Re: [PATCH v1 2/3] gpiolib: Rename gpio_set_debounce_timeout() to gpiod_do_set_debounce()

[Andy Shevchenko <andriy.shevchenko@linux.intel.com>]

On Tue, Mar 04, 2025 at 01:15:02PM +0100, Bartosz Golaszewski wrote:
> On Tue, Mar 4, 2025 at 1:00 PM Andy Shevchenko
> <andriy.shevchenko@linux.intel.com> wrote:
> > On Tue, Mar 04, 2025 at 01:31:35PM +0200, Mika Westerberg wrote:
> > > On Tue, Mar 04, 2025 at 01:16:54PM +0200, Andy Shevchenko wrote:
> > > > On Tue, Mar 04, 2025 at 01:11:57PM +0200, Mika Westerberg wrote:
> > > > > On Tue, Mar 04, 2025 at 12:59:25PM +0200, Andy Shevchenko wrote:
> > > > > > On Tue, Mar 04, 2025 at 11:18:04AM +0200, Mika Westerberg wrote:
> > > > > > > On Mon, Mar 03, 2025 at 06:00:33PM +0200, Andy Shevchenko wrote:
> > > > > > > > In order to reduce the 'gpio' namespace when operate over GPIO descriptor
> > > > > > > > rename gpio_set_debounce_timeout() to gpiod_do_set_debounce().
> > > > > > >
> > > > > > > To me anything that has '_do_' in their name sounds like an internal static
> > > > > > > function that gets wrapped by the actual API function(s).
> > > > > > >
> > > > > > > For instance it could be
> > > > > > >
> > > > > > >   int gpio_set_debounce_timeout()
> > > > > > >   {
> > > > > > >       ...
> > > > > > >       gpiod_do_set_debounce()
> > > > > > >       ...
> > > > > > >
> > > > > > > However, gpiod_set_debounce_timeout() or gpiod_set_debounce() sounds good
> > > > > > > to me.
> > > > > >
> > > > > > Then please propose the second name for gpiod_set_config_XXX to follow
> > > > > > the same pattern. The series unifies naming and reduces the current
> > > > > > inconsistency.
> > > >
> > > > > gpiod_set_config()?
> > > >
> > > > The problem is that
> > > >
> > > > gpiod_set_debounce() and gpiod_set_config() are _existing_ public APIs.
> > > > That's why I considered "_do_" fitting the purpose.
> > >
> > > I see.
> > >
> > > Hmm, we have:
> > >
> > > int gpiod_set_debounce(struct gpio_desc *desc, unsigned int debounce)
> > > {
> > >         unsigned long config;
> > >
> > >         config = pinconf_to_config_packed(PIN_CONFIG_INPUT_DEBOUNCE, debounce);
> > >         return gpiod_set_config(desc, config);
> > > }
> > >
> > > and
> > >
> > > int gpio_set_debounce_timeout(struct gpio_desc *desc, unsigned int debounce)
> > > {
> > >       int ret;
> > >
> > >       ret = gpio_set_config_with_argument_optional(desc,
> > >                                                    PIN_CONFIG_INPUT_DEBOUNCE,
> > >                                                    debounce);
> > >       if (!ret)
> > >               gpiod_line_state_notify(desc, GPIO_V2_LINE_CHANGED_CONFIG);
> > >
> > >       return ret;
> > > }
> > >
> > > I wonder if there is an opportunity to consolidate? ;-)
> >
> > Send a patch! I would be glad to see less functions and internal APIs in
> > GPIOLIB.
> >
> 
> I'm definitely in favor of consolidation instead of renaming to
> gpiod_go_set_debounce(). If anything a better name would be:
> gpiod_set_debounce_nocheck() to indicate the actual functionality.
> 
> How about first extending gpio_set_config_with_argument() to take a
> boolean "optional" argument and removing
> gpio_set_config_with_argument_optional() altogether? Both are internal
> to drivers/gpio/ so it would have no effect on consumers.

Consider this series as a report then, I am not going to spend time on it.
Thank you for the review.

-- 
With Best Regards,
Andy Shevchenko