I've been doing more debugging. First, I couldn't find any jumps between the two screens that _won't_ cause the mouse "detachment". For example (assuming screen width of 1680):
After messing with that, I was curious about where the cursor ostensibly was during "detachment". So, I dug around for applications to detect the cursor position and found a QT3 script on some forum, which I've attached. I don't know if this will work for everybody on Gnome, but it works under KDE4 and I assume also under KDE3. (See readme.txt in the zip).
The upshot: the cursor is registered as being stuck at 0,0 during "detachment"; despite me seeing the cursor moving on screen, the coordinates stay registered as being at 0,0.
I've been doing more debugging. First, I couldn't find any jumps between the two screens that _won't_ cause the mouse "detachment". For example (assuming screen width of 1680):
------------
#!/bin/bash
xte "mousemove 1479 5"
xte "mousemove 1900 5"
------------
------------
#!/bin/bash
xte "mousemove 1200 5"
xte "mousemove 2100 5"
------------
Also, this (apparently) isn't time related. Making sure not to move your mouse, try (again, adjusted for your resolution):
------------
#!/bin/bash
xte "mousemove 1200 5"
sleep 1
xte "mousemove 2100 5"
-------------
After messing with that, I was curious about where the cursor ostensibly was during "detachment". So, I dug around for applications to detect the cursor position and found a QT3 script on some forum, which I've attached. I don't know if this will work for everybody on Gnome, but it works under KDE4 and I assume also under KDE3. (See readme.txt in the zip).
The upshot: the cursor is registered as being stuck at 0,0 during "detachment"; despite me seeing the cursor moving on screen, the coordinates stay registered as being at 0,0.