Re: address@hidden: Re: address@hidden: mouse-autoselect-window needs a de lay]]

[Kim F. Storm]

martin rudalics <address@hidden> writes:

>> To me it makes more sense if this "move and stop" is the default method
>> (ie. selected with a positive delay value).  The negative value could
>> be for experts which want just the move delay.
>
> The absolute value of the delay is used to set up a timer whenever the
> mouse crosses a window border.  Usually though, the mouse pointer
> already appears at a completely different position at the moment I
> install the timer.  Unfortunately, I can't access that position easily.

Don't (mouse-position) or (mouse-pixel-position) give the current position
when you install the timer?


> When the timer triggers I do (provided other conditions apply as well)
>
> - for a positive delay select the window,

Do we really need that option?

>
> - for a negative delay check whether mouse-position is the same as the
>   last time I checked.

That's what I think should be the default ... don't select the window until
the mouse movement stops.  But detect that quickly!

>
> Now, the very first "last time I checked" occurs at the window border as
> I explained above.  Hence the first time the timer fires, the recorded
> mouse-positions are usually _not_ equal and I have to restart the timer
> until they are.  In practice this means that if I set the delay to -1
> and move to the center of another window in a single stroke, the window
> is selected after 2 seconds which is more than 1 seconds after the mouse
> stopped moving.  Since this is slightly counterintuitive I wouldn't want
> to make that the default.

Can't you just use a (fixed) 0.1 second timer the first time ?

>> But why not separate the two timeout settings, e.g. make it a cons with
>> the "move delay" in the car and "stopped delay" in the cdr.  That's
>> easy to set via customize.
>
> Currently, there is only one delay.  Do you think I should use two
> different delays?

No.  One timer "mouse-movement-stopped-timeout" would be ok.

-- 
Kim F. Storm <address@hidden> http://www.cua.dk