🚀 go-pugleaf

In article <2158@uoregon.uoregon.edu> jqj@drizzle.UUCP (JQ Johnson) writes:
>Not all pointing devices are mice.  With "absolute" pointing devices
>such as tablets, light pens, or touch screens, it is extremely
>undesirable to break the 1-1 mapping between absolute physical pointer
>position and logical pointer position.  Although I'm not averse to
>semiautomatic mouse cursor movement (e.g. in response to a mouse click
>or menu selection), I worry that such a user interface design makes the
>system far less flexible.

	As Jim Gettys pointed out, it is RARELY appropriate to
	warp the cursor, but there are a few situations that
	demand it.  Those I've seen tend to involve maintaining
	a semantic context when associated graphics change.

	One example is rotating or reflecting symbols in a schematic
	editor.  Semantic context for editing operations tends to
	be a hierarchy of objects marked by the mouse, such as...

		Drawing set
		  Specific drawing
		    Area within drawing
		      Symbol (e.g., image of an IC chip)
		        Component of symbol (e.g., label of a pin)

	If the cursor marks a pin near a corner of a rectangular
	chip symbol, and the user commands a 90-degree rotation
	by hitting a function key, leaving the cursor alone drops
	the bottom two objects of the semantic context.  Early
	experiments while adding rotation in FutureNet's DASH-4
	editors showed that this was EXTREMELY annoying -- the
	user had to chase the object he was editing.

	The solution that worked well was to warp the cursor to the
	graphic position that maintains the same low level context.
	I think the key notion is to do whatever preserves the context --
	usually this means NOT warping the cursor.


---------------------
Paul Raveling
Raveling@vaxa.isi.edu