Brad Campbell wrote:

> Matthew Mastracci wrote:
>
>> It's a bit of a pain using the mouse over VNC since everything is done
>> using relative position packets.  It might be possible to emulate a
>> Synaptics PS/2 touchpad (which supports absolute positioning and is
>> supported under Windows and X) to make this a bit easier, but I'm not
>> certain how difficult that would be.  It might be easier to whip up a
>> special VNC client that supports mouse capturing.  :)
>
>
> Just out of interest, the mouse behaves just as badly running qemu 
> under an X vnc server session.
>
> Oh, and did I mention, nice work!

Thanks!  The mouse issue is certainly a problem - there's no tracking 
between the virtual VNC cursor and the real guest cursor. 

I hope that having a host/guest interface port (see the other recent 
thread) will make the whole thing a lot easier.  If we know where the 
guest's cursor is, we can figure out the relative movement required to 
get us there.
 
It's tough for me to post it as a diff -u against CVS because I'm so 
weak with autoconf-related changes.  I had to hack my own makefile just 
to get it to compile.  I'll look into what it would take to turn it into 
a full-fledged patch.

Matt.