[PATCH wayland-protocols] unstable: add xcursor-configuration

[Jonas Ådahl]

On Mon, Dec 10, 2018 at 02:10:30PM +0000, Simon Ser wrote:
> On Monday, December 10, 2018 2:51 PM, Jonas Ådahl <jadahl at gmail.com> wrote:
> > There is an alternative more generic solution to this problem, and all
> > other related to what XSettings did in the past. If we'd end up
> > introducing a protocol like this, we might end up with many tiny
> > protocols for things previously covered by XSettings. Cursor things,
> > font things, and what not. Adding a new protocol for each these things
> > would be silly.
> >
> > The generic alternative is designed with sandboxing in mind (but
> > of course doesn't require it in any way), and is part of
> > xdg-desktop-portal, the same place where the compositor agnostic
> > screen casting API lives (org.freedesktop.portal.ScreenCast, implemented
> > at least by GNOME and KDE).
> >
> > It's a D-Bus API called org.freedesktop.portal.Settings[0]. The only
> > thing a compositor would need to do is to provide a
> > org.freedesktop.impl.portal.Settings implementation using whatever
> > settings backend might be in place.
> 
> Hi,
> 
> Unfortunately I don't think D-Bus is a good solution here. Here are a few
> reasons why:
> 
> * Not all clients use D-Bus. Adding support for your API would mean adding a lot
>   of code and new dependencies to many clients.

This reason comes up everytime D-Bus is mentioned, and I think it's
somewhat counter productive; I don't think this is a problem clients
should try to avoid so hard. D-Bus is more or less the universal IPC on
Linux, is used quite extensively to allow sandboxed client to
communicate with the outside world, and is almost guaranteed to be
available everywhere.

> * The idea would be to add support to libwayland-cursor so that many users can
>   benefit from the protocol. I don't think firing up a D-Bus connection in
>   libwayland-cursor as a viable solution.

Well, we could still have applications feed settings into it. It's not
tha complex to do that, and the addition to libwayland-cursor that is
arguably more complicated could cover the server side cursor buffer
loading that was talked about before.

> * Multi-seat support would be kind of alien.

Can at least be investigated how multi seat should be achieved before
being dismissed as being alien like. I can open issues for that if you'd
there is interest.


Jonas

> 
> What do you think?
> 
> Thanks,
> 
> Simon
> 
> >
> > Jonas
> >
> >
> > [0] https://github.com/flatpak/xdg-desktop-portal/blob/master/data/org.freedesktop.portal.Settings.xml