Re: [PATCH] media: Doc add missing documentation for samsung,exynos4212-jpeg

[Jacek Anaszewski]

On 07/15/2016 11:30 AM, Krzysztof Kozlowski wrote:
> On 07/15/2016 11:18 AM, Jacek Anaszewski wrote:
> > On 07/15/2016 10:33 AM, Krzysztof Kozlowski wrote:
> > > On 07/15/2016 10:28 AM, Jacek Anaszewski wrote:
> > > > On 07/15/2016 10:17 AM, Krzysztof Kozlowski wrote:
> > > > > On 07/15/2016 10:14 AM, Jacek Anaszewski wrote:
> > > > > > > However if these compatibles are exactly equal then
> > > > > > > only one should be preferred. It makes everything easier. Second
> > > > > > > can be
> > > > > > > still documented e.g. as deprecated.
> > > > > >
> > > > > > Still, both of them are present in the driver. Shouldn't it be
> > > > > > reflected
> > > > > > in the documentation?
> > > > >
> > > > > Right, it is a good practice, so how about:
> > > > >
> > > > >      - compatible    : should be one of:
> > > > >              "samsung,s5pv210-jpeg", "samsung,exynos3250-jpeg",
> > > > >              "samsung,exynos4210-jpeg", "samsung,exynos5420-jpeg",
> > > > >              "samsung,exynos5433-jpeg";
> > > > >
> > > > >              Deprecated: "samsung,exynos4212-jpeg"
> > > > >
> > > > > (or any other formatting)
> > > > > plus update to DTS changing it to 4210?
> > > >
> > > > Why newer 4212 version should be made deprecated?
> > >
> > > I don't mind the other way. However it seems logical to me that newer
> > > chip is compatible with existing one so the existing one (older) is
> > > used. When adding support for new devices, for most of re-usable drivers
> > > we use old compatibles. But as I said, it doesn't really matter to me.
> >
> > Frankly speaking marking a compatible deprecated looks weird to me.
> > It can be interpreted in the way that the device itself is deprecated
> > or it is not fully reliable.
>
> Marking a compatible or a property deprecated is commonly used, if
> needed of course. It has nothing to do with device being deprecated.
> This is documentation for bindings and deprecation affects only
> bindings. It is not weird or something strange. We already did this for
> some of Exynos compatibles (later removing them) and there are quite
> many examples in Documentation already.

If this is broadly accepted pattern, then I will not argue against.
Let's proceed as you proposed.

--
Best regards,
Jacek Anaszewski