Now that basic xrandr works with my configuration, I have found problems with the mouse cursor.  I have a monitor configuration where my laptop's LVDS is oriented normally but my external VGA-0 monitor is left-rotated into portrait orientation as follows:

    +-----+ +---+
    |     | |   |
    +-----+ |   |
            |   |
            +---+

Here is the pertinent output from xrandr:
    $ xrandr -q
    Screen 0: minimum 320 x 200, current 3120 x 1600, maximum 3520 x 1920
    VGA-0 connected 1200x1600+1920+0 left (normal left inverted right) 408mm x 306mm
       1600x1200      60.0*+   59.9  
    ...
    LVDS connected 1920x1200+0+0 (normal left inverted right) 0mm x 0mm
       1920x1200      60.0*+
    ...

There are two related issues.  The mouse cursor is left-rotated and appears correctly on the portrait monitor.  But it is turned on its side when moved to the landscape laptop display.  This is not _too_ disturbing, but could be better.

It gets even stranger, if you set the position of the VGA-0 to overlap the region displayed by the LVDS and move the mouse into the shared area, it will be oriented correctly on the VGA-0 and incorrectly on the LVDS.  Now I under stand from a video-memory perspective they are both being drawn in the same "direction".

The second more important bug is that on the landscape display the cursor's apparent and actual hot point differ.  Visually the mouse points to what seems to be the lower-left corner of its sprite space, but the actual hot point as detected by hovering over a button or clicking on something is the upper-left corner.  This is further evidenced by the fact that you cannot move the pointer clear to the top of the screen.