On Mon, Feb 12, 2024 at 02:56:43PM +0100, Krzysztof Kozlowski wrote:
> On 12/02/2024 14:39, Andy Shevchenko wrote:
> > On Mon, Feb 12, 2024 at 09:34:24AM +0100, Krzysztof Kozlowski wrote:
...
> >> - i2c {
> >> - #address-cells = <1>;
> >> - #size-cells = <0>;
> >>
> >> - pcf8574: pcf8574@27 {
> >> - compatible = "nxp,pcf8574";
> >> - reg = <0x27>;
> >> - gpio-controller;
> >> - #gpio-cells = <2>;
> >> - };
> >> - };
> >
> > In patch 3 you updated the lines that have lost their sense due to this one.
>
> How did they lose it?
Now they are referring to the non-existed node in the example. OTOH, there is
already hc595 case...
The Q here (as you pointed out that it's better to name nodes in generic way),
how these names are okay with the schema (hc595, pcf8574) as being referred to?
...
> > And I agree with others, please leave this example in place.
>
> What for? Why this binding is special and 99% of others do not need GPIO
> expander in the example?
Some people already tried to explain you their point of view, but I see that:
- the unrelated nodes in the schemas are not welcome (as per your talks
and documentation);
- the current file has other references that have no existing node in the
example;
- you are DT maintainer, so I believe you know this better.
With this, I'm almost (see above question though) satisfied with the series.
--
With Best Regards,
Andy Shevchenko
