Sure, but you essentially end up with virtual desktops, only you can't visualise them (because they are invisible). I don't see the added value.However, for the sake of argument, let me expand a bit on my previous idea. These groupings of windows would be running, essentially, virtual desktops invisibly. To get make them visible should only require a tiny bit of UI code.
Not really. You group them together based on the task at hand, just like you would with your groupings. Put all windows you work with on a particular desktop. Then you only need to sift through those, plus it's easy to zoom out and see at a glance what is where.To your specific example, I don't know why you'd have a task that requires 20 windows. However, for the sake of understanding, you'd still have those 20 windows to sift through on your virtual desktops.
You then change the background for each desktop so it's immediately apparent where you are at the moment. Visual memory ftw.
Mental imaginings is a barrier to better general usability. People have varying amounts of ability, in that regard, and a desktop should make as few assumptions in those regards as is possible, IMHO.Sure, but you essentially end up with virtual desktops, only you can't visualise them (because they are invisible). I don't see the added value.
If someone is a visual person, you go to overview and see the various groups of windows (the exact UI wouldn't be like Panorama, but, at a high level, that is the closest thing I've seen). The alt-tab can also be exposed in different ways (perhaps it would limit you to a particular task group, but that would need to be mocked up and tested). Assigning keyboard shortcuts to the task groups should also be possible.
I think you misunderstood what I meant. If you've used Panorama all the windows you have open aren't visible (even assuming you choose to go with the mouse navigation, which I assume you wouldn't) only the windows in the task group that you're currently using are visible. So, as I said, it is no more cluttered with windows that virtual desktops would be, but the advantage is that they are conceptually easier to use, IMHO.Not really. You group them together based on the task at hand, just like you would with your groupings. Put all windows you work with on a particular desktop. Then you only need to sift through those, plus it's easy to zoom out and see at a glance what is where.
Sure you can change the background. I prefer naming the groups, as Panorama does, but some people may prefer the more visual differentiation that backgrounds provide (since I basically never see my desktop it doesn't matter to me but, as you say, some people do care and the amount of code added shouldn't be a great deal).You then change the background for each desktop so it's immediately apparent where you are at the moment. Visual memory ftw
Ideally, I would like a way to arrange windows in each task group in a specific pattern and have them remain that way (a bit like tiling but not as rigid).
For the most part I would imagine you would only need a few task groups and those would almost never change (say, Research, Coding, Fun Time, for example). So, once you set these up, and setting them up is as easy as launching the windows in Overview, lassoing/drag-n-drop/alt-click/whatever the windows you want to group, name the group.
There's probably some other things that would be general productivity boosters that don't also incur significent mental overhead, but these changes aren't too difficult and would make GS more generally useful.
Reading about Green Island: A New Qt-Based Wayland Compositor, which also mentions Hawaii Linux Desktop Environment and Maui OS "the Wayland-friendly desktop that can be used on other Linux distributions too" got me curious so I did the natural thing and 'oogled the aforementioned terms.
There appears to be very little public information about these projects out there, but among the few hits was a somewhat soberingly sceptical viewpoint on Mr Fiorini's (PLF ?) track record for carrying projects through:
The thing is, I'd be interested in hearing how these projects get off the ground. Glancing over the rather spartan Maui project FAQ left me with more questions than I had before, like how are the other "desktop" environments (IIRC some GNOME developers have wanted the whole concept of "desktop" canned and banned...) either too much or too little configurable...He likes to start these little toy OS projects. He loses interest and they get abandoned after a little while. He has started (and given up on) several since OpenBeOS (now Haiku) began. Some pieces from previous projects get re-used in each reboot, and there are common themes. Fiorini likes Qt a lot, for example, so that usually features, and he's always interested in new-ish technologies born on Linux, DBus was one example, Wayland is another.
The on-again off-again OS project has been called BeFree, Vision, Mockup, and now apparently Maui. Maybe I missed some names in there.
NB: FWIW I don't know any of the characters involved even remotely and I apologize in advance if the quoted message was part of some (to me) unknown flame war, but the questions that were raised there did sound valid enough.
Let's try to move back to the actual thread topic, instead the bland Gnome/KDE flame war (people like different DEs, get over it)...
I quite like the idea of a distro being locked to a specific DE if it creates a slicker and more stable experience, although it's a brave/stupid move for a brand new DE!
"The idea behind Maui is to avoid traditional packages . . . Applications will be installed from bundles."
What is the difference? How are dependencies handled?
So- pretty much no dependencies will be needed. Until they explicitly tell, this is just a hypothesis though ..
Pier Luigi <---- first name
Fiorini <---- family name
The large Icons are good for quickly finding an app (and pinning them in a consistent place for commonly used apps is nice), but sometimes you're looking for a specific document window, not an app. When you have 5 or 6 different spreadsheets open that all look very similar (same template with different data filled in), the "plaintext with a small icon" with the title is much better than any graphical window display/thumbnail view. That's what titles are for: Quickly and unambiguously identifying documents. A similar case arises when referencing multiple pdf documents that are primarily text based and not easily distinguishable by appearance. I actually like Unity for certain tasks, but some things I've switched from LibreOffice to Google Docs just to be able to open documents in Firefox tabs with nice "title-centric" navigation, and I keep KDE installed also because certain workflows are just too inconvenient in Unity. In my experience (it might just be the kind of documents I work on) "reading titles" is by far the easiest way to "identify which one is which".
One thing that I think would be absolutely great in Unity (and solve almost all of my usability issues with it) would be the ability to pin windows (not apps) to the bottom of the screen. It could be similar to the drag to top or sides to maximize or half-maximize, but would pin a "taskbar-like" item to the bottom of the screen for that window. The taskbar could go away when it's empty (and have auto-hide options even when it's not empty). The biggest problem with the traditional taskbar is that it gets cluttered up with things that you don't really want there (hence all the "minimize to system tray" options that music players, IM clients, etc have added as a workaround). This "pin to bottom" solution would solve the problem by showing only the windows you deliberately pin there. There could also be an option in the launcher's context menu for each app to let you automatically pin all windows from that app. I would probably set Evince and LibreOffice to automatically pin their document windows.
The HUD, search features, and keyboard shortcuts can all exist, pretty much independently of whether you use a traditional(ish) taskbar or not. I don't really use them much (except keyboard shortcuts for a few common things). Maybe it's just that my memory is really bad (visual memory or otherwise), but I find it much easier to browse through a deeply nested tree where each level has a small enough number of options for me to scan the whole thing quickly, than to remember the name of what I want to search for. The act of navigation through the tree is sort of a way of maintaining some structure on my own thought processes, and I'd find myself almost non-functional if I tried to work in an environment with a flat file system that relied solely on search. Search is great for finding things I don't have (on the Internet for example), but organization is much better for me for finding things that I already have, and (more importantly) remembering that I have them in the first place.
So in that case, you move the mouse to the top/bottom of the screen, select your app from the taskbar, and return the mouse to where you wanted it. Keep that in mind - it's not just switching windows with one click, it's with one click and a bunch of mouse movement, quite a lot if you have a large monitor, or a laptop with a crappy touchpad. Under Shell, I hit the Start/Windows key to momentarily show those hidden windows, and click on the one I want. Yes, it's an extra key press - but a negligible one given one hand is already near the key, and it's usually a lot less mouse movement to pick the window I want. And so in practice, I actually find it a lot easier than a taskbar.
And a related benefit - the window previews in that view are live, not just screenshots. Which means that if I have some background task running in a terminal window somewhere, I can simple check on it by popping up that overview mode with a single keypress, see that text is still scrolling down the terminal, then return to what I was doing by hitting the same key again. It's actually really useful for doing one thing while periodically checking on background tasks (e.g a database backup, a software build, a file copy, etc).