[PATCH wayland] protocol: add repeat_info event to wl_keyboard

[Jonny Lamb]

(finally got around to looking at this again)

On Mon, 2 Jun 2014 10:22:40 +0300 Pekka Paalanen wrote:
> this looks good to me, but I'd like to see text about when this event
> is sent. I assume it is sent once as soon as a wl_keyboard object has
> been created, and this is also required/guaranteed. IOW, if a client
> creates a wl_keyboard and then does a roundtrip, it is guaranteed to
> have received the event at least once. Right?

Right! I've addressed all these issues.

> Or should sending it be optional? Should we define any defaults in case
> no event has been received? I'd assume not.

I agree not. I defined some defaults in weston but that's weston.

> Can it be sent later, too? (Likely yes.)

I added a note to say it can.

> Was this ever discussed before this email thread? I'd like to see at
> least one more ack (Daniel already ack'd) before I'm comfortable pushing
> this to wayland master, as this is part of the core protocol.

No it wasn't, although Daniel was the one who told me to implement it in
the first place.

On Mon, 2014-06-02 at 10:36 +0300, Pekka Paalanen wrote:
> Would we want to specify "no repeat" values?

I used Giulio Camuffo's suggestion of using zero for no repeat.

> Also, while it is just an event and error codes would not make sense in
> the protocol, we should probably still mention, that negative values are
> illegal - what to do with zero?

I added a note about negative values.

> Re-reading it, this seems funnily said:
> > > +           summary="time in milliseconds between keys repeating"/>
> I think it wants to say: "delay in milliseconds since key-down until
> repeating starts" or something. Not "between keys" since that sounds
> like repeat rate. :-)

Yes, good point, fixed.

-- 
Jonny Lamb