also should now work if the first element of a modifier
keymapping is 0. that won't normally be the case but xmodmap
was doing weird things on grace. if the first element is 0
it'll try the second element. if that's also zero then that
modifier will be ignored.
now tries to map ISO_Left_Tab without shift first then falls back
to Tab (note that if ISO_Left_Tab can be mapped but requires a
modifier then the modifier will be added). also changed attempt
to map ISO_Left_Tab as a backup to Tab to request the shift
modifier whether or not the primary screen requested it.
a screen. this allows the secondary screen to set it's
modifier state to match the primary screen's state. this is
not strictly necessary since each keystroke should adjust the
modifier state as needed to get the right result.
now but did notice one problem: when powerbook is primary and
num-lock is on the keypad works fine until shift is pressed
(and released); after that the keypad only works while the
shift key is down.
clipboard owner support (MS windows done, X windows partial)
added key transfer on ms windows
mutex fixes in CClient (had race conditions)
faster debug output in ms windows
changed temporary screen name to "secondary"
network fixes on ms windows (poll returned wrong result)
fixed transparent cursor on ms windows
