This page documents projects and ideas to use pie menus. There are actually already some early versions of projects using it to test and work on. The original idea (by Thilo Pfennig) was:
1. Wouldn't it be much better if most menus where made as circles or ellipses?
Contents
Simpler and slicker (more like small devices like MP3 players, iPods etc.)
Nicer for mobile devices which do not have a mouse or keyboard.
(Why? At the end of the day, pie menus and dropdown menus are both lists with the same two control primitives: 'next' and 'previous'? -AndersFeder 2024-10-23 10:59:08)
dropdown menus often come to the point where menu items dont all fit on one screen and scrolling is sometimes impossible or not easy. pie menus allow you to organize differently (hide entries,etc.) -- ThiloPfennig 2008-07-06 23:42:52
For people with disabilities (including people in a hurry), the pointer movements can be shorter than with long standard menus.
A circle can contain many more entries (even on many layers).
(I would like to contest this claim: an icon is an icon, whether its rendered on a circle or a vertical line? -AndersFeder 2024-10-23 10:59:08)
Simple as that: dropdown always has as many possible entries as the screen height. you can add submenus but that makes handling menus MUCH more complicated. A pie menu can behave differently. Like you dint have to list all entries at once. -- ThiloPfennig 2008-07-06 23:46:45
This hiding and not-showing can't be done with drop-down menus? -AndersFeder 2024-10-23 10:59:08
1.1. Ideas
Rather design the menu options than just list them.
Have input devices that fit to the menus?
1.2. Projects
1.2.1. circular-application-menu
added:2008-03-19
This is a circular-application-menu prototype mirroring the same structure as the existing application menu for the GNOME desktop. Currently it's just a command line application.
1.2.2. Chromadesk
LiZe and KevinKubasik have worked on Chromadesk whcih uses circular menu to launch apps. It's a kind of Main menu button in the middle of the screen, other menus (for example network, multimedia, etc) appear in circle when the main button is clicked, and apps (for example net browser, text editor) appear when the menu button is clicked. It uses Tango icons with eye-candy features (transparency, lightening, etc). Works with pygtk>=2.10, rsvg, and a composite desktop (I works on XGL, but it also seems to work with a simple X with composite option).
Play Music Video. The video shows only 3-items menus, but icons are automatically created and moved in order to draw a circle around the parent icon. There is a lot of problems (no text yet, etc), but it can give you ideas...
Code is awful and video is slooowwww.-- LiZe
1.2.3. Fittsmenu
A project by KarlLattimer:
I'm actively looking for feature requests and Fittsmenu is being developed to fit into MedicalStudio too, the counter rotation can be disabled at will and I'm also working on a scaling algorithm which will make the menu egg shaped 
- KarlLattimer
- One interesting question is how one could implement submenus:
- make an outside ring menu
- replace pie menu
- make a pie menu "above" the old (another idea is how this can look combined with 3D.)
- open a submenu not through clicking but through keeping position ?
- an important problem is that if the user makes unplanned quick movements that he should not
- Another thought: How can those menus handled the easiest way: A lot of entries in one ring or only very few? I think the nicest thing is if the elements to choose are from the same level.
- Maybe one could start use that testwise as a replacement to the "right click on background" action?
And/Or: Make it an application that is configurable. So that you may start with only oen entry "Configure" - and then you can add an action. Some actions like changing the background could be predefined, other could maybe be defined by the user. Such could be "open a terminal" or "open email program", "open web browser" ,... so a bit like a GNOME starter. And an entry also could be a sub menu. I would think that to configure it should have a simple wizard. Also every entry should/could be removed or changed with a right-click + edit/delete. the general idea could be to present the user the options that make sense in that very moment. I think it would be maybe nice if one could play with the timings and behaviour to find out what behaviour of the pie menu is feeling best (how long it keeps position, highlights, etc.) -- ThiloPfennig 2007-09-02 22:44:52
1.3. Pie Menu usages or links
There is an interesting study of this system(aka marking menus) by the University of Toronto. Useful stuff for interested developers to think about --Decoherence
There was a Bugzilla entry, hoping to raise interest in this idea, located at 355718--D. Le Sage, Hobart
SecondLife also uses pie menus also (Screenshot)
The game "Neverwinter Nights" did implement a radial menu. It has been considered a bad move as Neverwinter Nights 2" came back with an old style menu. As a user, I'm not sure the radial menu is a good move, because it's harder for the eye to scroll all the available choices. Submenus are also a problem, because it then requires navigation between all the menus. I'd like to have some feedback from others NWN players. Radial menus may also be a bit uncomfortable for new users. However, we can't really know until we start using it, and AFAIK neither Windows nor MacOX ships them at the moment, so this may be some added value. -- LuisMenina 2006-12-18 12:01:54
Radial context menu for Firefox especially noting the way it is labelled. Personally I think that it is harder to use because people are used to scanning lists for things rather than looking in a circle
Pie menus has been very successfully used in some games. They're very effective for not too huge actions lists (dived in sub pie menus) and provide the same interface for casual users and power users, making the learning curve of the later easier. There's a good paper about this topic by Jonathan H. K. Mak. It's focused on using only keyboard but it may be interesting anyway.
- I would recommend against extrapolating too much about desktop usability from games. Games usually have a limited set of verbs ("attack", "talk", "pickup" etc.), whose icons the player memorize and then utilize game session upon game session. Thus a steep learning curve is a good choice, because there is only a limited set of verbs (pie menu icons) to learn and once that is done, nothing new has to be learned. In contrast, the desktop has a large set of verbs that change with the application and the task to be done. Thus a steep learning curve would be a bad choice as it would just keep going up and up, as the set of verbs is constantly changing - there is always a new icon or ordering of icons to learn. In a game you play the same 'level' or 'dungeon', using the same verbs, over and over again, but on the desktop you are much less likely to play the exact same 'level' or workflow scenario twice, as that would be ineffective.
Aren't your points part of the problem? I was suggesting the pie menus primarily for nautilus/desktop. But I also am curious what people plan on GnomeMobile. -- ThiloPfennig 2008-09-26 09:47:04
- I would recommend against extrapolating too much about desktop usability from games. Games usually have a limited set of verbs ("attack", "talk", "pickup" etc.), whose icons the player memorize and then utilize game session upon game session. Thus a steep learning curve is a good choice, because there is only a limited set of verbs (pie menu icons) to learn and once that is done, nothing new has to be learned. In contrast, the desktop has a large set of verbs that change with the application and the task to be done. Thus a steep learning curve would be a bad choice as it would just keep going up and up, as the set of verbs is constantly changing - there is always a new icon or ordering of icons to learn. In a game you play the same 'level' or 'dungeon', using the same verbs, over and over again, but on the desktop you are much less likely to play the exact same 'level' or workflow scenario twice, as that would be ineffective.
1.4. Comments
JustinPealing: Circular menu's sound great, new computer users often have trouble with drop down menus - many people could find circular menus far easier. Power users would also benefit as long as there was consistency in the position of common operations, which leads to the question - what happens if you want to use a circular context menu close to the edge of the screen? Either the menu would need to change its shape (destroying consistency in terms of menu item locations), or the menu would need to move away from the target area (or the target area move away from the edge of the screen somehow!)
- Also - the above example looks great, kind of like the circular equivalent of the current gnome menu, but it needs to be in the middle of the screen to work effectively. Are there any way's that such a menu could be easily accessible at all times, without the need to clear all your existing windows, or use some sort of keyboard shortcut?
I like to see the projects working together if possible (maybe at least share the library? -- ThiloPfennig 2007-08-11 17:58:49
Circular menus look cool, but there are a few things to think about and research into before we can seriously consider them as a replacement for the PARC pull-down menu paradigm. For example, in pull-down menus, the items tend to keep their positions relative to mouse even if new items are added at the bottom of the list. Also, the size and position of each item is usually fixed and can be predicted and learned. These things aid mouse navigation; predictable is good. If circular menus were to rotate around or their items grow bigger or smaller too much, users would only get frustrated in the long run and there would be a cap on the maximum practical speed of item selection much higher than it is with pull-down menus. — ZoranRilak 2008-07-05 14:46:49
Well I dont agree that regular menus dont have this kind of problem. For instance my GNOME menus change when I install software. You only get predictability when you have a stable set of menu entries. So this would be if GNOME would consist of a defined set. It does, sort of, but the menus still are random. (Applications, Places, System). So I dont think that this problem is specific to pie menus. Also it is absolutely possible to make pie menus with the same entries with always the same size. -- -- ThiloPfennig 2008-07-05 20:04:31
- I suppose the formalism is like this:
When adding items to position x in a dropdown menu, all items above item x will retain their screen positions.
When adding items to position x in a 360 degrees pie menu, all items will move by a fraction of the size of the new item.
Personally, I can almost feel the pain from trying to read a menu which does not follow the normal reading direction just thinking about it. The only situation I think the pie menu can be a win is when item positions can convey some significance about the item itself - a menu for multi-way navigation (i.e. a compass menu, north-east-south-west) is the obvious example but you can think of other groups of signs that may be laid out like that, e.g. stream control (play-forward-stop-reverse), mathematical operations (add-multiply-subtract-divide) etc. So, if anything, I would suggest keeping regular menus as they are but provide a pie menu widget as a convenience for these special cases. - AndersFeder 2024-10-23 10:59:08
There is no such thing as a "normal" reading direction. There are reading directions in the western way, arabic, japanese, etc. I think one could play with the idea - I am quite sure that it is powerful. I think one thing that would have to be included when one would get seriously implementing it is that one should think twice what to put in a menu. So pie menus with all those random icons of today does not make much sense. But today we have this phenomenon: On english systems "Accessories" is the first item in applications menu - on german system its called "Zubehör" and is the last item. So to organize alphabetically does not make those small tools more accessible - and also on different systems you will find different locations. I think with dropdown menus there is no real better solutions for this problem - as when you give up alphabetical sorting it will be harder. Only thing you can do is to think differently on what you present wheren and WHEN. Time and behaviour is also something one can be used in organizing items. The only inventions in the past have been to organize 3D or search interfaces. I am not saying pie menus are IT - but I think the conservative drop-down approach is stupid and 3d and search interfaces are just cosmetics. -- ThiloPfennig 2008-07-07 00:01:22
Aha, now we are getting somewhere. Instead of speaking of pie menus specifically should we talk about 'rich menus'? Menus with more detailed and explicit semantics? This I think could be a very interesting innovation of traditional drop-down menus. Incidentally, it would fit perfectly into the SemanticDesktop (though not require it). (Regarding reading directiion, I'm pretty darned sure no-one on this planet considers it normal to read in circles - and that should tell us something about human cognitive predispositions IMHO) -AndersFeder 2024-10-23 10:59:08
- I suppose the formalism is like this:
As a disabled user (with both hands and eyesight affected) a circular menu with icons and text would be easier for me; less movment, easier to navigate. Size, however, would have to be capped.