Re: [PATCH v2 4/4] pwm: sysfs: Utilize an array for polarity strings

From: Andy Shevchenko
Date: Mon Aug 29 2022 - 03:59:53 EST


On Sun, Aug 28, 2022 at 9:19 PM Joe Perches <joe@xxxxxxxxxxx> wrote:
> On Sun, 2022-08-28 at 20:40 +0300, Andy Shevchenko wrote:
> > On Sun, Aug 28, 2022 at 4:46 PM Joe Perches <joe@xxxxxxxxxxx> wrote:
> > > On Sun, 2022-08-28 at 09:40 +0300, Andy Shevchenko wrote:
> > > > On Sunday, August 28, 2022, Joe Perches <joe@xxxxxxxxxxx> wrote:
> > > > > On Fri, 2022-08-26 at 20:07 +0300, Andy Shevchenko wrote:
> > > > > > Code is smaller and looks nicer if we combine polarity strings
> > > > > > into an array.
> >
> > > > First of all, please remove unnecessary context when replying.
> > >
> > > I am _very_ aware of context.
> > > I specifically left the code in.
> > >
> > > > > It's less robust though as PWM_POLARITY_NORMAL and _INVERSED
> > > > > are now required to be 0 and 1. As the only 2 values in
> > > > > an enum they are, but that's not really guaranteed unless
> > > > > you read the enum definition.
> > > >
> > > > So, what do you suggest here and in many other similar places (yes, ABI
> > > > implied) in the kernel?
> > >
> > > Leaving the code alone.
> >
> > It's good that PWM maintainers look at this differently.
>
> The enum is not userspace so it's not ABI.
>
> The PWM maintainers are free to do what they want but I
> prefer obviousness over compactness.

Why do you not start "fixing" other similar places in the kernel?

--
With Best Regards,
Andy Shevchenko