On Sat, Jun 2, 2018 at 12:28 AM, Daniel Drake <drake@xxxxxxxxxxxx> wrote: > On Thu, May 31, 2018 at 4:43 AM, Andy Shevchenko > <andy.shevchenko@xxxxxxxxx> wrote: >>> + depends on ASUS_WMI >> >> I'm not sure it's a good idea. >> >> Rather you may use ifdef inside header which also provides a stub for >> the function. But be careful with corner cases, like ASUS_WMI=m && >> HID_ASUS=y. > > What do you see as wrong with it? The driver already depends on e.g. > LEDS_CLASS for the functionality that it sometimes uses there. > > In my opinion the ideal solution to the corner cases you mention is > the Kconfig "depends on" which is exactly what the patch is > proposing... depends on in this case adds less generic dependency to the more generic module. Thus, we imply that we have more users when dependency is satisfied than otherwise. Which is simply not true. Thus, you add a bulk dependency to unsuspecting users w/o any need. -- With Best Regards, Andy Shevchenko
