Re: UAPI Re: [PATCH 1/3] drm: Add DRM_MODE_TV_MODE_MONOCHROME

[Daniel Vetter <daniel@ffwll.ch>]

On Wed, Feb 28, 2024 at 04:22:56PM +0000, Simon Ser wrote:
> On Wednesday, February 28th, 2024 at 17:14, Maxime Ripard <mripard@kernel.org> wrote:
> 
> > > I don't know what the rules were 8 years ago, but the current uAPI rules
> > > are what they are, and a new enum entry is new uAPI.
> > 
> > TBF, and even if the wayland compositors support is missing, this
> > property is perfectly usable as it is with upstream, open-source code,
> > through either the command-line or X.org, and it's documented.
> > 
> > So it's fine by me from a UAPI requirement side.
> 
> That is not a valid way to pass the uAPI requirements IMHO. Yes, one
> can program any KMS property via modetest or xrandr. Does that mean that
> none of the new uAPI need a "real" implementation anymore? Does that mean
> that the massive patch adding a color pipeline uAPI doesn't need
> user-space anymore?

xrandr only supports properties on the connector, so it's right out for
the color pipeline.

Also "we use xrandr for color properties" very much doesn't pass the bs
filter of "is it a toy".

My take would be that this escape hatch is also not valid for all
connector property, stuff that is clearly meant to be configured
automatically by the compositors cannot use the "we use xrandr" excuse,
because users can't type fast enough and hit <Enter> precisely enough to
update a property in lockstep with the compositor's redraw loop :-)

> The only thing I'm saying is that this breaks the usual DRM requirements.
> If, as a maintainer, you're fine with breaking the rules and have a good
> motivation to do so, that's fine by me. Rules are meant to be broken from
> time to time depending on the situation. But please don't pretend that
> modetest/xrandr is valid user-space to pass the rules.

I think it bends it pretty badly, because people running native Xorg are
slowly going away, and the modetest hack does not clear the bar for "is it
a joke/test/demo hack" for me.

I think some weston (or whatever compositor you like) config file support
to set a bunch of "really only way to configure is by hand" output
properties would clear the bar here for me. Because that is a feature I
already mentioned that xrandr _does_ have, and which modetest hackery very
much does not.
-Sima
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch