I am experimenting with a graphics pen and tablet and it got me thinking about the difference between it and my mouse. One huge difference is that the tablet's working area covers the whole screen, so that you can tap on an absolute position. My mouse always outputs relative x/y from its last position. This is obviously a huge benefit because it does not restrict the mouse to one part of your desk. I am wondering if this was how the original mice/trackballs/whatevers did, or were they absolute position?
The first mouse tracked relative motion along two axes, and as far as I know all standalone mice produced since have followed suit. It would be difficult to build and use a mouse relying on absolute positioning: it would have to track its movement very accurately, with no slippage, or else allow for regular recalibration; as you mention, it would only be usable in a specific area; and it wouldn’t support varying sensitivity (i.e. slow movement being translated at higher resolutions than fast movement).
Even early optical mice, which used specific mouse mats (as used for example on Sun workstations), didn’t track their position on the mat itself, only their relative movement.
There are mouse-like devices which produce absolute coordinates: pucks on graphical digitisers. These are used on large tablets, and are not practical replacements for mice in most cases.
It was by no means a mass market device, but Hayward and Ramstein's Pantograph (1993) encoded linkage positions as absolute coordinates. It also provided force feedback, and could ‘drive’ itself based on screen content
Some of the HP Omnibook series of laptops and sub-notebooks from the mid 1990s had a curious pop-out “mouse on a stick”:
While hardly part of the original mouse timeline dating back to the 1960s, this HP mouse used encoders built into the computer body. The encoders — as shown in this Omnibook repair video from 10' 40" on — appear to track the extension and angle of the mouse stick. In order to produce PS/2 mouse compatible movement counter signals, the (∆r, ∆θ) from the internal encoders would have to be temporarily converted to an absolute (X, Y) position from which (∆x, ∆y) signals were derived.
From memory, the Omnibook mouse would continue to produce (∆x, ∆y) signals if it hit the relevant end stops, so there were perhaps some additional limit switches in the mouse hardware. It was a fairly terrible mouse, and completely unusable by left-handed people.
I am experimenting with a graphics pen and tablet and it got me thinking about the difference between it and my mouse. One huge difference is that the tablet's working area covers the whole screen,
No, that's scaling of your software. The tablet has its own coordinate set, which gets adjusted to your document and/or screen. Usually by the drawing application using it in absolute mode. For screen the driver may be configured to do it, or offer it as relative to fit usual mouse handling.
so that you can tap on an absolute position.
Jo. After all, with a tablet, the detection is not done by the moving device, but the fixed surface.
It's imperative for drawing tablets to work absolute, as for one there's no detected movement when the pen is (way) up. Only absolute detection will work to catch it when going down again. But more importantly, with a mouse a user usually 'homes in' to a target - meaning the movement is controlled via an optic feedback on the screen - while with a tablet the user expects to hit the point like with a pen on paper. No matter how much movement has happened in between.
My mouse always outputs relative x/y from its last position.
In general, without a fixed reference can only detect relative movement. Even a 'simple' device like an odometer only adds up data delivered as relative. And like everything working relative, it adds up errors. Thus relative recording isn't a great idea with tablets anyway, as the user assumes the device to read the exact spot he targets.
This is obviously a huge benefit because it does not restrict the mouse to one part of your desk.
The same can be done for tablets by scaling and panning. Take touch pads for example, here a cursor can be moved several times in one direction by repeated moves. Despite the fact that it delivers absolute coordinates. Similar is done for drawing tablets.
I am wondering if this was how the original mice/trackballs/whatevers did, or were they absolute position?
Always relative. Anything else would be incredibly complex and end up in a system resembling a drawing tablet. There have been combinations that looked like a mouse, but had to be used on a tablet (or with a fitting tablet like sensor setup), but they were incredibly expensive while combining disadvantages of mice and tablets.
For a generic, not very exact input device, relative is the way to go. Even more when the task is about relative positioning anyway.
The Commodore 1351 mouse, created as an afterthought for the C64 and C128 and (ab-)using those systems' analog paddle inputs to transfer mouse position data (since no mouse support was planned when those computers were designed), maintains an internal sort-of-absolute position on a wrapping 64x64 pixel grid which it then provides to the computer. While mouse-supporting software running on those computers will have to convert those 64x64 grid positions to really absolute values by correcting for the wrapping and optionally applying some sort of acceleration algorithm, this type of mouse does not transfer relative position in terms of signed x/y displacement values.
As Stephan Kitts mentions, the mouse puts out relative coordinates. In reality the mouse sends its x/y movement (not coordinates) in mickeys (yes that's the name of the unit) to the PC in the form of interupts. The software can intercept these interupts and process them to do whatever: move a cursor, scroll, move an item in a game or more. Not necessarily anything to do with position on a screen. That is only a (now the most common) usage of a mouse.
When I was an older kid or early teen, my uncle took me to his office and let me play around on his computer. It was a dedicated CAD workstation, with a mouse that did actually encode absolute position on its pad (or whatever you would call it). The tail of the mouse was attached in the opposite position from modern mice, coming out under your wrist as you were holding it and plugging in to the "mouse pad" at the bottom, while a second cable connected the pad to the actual computer. At the top of the mouse, where the tail would normally connect, there was instead a crosshair that could be used to select icons that were actually printed on the "mouse pad" itself.
I remember getting frustrated because none of its four buttons were labelled, so you just had to remember which button did something with the mouse cursor on the screen, and which activated the function that the crosshair happened to be pointing at.
This is all from memory from quite a while ago, so I don't really have any more information about it, like a manufacturer or model or anything.