Johannes Stezenbach wrote:
Holger Waechtler wrote:
one more point came to my mind: the 2.6 driver API provides a clean way
to upload demod firmware files, no need for dirty hacks and workarounds
and direct access of internal dvb_adapter or i2c_adapter struct members.
That's the whole point why we want to use kernel-I2C. We get
firmware loading (and the whole sysfs foobar) for free.
Well, so I thought...
why insert an abstraction layer inbetween that does not provides any
additional functionality, needs special care because of distributed
locking, has a more complicated-to-use API and just hides functionality
that could get used easier by accessing the cleaner lower-level API
directly?
just my 2¢,
Holger