FC5: first impressions

Jon Nettleton jon.nettleton at gmail.com
Sun Mar 26 15:21:08 UTC 2006


On Sun, 2006-03-26 at 16:55 +0200, Nicolas Mailhot wrote:
> Le dimanche 26 mars 2006 à 09:43 -0500, Dimi Paun a écrit :
> > On Sun, 2006-03-26 at 14:20 +0000, Bill Crawford wrote:
> > >  True enough, but the focus stealing prevention is something that's
> > > been needed for a long time. 
> > 
> > Not for me, it hasn't. I know for sure I don't want it.
> > The typing-the-root-password-in-the-newly-opened-im-client
> > scenario doesn't keep me up at night. In fact, it's a 
> > non-issue for me.
> 
> Has anyone ever tried something like "if current window has been
> receiving keypresses in the last x ms do not steal focus from it?"
> 
> The main problem seems to be the race between the system showing up a
> new window and the user realising the new window has stealed the focus,
> so why not just refrain from taking the focus if something is being done
> in the current window ?
> 
I brought up using that behavior some time after Test 3 was released.
It was met with the typical, "Having inconsistent window focus will just
confuse the users more".  I know that when I launch an application I
automatically expect it to show up in the background.  

I think the check for keypresses/mouse clinks between application launch
and it being drawn on screen makes a much more logical interaction with
the desktop.  If you ask your secretary to get you a file, and then you
start working on something else, she doesn't just come over and drop it
right in front of you.  If you ask your secretary to get you a file, and
then you sit there twiddling your fingers, she will probably hand it
right to you so you can start working on it.

Metacity already has an algorithm that somewhat tries to implement this
with the intervening_user_event_occurred function.  I don't think that
this is working quite right.  I spent some initial time working on it,
but then got side tracked.  I would be more than happy to pick this back
up if there is any interest in people testing this behavior and seeing
if it is natural, or too confusing.

Jon




More information about the devel mailing list