> I should mention that the current Squeak idiom (idiocy!) of using the 
> select button to open a menu when on the main display is really stupid 
> and can only be explained by the initial connection to Mac single 
> button mice. Which is another dim idea.

Well it all depends from where you are coming from.

On the Amiga, there was only two mouse buttons. The left-one was for 
all purposes while the right-one was used to bring the menu in the top 
menu bar, which acted as a status bar and a way to drag down the screen 
to see what was going on behind. In DeluxePaint, the right mouse button 
was used to draw with a secondary color and the user had to move the 
mouse to top of the screen to open the menu bar. It was practical on 
Lo-res screen modes (320x256) as the user didn't have to move the mouse 
that far but became tedious when the first graphic cards were available 
with Hi-res (800x600 and over) screen modes.

One nice thing was that the Amiga menus permitted to click on menu 
items with the left-mouse button, even in different groups. So the user 
could open the menu bar with the right button, and click with the left 
button on "Find > Next", "Edit > Copy", "Project > Close", "Project > 
New", "Edit > Paste", "Project > Print" in one go and these items were 
executed in that order. (It seems that RISC OS borrowed that feature).

So things were quite simple and the two mouse buttons had two very 
different behaviours. With the special Amiga keys, it was even possible 
to bring the menu while using only the left mouse button.

What I have learned during this period with end-less hours passed in 
front of DeluxePaint is that computers feature 100+ buttons. But today 
operating systems/applications make poor use of them (don't even get me 
started about shortcut keys...). I don't know when things started to go 
wrong but suddenly the keyboard was something solely used for inputing 
text, not for acting on objects, at least on the Windows side of 
computing.

Sometimes, a one-button mouse is working amazingly well when making 
good use of the keyboard. For example, if I want to wire two shapes in 
OmniGraffle, I just keep down the L key and draw the connection line.

It even makes more sense to have a one-button mouse with Squeak: morphs 
have different menus, depending on the use of the Control, Alt or 
Command keys. Between delegating some of the menus to the mouse and 
delegating all menus to the keyboard, I prefer a single-button mouse 
with the last solution because it puts emphasis on actions and not on 
contexts.

To illustrate furthermore this issue, take the polygon morph. To add 
new points, the user clicks on the green triangles, which is an action, 
but to remove a point, she has to move a point to a neighbour point, 
which is context. The first one is obvious, the second requires to 
browse class comments to figure out... Sure a context menu could do the 
trick, but then the user is confronted to many options, not actions 
(that's why the menus have to be revisited in Squeak as they appear to 
have an infinite number of options).

Dan talked about pen computer and it reminds me how poorly supported 
are these devices in current systems. Wacom pens come with four buttons 
(the tip, two buttons on the side and the eraser) making it an 
incredible device for inputing stuff with gestures. The side buttons 
could be used to apply immediate and delayed actions (and not for 
opening context sensitive menus) and possibility are endless when using 
pressure sensitivity. The current PDA and other pen based devices are 
frustrating to use because they don't make proper use of a pen. The 
Newton was interesting because it created a complete metaphor around 
the pen input by using conventions found in text revision made on 
paper.

Nintendo makes a bold move into this area with the DS and some of the 
games are amazing. Pen and voice form a pretty combination.