[Spice-devel] [PATCH spice-gtk 2/2] gtk-session: always sync modifiers for client events

[Jonathon Jongsma]

----- Original Message -----
> From: "Marc-André Lureau" <mlureau at redhat.com>
> To: "Jonathon Jongsma" <jjongsma at redhat.com>
> Cc: spice-devel at lists.freedesktop.org
> Sent: Friday, April 4, 2014 12:16:26 PM
> Subject: Re: [Spice-devel] [PATCH spice-gtk 2/2] gtk-session: always sync modifiers for client events
> 
> 
> 
> ----- Original Message -----
> >         - if the guest display widget is focused and the user presses
> >         Caps lock, we will send a keypress message (which will cause the
> >         capslock modifier to toggle on the guest) and then we will
> >         immediately send a modifers message. So we'll send duplicate
> >         messages for the same event. In practice I hadn't noticed any
> >         problems because of this, but I wonder if this is partly what
> >         caused the issue you observed?
> 
> Yeah, I noticed that too, but I don't think it's a big problem sending twice
> the current state.

Probably not. I can't currently think of a race condition that would cause problems here.

> 
> 
> >         - if the capslock on the client is mapped to something else
> >         (e.g. 'control'), and the capslock key is pressed while the
> >         display widget is focused, we will send a keypress event to the
> >         guest, which will cause the guest's capslock state to change.
> 
> 
> >         However, the client's capslock state will not change, so when we
> >         receive the 'inputs-modifiers' message from the guest, we will
> >         send down the old value again.  On the other hand, this is a
> >         corner case because it requires a non-standard keyboard layout,
> >         and the alternative (not overriding the guest modifiers state)
> >         results in an out-of-sync modifiers state between the guest and
> >         client. So I'm not sure what the right answer is here.
> > 
> > thoughts?
> 
> That sounds still ok to me, we want the guest to be sync with the client
> state. Doing something else will be even more annoying, wouldn't it?
> 

Yes, probably, but I thought I'd mention it for others to consider.