UI Patterns and Techniques

One-off Mode

Use when:	You're building a graphical editor, and there are certain operations -- such as creating objects -- that users don't normally repeat or iterate over. Usually, the user will perform the operation once, then immediately want to do something else, like manipulating the object just created.
Why:	Users will find it annoying to switch into a mode, do one little thing, then explicitly switch out of that mode again -- this often involves clicking on small "hit targets" in palette windows far away from the working canvas. (Here, mode is defined as an application-wide state that temporarily changes the behavior of the mouse pointer.) Too much "clickiness" in an interface is a known irritant! Instead, the interface should do what makes the user's job easier, even if it's not conceptually tidy or easy to program: when the user enters the mode in question, stay in it for only one operation, then automatically leave the mode.
How:	The hardest part is deciding which operations ought to behave like this. Object creation typically does; zooming, lassoing, paint strokes, etc. typically don't. Find out what graphical editors your users tend to use most, and see what they do. An example might make all this clearer. Consider a drawing tool in which mouse-clicking on the canvas normally selects objects underneath the mouse pointer: User clicks the "Create rectangle" button on a palette. The UI goes into a special creation mode, indicated by a rectangle cursor -- now, clicking on the canvas will mean object placement, not selection. User clicks once on the canvas, to place the upper-left corner. User clicks again on the canvas, to place the lower-right corner. The UI, having counted two clicks and created the rectangle, leaves its rectangle-creation mode and goes back to the default mode, in which clicking means selection. User is now free to select objects, move them, resize them, etc. without having to go back to the palette to change mode.