Re: [rudalics@gmx.at: Re: [simon.marshall@misys.com: mouse-autoselect-window needs a de lay]]

[martin rudalics <rudalics@gmx.at>]

 > I think a delay is the wrong thing to do here since it makes stuff
 > unpredictable.

Delays are the standard approach window managers adopted for focus
follows mouse and window auto-raising.

 > The right fix, in my opinion, would be lazy focus
 > change: only when a keyboard or mouse event occurs in the new window,
 > is the focus changed.  That has the disadvantage that the user may be
 > surprised by the change since the old window appears to have focus
 > before typing a key.

That could be added.  But it would violate the principle of least
surprise.  In my experience, the `mode-line-inactive' face and the
`cursor-in-non-selected-windows' option handle feedback pretty well.

 > In order to mitigate the surprise, it might be reasonable to visibly
 > unfocus the old window (by the different highlighting of the mode line
 > and the different cursor type), but not refocus a different window
 > before an event occurs.

That would be disconcerting, IMHO.

 > A more radical approach would be to move toolbar and menubar just
 > above the currently selected window.  This would also require less
 > mouse movement, but would likely earn us an award for the most weird
 > user interface look ever.

I recall someone proposing the use of popup menus instead.  There's also
the possibility to access the menubar via F10 or the ALT key.  However,
clients of `mouse-autoselect-window' are probably inclined to access the
menubar with the mouse too.

 > I don't think it is reasonable to expect to sort this out before the
 > release.  I think that there are several viable possibilities, and
 > we'd need the feedback of testers trying each of those out for several
 > weeks before it would be viable to decide on a sensible strategy.

Simon and me have been testing this intensively for an entire month.
Emacs developers apparently either don't use `mouse-autoselect-window'
or don't use the menubar.  Otherwise they would have complained earlier.
Hence, for testing this any further we need a broader audience.

 > I don't think that the delay stuff is a good strategy: it brakes and
 > confuses the user when he indeed wants to change focus, and it causes
 > hectic when he doesn't.

That's what I thought before I started implementing this.  Currently, I
use a value of -0.5, that is, wait for about a second after I started
moving the mouse and select the other window iff the mouse doesn't move
any more.  So far I didn't notice any confusion with these settings.