Dear Thomas,

I have made a "clean" experiment by running fvwm for a user that
has absolutely no fvwm config in the homedir. The only thing comes
with the distribution and resides under /etc . The behaviour is the same.

> * If you have a window which is wholly on VGA-0 that moving the mouse into
>   monitor doesn't focus a window under the pointer?
> * That it only works if a window spans both monitors?

Yes. Put another way: focus follows mouse only for windows that
overlap with the "main" monitor LVDS (i.e., my laptop screen).

Let me list specific steps, may be it will be more clear.

1. I start X and fvwm on my laptop (no secondary monitor connected yet!).

2. I connect another monitor to the VGA output. Now I have two
   identical copies of the same picture.

3. I run xrandr to map the rightmost half of my virtual screen
   to the second monitor. Now my virtual screen is split between
   two monitors: if I move my mouse beyond the right edge of my laptop screen,
   it appears on the left edge of the second monitor.

4. Programs start to understand I have more physical space: for example,
   firefox loves to ask questions on the second monitor and xdvi
   chooses a larger width than usual.

5. Now everthing works OK *except* that any window that is contained
   fully in my second monitor (i.e., right half of the virtual space)
   cannot be focused by moving my mouse into it. Nevertheless, I can use
   buttons and scrollbars in those windows without any problem; however,
   focusing does not happen (in particular, I can't use keyboard in this
   window). 

   However, DEfocusing of the previously focused window DOES happen, so
   when my mouse touches those buggy windows, no window becomes focused. 

6. For clarity: the fvwm pager has cells corresponding to the
   size of a single 1024x768 monitor, and thus my virtual 2048x768
   screen spans two cells: two monitors occupy two cells of the pager; 
   the left cell is marked as active and
   the right cell is marked as inactive 
   (and I can move the mouse to this cell, yes). 

Best regards,
Edward

P.S.
> Have we worked together in the past?  :P
I assume every software developer or computer scientist is a colleague to every 
other software developer or computer scientist :)

On Wed, May 11, 2011 at 03:03:08PM +0100, Thomas Adam wrote:
> On Wed, May 11, 2011 at 10:01:14AM +0400, Edward A. Hirsch wrote:
> > Dear Colleagues,
> 
> Have we worked together in the past?  :P
> 
> > I have fvwm 2.5.30, X (Debian packages 1:7.5+8) with
> > virtual resolution 2048x768 and two 1024x768 monitors:
> > 
> >    /usr/bin/xrandr  --output LVDS --auto --rotate normal --pos 0x0 --output
> >    VGA-0 --auto --rotate normal --right-of LVDS
> > 
> > My problem is that moving my mouse to the second monitor (VGA-0) does
> > not bring focus to windows there UNLESS at least a tiny part of a window 
> > is located at the first monitor (LVDS). That is: it does not work for
> > windows that are fully contained in VGA-0.
> 
> Can you explain this again, please?  Draw a diagram if it helps.  But I've
> been using Xinerama for instance for yonks and not had this problem you're
> describing.
> 
> Are you saying that:
> 
> * If you have a window which is wholly on VGA-0 that moving the mouse into
>   monitor doesn't focus a window under the pointer?
> * That it only works if a window spans both monitors?
> 
> I'm dubious about this to be honest.
> 
> > The bug persists for at least a year; X version changed, kernel version
> > changed, everything changed, so I guess it is specific to fvwm.
> > 
> > I am attaching my ugly configuration files for fvwm and for X.
> 
> This is unacceptable.  Please get a *minimal* config which shows your
> problem.  I suspect it to be only about three lines if it's what I'm
> assuming the problem will be, but I am not about to second-guess. 
> 
> -- Thomas Adam

-- 
Edward A. Hirsch
Steklov Institute of Mathematics at St.Petersburg
27 Fontanka, St.Petersburg, 191023, Russia
Email: [email protected]
URL: http://logic.pdmi.ras.ru/~hirsch/

Reply via email to