It should be possible to enable focus-follows-mouse with no delay between the mouse entering a window and that window being focussed. The UI currently imposes a minimum 5ms delay.
Created attachment 2266
Allow setting 0ms focus delay
This patch just changes the range of the focus delay slider in the UI so that its minimum is 0ms, and ensures that the value selected here isn't clamped to a minimum of 5ms.
(In reply to comment #1)
> Created an attachment (id=2266) [details]
> Allow setting 0ms focus delay
> This patch just changes the range of the focus delay slider in the UI so that
> its minimum is 0ms, and ensures that the value selected here isn't clamped to a
> minimum of 5ms.
No, 5 is fine, 0 would not bring anything perceivable.
I would very much like to have no focus delay. How about leaving the minimum value for the delay at 5ms, but having an additional option to disable the delay altogether?
(In reply to comment #3)
> I would very much like to have no focus delay. How about leaving the minimum
> value for the delay at 5ms, but having an additional option to disable the
> delay altogether?
No, I am against adding yet another option for that. Can you elaborate on why you need 0, how do you see/feel the difference between 0ms and 5ms? Any real life use case to backup that need?
(In reply to comment #4)
> No, I am against adding yet another option for that. Can you elaborate on why
> you need 0, how do you see/feel the difference between 0ms and 5ms? Any real
> life use case to backup that need?
I find it a little disconcerting having the small delay between moving the mouse over a window and it becoming focussed; it means I need to consciously wait for the window to become focussed before typing. Often I like to flick the mouse between xterms and type things. While it may not be a practical problem, it is a little jarring.
(I think the actual delay turns out to be greater than 5ms, despite the slider being set down to its minimum. Maybe there's some intrinsic minimum that I'm reaching? 5ms should be nearly imperceptible, I suppose.)
I'm happy to provide a patch if necessary.
I agree it should be possible to make it have no delay. Xfce is supposed to be fast so why waste 5ms? Considering how much effort some people put trying to make their PCs faster (overclocking etc) then wasting 5ms for no good reason seems rediculous.
The minimum time it takes for a human being to perceive and process a visual stimuli is in the range of 15 milliseconds at least. A 5ms delay should not be noticeable.
As explained in the discussions at http://stackoverflow.com/questions/7882713/can-a-human-eye-perceive-a-10-milliseconds-latency-in-image-load-time a delay may at worse make a visual change appear a frame later, i.e. 17ms later for 60 FPS than the original action causing it.
Now onto what's actually happening with the focus on mouse hover. It takes an average of 300ms, when fully focused, to press a keyboard key after a very visible visual stimuli. That's if you're really trying to go fast. So, missing a frame is no big deal at all because it takes you 15 times as long to start typing anyway. The performance gain from removing the 5ms delay would be unsignificant, even if you were to perceive that the UI is more responsive.
I could do tests to check if people actually find Xfce snappier without the 5ms delay, in which case it might be worth removing it. But it won't actually make your use of Xfce any faster.
(see http://cognitivefun.net/stat/1 for proof of stimuli processing delays)
As far as I am concerned, this is wontfix.
WONTFIX as the inconvenience of the 5ms delay does not warrant a change to the codebase.
I'm no longer an xfce user, so WONTFIX is fine by me.