sfw@xpiinc.UUCP (Steven F. Williams) (10/21/88)
The Xlib Reference Manual (O'Reilly Assoctiates books) says that setting cursor = NONE in a call to XDefineCursor will cause no cursor to be displayed. The sample server causes the parent window cursor to be displayed, as if XUdefineCursor was called. The sample server Xlib implementation of XDefineCursor uses a ChangeWindowAttributes protocol request to set the cursor. The X protocol spec (R2) says "Changing the cursor of a root window to None restores the defalut cursor". The Xlib implementations seems to contradict the Xlib Reference Manual and the Protocol spec just is unclear. Steven Williams sfw@xpiinc.uu.net
swick@ATHENA.MIT.EDU (Ralph R. Swick) (10/22/88)
O'Reilly is wrong. XUndefineCursor is equivalent to XDefineCursor(cursor=None), both of which actually use the ChangeWindowAttributes request. Setting the cursor of a root window to None is a special case, since the root window doesn't have a parent.
aat@masscomp.UUCP (Andy Takats) (10/24/88)
In article <8810211915.AA17030@LYRE.MIT.EDU> swick@ATHENA.MIT.EDU (Ralph R. Swick) writes: >O'Reilly is wrong. XUndefineCursor is equivalent to ^^^^^^^^^^^^^^^^^^ >XDefineCursor(cursor=None), both of which actually use the >ChangeWindowAttributes request. Setting the.... I'm not particularly fond of the O'Reilly manuals, but I don't think it's fair to say something like this when the MIT C Library Reference manual states (R2) for the XDefineCursor call: "....You can pass None if no cursor is to be displayed." That sounds like more than just O'Reilly is wrong. Andy Takats ..!{uunet,harvard}!masscomp!aat