioning as expected. Well, 
> this is really hard.
>
> Well, this is the disadvantage of not having code duplication.
>
> But the real aspect of it is, that if all modules were to have code 
> duplication, imagine the sixe of the kernel, and maintainabilty issues.
>
> The adavantage of not having code duplication is that bugs can be 
> easily identified out, as the code review does not happen for a 
> particular hardware, but happens in a generic manner.
>
> The only way compatibility can be ensured is subsystems play nice to 
> each other. A while back it was quite difficult to get dvb-kernel and 
> CVS to play nice to each other. Anyhow that situation has improved a 
> lot, and it is not as bad as it was earlier. Anyway, both aspects has 
> their own merits and demerits.
>
> Manu



What about putting the glue code into a pseudo-device module that both
the V4L and DVB modules depend on?

-Philip