[RFC libinput] Import buttonset interface

Peter Hutterer peter.hutterer at who-t.net
Tue Apr 26 05:20:48 UTC 2016


On Fri, Apr 22, 2016 at 10:51:40PM +0530, PrasannaKumar Muralidharan wrote:
> > fwiw, it's very frowned upon to take someone else's patch, rebase it and
> > then send it on as your own when it has very little changes. Please keep
> > original authorship while it makes sense and add the author as co-author
> > where substantial changes have been made.
> 
> I definitely did not intend to claim this as my patch. I will set the
> correct author when I send the patch again with modifications.
> 
> > tbh, I'm not quite sure what to review here though, without a backend
> > implementation this is quite theoretical. I'm assuming you just want a
> > quick review to make sure nothing obvious got lost in the rebase?
> > If so, yes, looks ok, thanks.
> 
> Yes. This is one of things I wanted to understand.
> 
> > There are a couple of questions that arose from the
> > wacom tablet button switch, specifically:
> > * what use do the linux/input.h button codes have in the buttonset
> >   interface? We will run similar issues with the evdev button code ranges as
> >   we did with the pad, where the button code is meaningless.
> > * the various axes are nice to have, but without a user I'm hesitant to
> >   merge them - we'll need to know how they could be used first
> 
> I am dealing is similar to rotary encoder, buttonset interface is the
> required for it. I am not aware of the issue mentioned above. I would
> prefer to have some guidance on what should be changed or removed. The
> main objective of the patch sent is exactly this - I wanted to know
> what I have to do to get the buttonset interface merged.

simply said, I don't know. It's one of the things where we have to write
most of the interface, see if it makes sense and then remove the bits that
we don't actually need. your main target right now is a rotary encoder
but I do encourage you to think of how to deal with other devices that are
generic devices (tv remotes, maybe a 3d mouse, dj mixers, etc.) and
incorporate them. an interface just suitable for a rotary encoder won't get
merged.

Cheers,
   Peter
 
> > * ring and strip are absolute, they are wacom-only axes and until we get a
> >   device that needs either, we should leave them out.
> 
> Will remove them.
> 
> > * the seat button count should be removed, it doesn't make sense for a
> >   buttonset device
> > * get_(x|y)* does not need an axis argument, I struggle to think of a device
> >   that has two x or y axes on the same event node.
> >
> > Other than that I still think that the general approach with numbered axes
> > and axis types makes sense.
> 
> Okay. Will change accordingly.
> 
> >> This change is compile tested.
> >
> > That's not a confidence-inspiring statement... :)
> 
> Agreed. I will follow up with dial input support and will have a
> proper working version by then.
> 
> Thanks,
> PrasannaKumar


More information about the wayland-devel mailing list