Heh as with most announcments where ubuntu makes changes to the UI the tech community is pretty dismissive, I for one love the fact that canonical are pushing forward a linux desktop experience in a way nobobdy else is, I didnt like unity much, but I trust these guys to fix it.
I find well done command interfaces much better than traditional menus for quite a lot of reasons
* they scale better, 10 items is a similiar experience to 100
* they are more discoverable, just start typing a word and go through the list of match (where help can be included)
* they can show information, each command can have a descriptive sentence (and a link to help), icons in a menu sometimes get tiny alt text
* they can adapt to your behaviour, if I use firefox all the time, pressing f in alfred give me firefox
they are more discoverable, just start typing a word and go through the list of match
I don't think that's discoverable. That is searchable: they do make it easier to search for a command you require without having to know under which menu to find it.
But they seem to be less discoverable in the sense that you cannot just browse through the menus to discover features. Having said that, there are probably ways to provide the discoverability of current menus, e.g., just list all the main menu items under each other, then typing a main menu item will show its items...
Still, discovering new features is not something you do nearly as often as using features you are familiar with, so the trade off seems to be a good one.
If the commands correspond to some set of objects (reserved words, defined functions, builtins, on-disk executables), then discoverability is indeed high.
Conventions also matter. Most Linux and GNU utilities offer:
- manpages (or, deprecated, info pages)
- a brief help with some variant of "-h|-?|--help"
- package information under /usr/share/doc/, generally with a pointer to upstream.
They'll often include readline command editing, command completion, and other features, all very useful.
Systems such as Debian's 'dwww' (built in web-enabled documentation integration) allow accessing all of this and more through your browser, if you prefer.
Not all utilities offer this. I've seen more than my share of proprietary CLI tools with very, very poor documentation and discoverability, so this isn't something that's inherent to CLIs. But it's certainly doable.
I see some 3200+ commands available on my search path, plus shell builtins and functions. My shell history includes fewer than 200 ($HISTSIZE=3000).
The situation for GUIs is very nearly always worse. Except for minimal interfaces, it's difficult to present more than a dozen or so options at a time. Some level of predictive completion may help, but in doing so you're starting to bridge the divide between a pure menu and a CLI.
I've found myself preferring CLIs increasingly, even in nominally GUI tools. The vimperator plugin for firefox claws back crucially useful vertical real-estate while offering a very powerful interface to my browser.
Perhaps I'm misreading you, but what's wrong with "bridging the divide between menus and CLI's"?
That kind of thing sounds like an excellent fit, especially for the programs shown in the accompanying video. Some things just work better with a visual interface - web browsing, graphics work, etc - so the ability to get away from the graphical sparseness of a CLI implementation while keeping its speed and expressiveness sounds like an excellent compromise to me.
The HUD proposal of Shuttleworth, on reading, is actually somewhat similar to how vimperator presently works. I can activate the statusbar with a ':' (same as in vi), and either hit tab or start typing. Though I don't get teh shinay overlay transparent displays, I'll get a list of completions for commands, URLs, bookmarks, etc.
Which is great when you're in front of a keyboard. When I'm using my phone (Android), I'm often annoyed at how niggly the interface is, and how much typing I have to do, in an environment which really does NOT support it well.
There are also users who simply learn things by rote. They don't know the commands, they don't understand the process, they simply click on a known element or (rarely) type something -- rarely more than their username or password. "My mom" or "Aunt Tilly" are the classic (if sexist) examples, but the former is certainly accurate in my case. She actively fears computers and loses about 50% of her confidence the second you park her in front of one.
As much as I think I'd like the HUD, I really do NOT think it would work for her.
Which is where Ubuntu is getting schizoid. For the past five years, it's seemed that the distro has been aiming for wider and broader appeal, with ease-of-use functions that appeal to the 99% (or more accurately, the 99.99%) -- the technologically naive (or illiterate, if you prefer). Which really pisses off this particular element of the 0.01% when I'm stuck on that interface.
It appears that the distro's taking a hard tack in the opposite direction now.
My own suggestion: do NOT aim for a one-size-fits-all interface. Simply ain't gonna work. Sure, all you Mac users will try to convince me otherwise, but aqua's an interface I find myself grossly stymied in.
Even if you know about a feature you use occasionally, you may not know that feature by name, so being able to drill-down to an obscure program or command by topic is important - especially when you have things like a scanner program name frickin' "Xine", etc.
And considering the announcement here is a road to create a searchable-not-discoverable interface on every individual gui, it seems quite like the project will be something of a train-wreck.
Further, there's no great contradiction between discoverable interfaces and searchable interface. The more "pedestrian" Windows 7 interface also allows one to discover easily as well as google-instant style search.
I suspect those of us who use and swear by Gnome-Do/Synapse/Quicksilver and the like will love this. For those who don't, think of this is as 'Google Instant' applied to the desktop interface. It feels like a whole layer has been removed from between you and what you want to do.
It's so much more efficient and flow-of-thought oriented, and opens up new opportunities for innovation on the desktop (better fuzzy search algorithms, speech control, etc). Glad to see Canonical pushing the tech here.
I use Gnome-Do, Launchy, and Quicksilver on all of my machines, and I think this is a pretty horrible idea. The lack of similarity between program names is why the aforementioned programs are useful; the similarity between menu names are why this is questionable at best. I don't think "Close Project" and start typing it, I think "File menu" and scan for it, until I know the keyboard shortcut for it. Typing as much of "Close Project" is necessary to make it realize that that's what I mean, then arrowing (or whatever) to it, is slower than mousing to it. Once I know the keyboard shortcut for it, this "HUD" silliness is useless.
And if HUD ever gives me "Close Project" when I meant "Close Window", I'm going to throw it out the goddam window.
I like that they're trying new things, but "people like Quicksilver, so people will like something entirely unlike Quicksilver" is terrifyingly bad reasoning.
On this topic at least, my world is sufficiently broad, thank-you-very-much. I appreciate the condescension, though.
The search set of applications, which are generally fairly distinctly named (especially since the Linux world finally got away from G-everything and K-everything), and menu options, which generally are not because they operate on a small set of concepts, is so vastly different in requirements and behavior that it's like trying to jam a square peg in a round hole. So, yeah, it is (almost) entirely dissimilar.
Well, that's where 1) fuzzy search and 2) intelligent, usage-based sorting kick in. Not sure about other mentioned launchers, but Kupfer (which I use) indexes not only applications, but a myriad of other indexable items, such as files, media player items, window actions and so on, so we are talking about hundreds of items, some of which certainly similarly named. It would be completely unusable without great sorting and searching algorithm.
Even casual users find typing faster than mousing.
Casual Linux users, perhaps? I can't imagine anybody who's ever watched a casual computer user use a computer saying anything like the above quote.
Most people have the mouse in their hand the entire time, and usually don't have their other hand anywhere near the keyboard. Typing involves not only clicking into a box and dropping the mouse, but actually leaning the entire body forward in order to reach the keyboard.
Even I, computer programmer by trade, spend the majority of my life in "first person shooter mode" rather than "typing mode" because most of a web dev's life is debugging things that happen in the browser (a mousish place), and pretty much anything you need to do along the code-analysis line can be done with a combination of pointing and left-handing.
If you make me drop my mouse every time I need to interact with an application on your OS, I don't think I'll use your OS.
You mean casual long-term windows/mac user, I suspect.
A user who's been taught by the OS not to use the keyboard much won't use the keyboard much - if they have to click into a box to do it in general, that's going to encourage it.
Ubuntu is attempting to make things better, not just copy the other desktop OSen. Maybe it'll work, maybe it won't. But the fact it's not what you immediately expect currently doesn't guarantee it to fail.
No, but they are not going to get casual users of other varieties by sitting around copying what's already out there. I'm reserving judgment on HUD for now, but I have a great deal of respect for an OS willing to push the boundary.
Of course, I really want Ubuntu to get into the touch space for that reason as well, because that's where casual users seem to be going.
There are actually plenty (for some value of plenty :)) casual Ubuntu users. Most of them have had Ubunutu set-up by somebody competent, but once they have it, they can use it as casually as they like. The only reason they even need somebody to install it is that they don't realize it's an alternative.
Once you become proficient with the index finder mousing (it doesn't take long) you'll never have to remove your hands from the keyboard and may even find yourself typing and mousing at the same time on occasion. I really don't see how these aren't more popular. I just wish I could find a wireless one for my home theatre setup.
moving your hand away from the keyboard is horrendous
... for you.
Every laptop keyboard in the world has a built-in pointer, and yet there's still plenty of demand for mice. So yes, while I understand that you and many people like you keep your hands on the keyboard 24/7 and are therefore several orders of magnitude more productive than the rest of us, you should probably realize that you're a very small minority of the general population.
Actually, some of the most popular laptops (eg: every macbook ever sold and every laptop on display at my local best-buy) do not have this type of built-in pointer. Trackpads are very common. Index finger pointing sticks are not. And they are much more effective than trackpads.
Given the choice between keyboard with trackpad vs keyboard and mouse, I too would choose keyboard and mouse.
I believe that I'm in a small minority of the general population simply because most people haven't been exposed to the less common alternative.
Of course I may be wrong, and perhaps most people have tried this type of keyboard and just don't like it. In any case, it's not about productivity for me. It's about annoyance reduction, comfort, and convenience.
I always found the small stick on IBM laptops utterly imprecise to the point of being unusable. OTOH I find that current Mac trackpads are as good as a mouse most of the time, even slightly better because of three and four fingers gestures.
Actually, one advantage of trackpads over the center-mounted pointers is that you can perform gestures using a trackpad. Two fingers to scroll, pinch to zoom, slide between desktops, etc. etc. You could use keyboard shortcuts for those, but when I'm Web browsing I rather enjoy using the scroll/zoom gestures on my Macbook's trackpad.
> Trackpads are very common. Index finger pointing sticks are not. And they are much more effective than trackpads.
Meh. Some people love them. Some people hate them. Most people who have both the nub and a trackpad seem to use the trackpad. I personally find that modern trackpads are far more usable, even ignoring the benefits of multi-touch and the improvements of the new clickpads.
HCI is moving toward direct interaction. The nub is the opposite direction. Touching the screen can provide something very close direct interaction. A trackpad or mouse adds a later of abstraction, as your motions are not directly on the screen, but instead translated to the screen. The mouse cursor becomes a proxy for your finger. The nub adds yet another layer of indirection, as your actions are reduced to prodding a tiny joystick. No longer is the mouse cursor a proxy for your finger, because it does not move with your finger.
I want to move the cursor a distance based on the distance I move, not based on the duration I move - it's comparable in wrongness to steering a racing car game with the arrow keys.
It depends on the total amount of pressure, not time... Just as using a mouse or a trackpad depends on the distance and not the time you move your finger. Exactly the same thing.
The difference, in practice, is just that you can't put as much force (comfortably) as you can move your finger/mouse with speed. That's why I consider the index finger pointing stick a complement to a real mouse. The same goes for the trackpad where you must have a low sensitivity to get any precision at all, or you could use acceleration but that's just awful (now the distance you move your finger have nothing to do with how long the cursor will move, now you also have to take time into consideration).
And as a complement the index finger pointing stick is vastly superior to the trackpad since you don't have to move your hands at all to switch between them. And for precision work or stuff that require "performance" a real mouse is the way to go anyway.
If it's anything like what it seems, you shouldn't need to type more than a few characters. Easily done with one hand, then you can hit enter or click the option.
Say you type "d". It brings up "delete", "documents", and "return to desktop" (with the fuzzing). Type one more character and you get the one you want, or click it. That's how it works in Windows 7, and it might be the best thing Microsoft has ever done.
That's how it sometimes works in Windows. But when I type vnc and there are no results, then vncviewer.ex and no results, then vncviewer.exe and one result with icon, and that's repeatable, I'm less impressed.
When I first had to use Windows 7 and used the new "Start" Button (or whatever they call it nowadays) I actually thought this could be a real step forward. I have been using nothing else but dmenu in my work environments and it is a pleasure. This is really just a case where bad implementation ruins the whole experience. Apparently it wasn't judged worth the effort and maybe usage-statistics show that most of their users never use that functionality. It's a loss anyhow.
As a programmer you should really try unhooking your mouse for 2 weeks straight then report back your findings. If you see the potential after that, then give something like HUD a try. I don't mean to be rude or antagonistic, but first person shooter mode sucks.
I think HUD looks pretty kick ass. I barely ever use the touchpad on my macbook and I'll never go back to using a mouse. It's like a context switch. Spotlight is great, but also having contextual search within the application looks so much better.
I already hate the chimes and beeps and twiddles that echo around the office. The very last thing in the world I want is to have it filled with the previous sounds and "computer: open file january report"
And, personal preference aside, there is no way that I can speak the _commands_ as fast as I can type them. For long paragraphs of natural prose, voice recognition is okay; for specific things, not so much.
Even when I use a laptop my hands are positioned off the keyboard. In fact, I find the transition from the Mousepad to the Keyboard more disruptive than on a desktop because you usually have ot move your non-mouse hand off the keyboard to make room for movement.
And, Laptops are great for consumption. If I'm doing some serious creation I get the hell off my laptop and go up to my office at home and work on my desktop.
Man we really despise innovation attempts for some reason today on HN. Canonical is trying to push computing into the future and progress user interface by doing really hard stuff. They're sailing uncharted waters and giving us all a lot of good research, and all for free. This is honestly awesome. Not because it's going to be a UI revolution but because we get to see a well funded company, that cares about user rights, genuinely trying to innovate and make our lives better.
Plus the HUD sounds great. Anyone here use Quicksilver or mac spotlight or gnome do? I use them non stop, I can launch/search/command my computer using a simple interface without leaving my keyboard. Now I have this built into each application. And soon developers will start to build applications with this in mind and it'll get even slicker.
Theres no reason to poo on Canonical. They are trying super hard in the face of an ungrateful tech community. It's not about developing the system you personally would enjoy. It's about innovation and the future.
I think the consensus is that everyone wants Ubuntu to be the "mainstream" Linux, which is most like the other mainstream desktop OSes and can be put up on a pedistal for its stability. All the recent changes like Unity and now this shake up a paradigm that has been in place for 2 decades.
I know for a fact that things like Unity keep me from installing Ubuntu on every family members PC as a general purpose OS. They are too different from Windows (nobody on my side really got on the OSX bandwagon) and then because of these constant huge UI changes things break that they can't fix and it would frustrate them to no end.
I don't want to speak for everyone, but I want to see experimental projects like this in some kind of Ubuntu test bad like Firefox has Aurora / Nightly to test new features. Things that work and are widely popular could be pushed into mainline releases, and things that aren't can be scrapped, maybe revisited later.
But throwing these paradigm shifts into release products makes everyone lean away from Ubuntu as a general purpose OS, which is what it was becoming best at. I think everyone just mourns the loss of potential.
the only consensus i'm seeing is that everybody wants to hate ubuntu. a mainstream product is the perfect place to test new UI concepts, because mainstream users don't care about these things and they don't have any expectations. despite what you might think, being very similar but not quite windows is not a good thing for a mainstream user.
everybody i've tested unity on has adapted to it really quickly. the very basics need teaching (things like showing them how to open the launcher), but beyond that users are pretty good at discovering things for themselves. the problem with making things very like windows is that people expect it to be exactly like windows, and panic when something isn't where they expect it to be. unity removes expectations and the user starts off with a blank slate, and they can learn fairly quickly. also, mainstream users don't panic over change they way you imply they do. they just don't notice change. there's a presentation from google that mentions when testing google instant, many users didn't even notice anything different.
the people who are hating on unity and ubuntu are not speaking for the mainstream, they are speaking for the power users who have a library of learned behaviours that they don't want to unlearn. a mainstream user doesn't have a whole lot of learned behaviours to overcome, and they will benefit more from a UI improvement than any other because so many of them are essentially re-learning the system every single time they try to accomplish something. lots of people say they want ubuntu to be built for the mainstream, but what they actually mean is that they want is a distro built specifically for themselves.
> mainstream users don't care about these things and they don't have any expectations.
You talk like a "mainstream" user has never used a computer before.
> the very basics need teaching (things like showing them how to open the launcher),
This is the reason why, in Windows, the start button is called the start button. After all these years, the start button doesn't have a label anymore because everyone, everywhere, now knows you that click the button in the bottom left corner to do anything.
for all intents and purposes, it's pretty safe to assume that a mainstream user has never used a computer before. after a couple years of onsite tech support, the biggest thing i took away was that when the average user sits down at a computer, they don't remember any usage patterns they may have learned during their last session. the way to make your UI usable is to assume that every user is using your software for the first time, every time.
there was a good rant on the verge a while back about the condescending UI, but from everything i've seen a little condescension is an essential part of a good UI.
I was using Mint until last year until I switched back to Ubuntu again last december, for 11,10. I was not really looking forward to that since I had heard so many things about Unity being a horrible choice and so on. I was pleasantly surprised that it was far from being the broken interface many people were screaming about. It may not be perfect (I still prefer KDE for several reasons) but it is, nevertheless, functional and it does the job.
Mainsteam users who can't adapt to those kind of changes are not going to use Ubuntu anyway. Windows is already very good at keeping its interface consistent from one generation to another.
I see no reason to hate Ubuntu for their choices - I don't think they are after market leadership, they want to make a place for themselves, and differenciation is key to achieve that. And at heart, Unbuntu still retains a number of qualities like stability and relatively good compatibility across a large range of hardware.
I don't think people here are giving Canonical a hard time because they're trying to innovate. I think it's because while Canonical is innovating, they're shoving alpha-quality innovations down the throats of their users. They're making their users their alpha and beta testers, in effect.
This is a perfect example. If Canonical's past innovations are any hint, HUD will be released broken and unfinished--and in an LTS release no less. If they completely replace the menubar in this LTS, users will get mad (and rightly so), and Canonical will throw up their hands and say, "But guys it's not done yet! Give it a chance in the next release!" Which is what they always say.
That's no fun for people who just want to get work done.
> will ultimately replace menus in Unity applications
Why? Obviously, HUD and menus have different functions, and different use. GUI menus might be more appropriate for GUI applications, since you have your hand on the mouse already and you don't need to move it to the keyboard to type the command.
> fuzzy matching, and it can learn what you usually do so it can prioritise the things you use often.
Every time that I saw something like this in any application/OS (Android is a prime example), it was such an epic fail that it is almost beyond words.
1) I usually learn faster than a computer, so the way I see it is not computer "learning", but computer arbitrarily "changing" its behaviour. So, instead of memorizing "to call Jack, I have to press J and Down and Down", I have to always pay attention when the interface changes...
2) If there are some actions that I do much more commonly, then give me a way to access these actions more directly (interface buttons, scripting, keyboard shortcuts). But, in any case, my shortcuts will be almost always better than "automatic" ones, and also, mine won't change.
Fuzzy matching can be good, but I've yet to see a good implementation, and it, above all, has to be consistent! Not as in Android, where matching is "fuzzy" for calling, but exact for SMS...
that's true, but sometimes it's a bit too smart for my taste... There is an internal company website that I often visit, say test.company.com. Chrome always autocompletes this with test.company.com/admin, even though I quite often also visit other parts of the domain, say /backend, /statistics, and so on... Now, the action is self-reinforcing, as I would type "t", press -> to autocomplete it, delete until there was just "test.company.com/", then select one of the options that only now show in the dropdown menu.
Actually, for me this is a perfect example--since I hate the combined search/URL bar in chrome. I'm conditioned to type urls in the URL bar, not search queries, and the only matching I want done is against the urls in my history. So instead of typing "hacker" I type "news." and usually, hacker news is way down on the list until I type "news.y", and even then on some browsers, it doesn't appear at the top (even though it is the only thing in my history with a URL that has that pattern in it.) This Ubuntu HUD would worry me heavy GUI apps like GIMP that have organically developed relatively complex menus. It seems to be a good thing to be able to search for the menu item, but having to remember all the different menu names could be problematic, not to mention the dependence on the keyboard on a app that is primarily mouse driven. I'm assuming context/right click menus won't be impacted by this?
> I usually learn faster than a computer, so the way I see it is not computer "learning", but computer arbitrarily "changing" its behaviour. So, instead of memorizing "to call Jack, I have to press J and Down and Down", I have to always pay attention when the interface changes...
Thanks for putting this in words. This is an example of an innovation that reduces the learning curve to a device, but almost certainly makes it slower for the experienced user. One of the reasons that I use GNU/Linux on my personal computers is precisely because FOSS interfaces tend to not make this trade-off (as evidenced by Emacs, Vim, Blender, Inkscape, and many more). So, I am a bit luke warm towards this new HUD interface (then again, I haven't forced myself to swallow the Unity pill yet, either).
My keyboard doesn't randomly swap around the keys on my keyboard, and I appreciate that because I spent the time learning to touch type. Arguably, adaptively moving keys could be faster for 'hunt and peck' typists, but not for people who have spent some time learning how to type.
> Obviously, HUD and menus have different functions, and different use.
A hybrid wouldn't be impossible. In the example screenshot, the menu breadcrumb trail is shown. One possible solution would be to show the top level menu items in the HUD when it's first pulled up and allow the user to navigate it via mouse.
I think there's an interesting behind-the-scenes change that is being demonstrated here: the transition from "every app handles displaying its own menu" to "apps just export their menu structure over DBus, and another program handles displaying them" is what makes it even possible to add the HUD to all applications instead of just developing a single app that uses a HUD instead of menus and leaving the rest behind. Now that all the applications are exporting their menus over DBus, if you don't like Canonical's menu interface, you can write your own and it will work with all applications. That's pretty cool.
For instance, any one of the Quicksilver-alikes available for Linux could add a plugin that turns it into a frontend for all your application menus. The KDE Alt+F2 run dialog could do likewise. Hell, you could write a menu interface in Emacs Lisp if you really wanted to, and control all your applications from the Emacs minibuffer.
I am not sure whether this would be possible, as menus are contextual - the launcher (QS-clone, Emacs, whatever) would have to keep your current application focused, i.e. not take over its focus in order to display its menu, right?
I believe that all applications that support this do so by exporting their menu structures over DBus at all times, not just when they are focused. For context, a launcher could provide access to all application menus, but prioritize results based on which applications were most recently accessed. Or just keep it simple and go with only the most recently focused one.
This is based on a stupid idea that keyboard shortcuts are fast only because they are done on a keyboard. But this is simply false -- they are fast because they are simple gestures I can train and perform without thinking about it. Same applies to mouse -- in a well designed GUI, one can easily train to select an option or go through a dialog box with a blink of an eye.
Now, typing anything is just a complex procedure involving quite a lot of attention; what's worse, it cannot be trained since this HUD's response for a certain input will be mostly unpredictable depending of the state of its fuzzy AI. Not to mention that it additionally makes one constantly waste time on reading HUD window to check what action it is planning to execute.
Finally, when drawing or using any heavy mouse-dependent application one usually can fire keyboard shortcuts with one hand while the second is constantly holding mouse -- typing requires two hands, so it makes one waste additional time to grab the mouse back and re-adjust hand.
Have you ever used things like QuickSilver, Gnome-Do, etc. The work great and complemnt Shortcuts and Menus.
I think this is very clever. Think you are using a complex software like OpenOffice - you know some feature you want to use, but don't recall the shortcut, just type in the name and to can use it. No movement away from the keyboard neccessary - very powerful idea.
Funny cause HUD reminds me of Maya who has one hell fast mouse based UI/Widgets. Unfortunately, almost nobody in the software industry (I mean beside some 2d/3d progrmas) knows or follows this kind of design.
You make me wanna design a keyboardless system, or maybe a two thumb only input system.
I'm on OS X right now, and I use Spotlight for absolutely everything. The help menu has a feature to Spotlight within menus, but it's not great if you don't know the literal name of the command (like save or open). Things can get complex quickly in robust applications.
What about appending alternative names to menu commands as metadata, so when a user searches for 'paint bucket' in Photoshop, they're driven to 'fill'. I'd imagine that's the easiest way to map intent to action. Further still, make menus public data-driven — learn common intent by what users search for and end up clicking on.
The use cases for this might be small, but the benefit of keyboard shortcuts is in discovery. Users should be able to articulate to applications what they intend to do.
Edit: If this exists already, we're living in the future.
I don't remember any example right now, but lot of system utilities in Windows 7 has such an alternate description in metadata discoverable through Windows start search input.
I've found similar behaviour in IOS - I've typed 'pocket' into IOS spotlight to launch PocketCasts app forgetting that it changed name to Casts. The app has been found nevertheless.
I may have missed it, but have they actuelly had real users(i.e. not ui designers) do some usability testing (i.e. actual real work) on this? Coming up with it an pushing it into a public release in 4 months looks risky to me. Given how frustrated some users were with Unity, I really hope they learned the lesson.
This as presented this has a very high risk of turning into a "guess what the designers were thinking" game. While very flattering to the designer, it is frustrating as hell to the user. Arguably the current interfaces offer pretty much the same issue : a "guess in which menu I put the option" game - however there is one crucial difference : we got used to the current one, and actually have a really good idea how the designers were thinking - and where they put that option.
My main concern with this is discoverability - menus are helpful in that they convey a very precise listing of the app's possible actions (when designed well), something that's hard to otherwise come by.
Blender is perhaps the extreme example of this - it has an incredible amount of features, so they can't be placed into menus, and you spend some time learning where (in the hundreds of panels) or which of the thousands of keyboard shortcuts to use.
That said, I find keyboard-driven user interfaces very powerful, so I'm curious to try this out. Fortunately, there's been a way to turn off Crazy Ubuntu Features in most releases if you know how, at least so far.
> For a start, we haven’t addressed the secondary aspect of the menu, as a visible map of the functionality in an app. That discoverability is of course entirely absent from the HUD; the old menu is still there for now, but we’d like to replace it altogether not just supplement it.
Blender actually has a HUD-like feature already. You can just hit space to bring up a searchable list of every command in the program. As a new user, I found it a lot easier to find things with the HUD than with the enormous menu and panel system.
Funnily enough, my experience was the opposite. I spent a while playing with Blender 2.49, and although its UI wasn't the most intuitive, at least I could use the menus to get a feel for what functionality was available. This new unified drop-down has the effect of grouping commands by textual similarity instead of conceptual similarity, which is much less useful. Even if you know what action you want to take, there doesn't seem to be an easy way to do it unless you know exactly how Blender spells it.
I upgraded to Ubuntu 11.10 yesterday, and the experience sucked. It broke Firefox and Emacs, and it took me a lot of fiddling to get back to a desktop I can live with. Turns out, for instance, that you have to alt right click the bottom and top bars in Gnome classic, to modify them. Not exactly an easy thing to figure out if you've just been right clicking them for years.
Time for Ubuntu to stop dicking around with all the UI stuff and go back to making something that is solid. I really liked it when it first came out: it was a fairly dependable system that had regular updates, and was 'good enough' in the polish department. I could use it on desktops and servers alike and be pretty happy with it.
I "sidegraded" to mint 11 right after the dubiously-named "upgrade" to 11.10 which broke the ubuntu 'classic' desktop. Mint had a normal interface and behaved as one would expect. Given how close the base was to ubuntu, the transition went smoothly. And I like green better than whatever undefined color ubuntu is anyway.
I recently upgraded to mint 12 and they are picking up a bit of the gnome3 suckiness - but it feels like they are really being unwillingly dragged into it and are fighting tooth an nail to keep it at bay. It is a bit worse than mint 11, but still a lot better than ubuntu 11.10. Maybe I should give that MATE thing a shot.
The only thing I kinda miss from ubuntu is the inplace upgrade process. Mint community seems to have some philosophical rejection of it and recommends a reinstall.
I am feeling that Ubuntu is pushing realy HARD to get into tablet space - throwing-users-under-the-bus-if-need-be kind of hard. I can see the advantage of a Unity interface there, but come on, i had to manually tweak the laptop with powertop to even get close to the battery life I was getting with windows. This will never work for an average-joe-grade tablet.
I know, I know It is free, if i dont like it, I can switch to something else. Guess what? I did. According to distrowatch i'm not the only one. It is a real pity to see a good distro go down like that.
> Mint community seems to have some philosophical rejection of it and recommends a reinstall.
I was doing in-place upgrades of Debian in the late 90ies, and they generally went quite smoothly. Not being able to to do that gives me a strong urge to use blunt terms like "horse shit". In-place upgrades are one of the reasons why you have an advanced package management system in the first place, and abandoning that is a huge step backwards. I do, after all, use this stuff on servers too.
Don't get me wrong. It is possible, the community website provides an accurate walkthrough of how to do so and it looks totally scriptable. Mint just doesnt seem to have the "fire and forget" option ubuntu has. Their justification is that there is a risk of data loss/system corruption (see section E1 in link) :
> The only advantage Ubuntu offers is that it makes the process trivial and fully automated. Though, considering the risks and the way it upgrades your system, this should be considered dangerous.
I also sidegraded to mint 11 last year, and I tried upgrading to mint 12 but I didn't like the extra fluff they added, so I'm sticking with mint 11 for now. At some point I might just do a headless install of Ubuntu 11.10 and then install the LXDE and a few choice apps from there. We'll see, I'm open to trying something new, but I prefer minimalism.
"Time for Ubuntu to stop dicking around with all the UI stuff and go back to making something that is solid."
Yeah, no kidding. Their problem isn't that they're just not X steps ahead of everyone else, it's that they aren't even close to the same level as OS X or Windows. They need to stop trying to be so clever and focus on the basics.
I dunno. I run 10.10 at home and 11.10 at work (both relatively new machines). I really, really, really like hitting the super key and getting the dash, typing fire and hitting enter and opening firefox. In addition, you can hit super and a number and it opens the app in that number.
Maybe i just didnt love gnome2 enough. That being said, I really hate the way Unity appears to have broken the readline interface (for emacs shortcuts everywhere). Maybe I'm missing something, but I couldn't get that working in either 11.04 or .10.
How do they get close if they don't try? Perhaps it's just the contrarian in me, but I'm glad to see some movement here. Admittedly, though, it does seem a bit "shoehorned" in, and not terribly discoverable, but I'll withhold judgment until I've tried it. I can't live without something QuickSilver-like at this point, so I should be well within their target audience.
Trying it is awesome, but only if it's an extra, not something that's happening in place, of, say, testing their distribution's upgrade process, testing it on disparate hardware, and other tedious things like that that make for a pleasant experience.
Entirely agree, people always defend Ubuntu's action by saying that they are after a different class of user (e.g the developers mum rather than the developer) and this is why they are changing the UI.
What they fail to realize is that even with the best UI in the world if 6 months down the line joe public clicks the "yes please upgrade me to the new ubuntu" button and then magically his wifi and sound break and his UI is replaced with something totally different that requires re-learning they are going to lose confidence pretty fast.
My wife, who is certainly smart and inquisitive, but not a 'power user', has been using Ubuntu just fine for several years with little intervention on my behalf, and she's going to have the same reaction I did when we upgrade her computer:
"Where the fuck did everything move to?!"
Which I guess is fine if you're willing to throw current users under the bus in the hopes of attracting lots of new users who will have never used the older versions.
(Admittedly, the neutrality of the experiment has already been somewhat compromised because she heard me swearing up and down about the upgrade last night)
I started with Ubuntu in 2006 and I don't remember a time when they were ever "solid" or didn't break things up upgrades. Rule of thumb has always been to partition a separate /home and do a clean install every time. Maybe back when they were closer to Debian, but then that's just Debian's stability (not Ubuntu's).
Alt+Click is such a bad idea. How is a new user ever going to discover that? I am probably using Unity wrong (only occasionally on my old notebook), but I frequently am only able to do taks because I now they existed from previous versions. For example launch the upgrade manager.
Right. In Gnome 3, you don't even seem to get launchers that you can add where you want. You get some kind of activities thing. I want to press a button and get a new rxvt, not go play activities with the teletubbies.
Xfce looks like it may be an interesting alternative at this point. Gnome "fallback", or "classic" mostly does what I want, but with a name like that, it doesn't sound like something that will get much love/maintenance in the future.
Unfortunately, you (and I) are no longer GNOME's audience; We aren't who they're interested in targeting anymore. This would irk me less if GNOME 3 was a new/separate project instead of the future of GNOME 2, but c'est la vie. I've been trying to use XFCE instead, but in reality it falls short of GNOME 2 in so many ways it could never be a replacement.
Fortunately, some people have forked GNOME 2 and plan on continuing its development under the name MATE, and they have Ubuntu/Debian repositories up.
Hmm, I did not like the idea initially, but then I started thinking of how Emacs and (more importantly) Sublime Text work ... I must say, this very system works amazingly well for Sublime Text, why am I presuming that it is an automatic failure for menu bars?
Actually, I think that menu bars in general are a horrible crutch for pretty much every task. There is a reason why Windows introduced the "Ribbon". The Ribbon is Microsoft's way to get rid of the menu bar. Maybe Ribbons are not perfect, but they definitely improved discoverability of content.
ido-mode in Emacs introduces fuzzy matching for file navigation and it is awesome. CMD-T is a similar mechanism in Textmate. Sublime Text is pretty much built on the concept of fuzzy matched palettes.
All these are programs I very much like. Maybe Ubuntu is actually on to something there.
Seems like there have been mostly negative comments here, admittedly I haven't used this yet but I think it's a great idea. Ultimately this is not supposed to replace key-bindings but rather allow me to quickly access features that I normally don't access without hunting for it in the toolbar menu. Sounds especially useful in applications where you heavily use keyboard (like text editors for example).
I believe/hope that experience will be similar to the 'command palette' of Sublime Text 2 where you can execute a feature that you don't use very often (e.g. convert to lower case <ctrl+shit+p> low<enter>) without hunting for it in the menu.
For me theory sounds good, I need to test it myself to decide whether I like it in practice or not.
It's Google vs Yahoo on the menubar! Currently we use the Yahoo method (hierarchical information based on someone else's curatorship), Mark is proposing the Google model (everything through a single search box).
Either one will not be as universally useful as combining them.
I'm not sure I agree with this direction, as this system seems like it has a high cognitive load, that doesn't allow a lot of relegation to muscle memory (ie it seems like you have to keep the feedback loop between your keying and the options that appear open for the duration of the command-invocation), which is what I find attractive about keyboard interfaces in general. But I could be wrong. I'll have to try and see.
In any case, I'm glad that the direct-manipulation paradigm, which I find to be generally and intrinsically impoverished, isn't the only avenue UI innovation is going down.
To this, this seems like a lot more unix-y way of doing things, and certain for someone like me who prefers to keep their hands on the keyboard, this is top-banana (yes, I know there's ctrl+alt+shift+X to doing something in one app or other, but remembering every combination in every application is a PITA).
If you've used Cloud9IDE's command line, I see this more like that -> you type git, and a popup shows you the completions for git.
There's a lot of haters, but I'm really keen on unity. It's like having gnome-do baked in. Sure, it's not quite polished, but at least it's trying different things than the identikit window managers out there.
Does it bother anyone else that this is actually the opposite of an actual heads-up display? The whole point of a HUD is that the user is given information without going out of their way to get it. This gives you less passive information than a traditional menu does.
I like the idea in principle. I just don't like nondescriptive names.
> There’s still a lot of design and code still to do. For a start, we haven’t addressed the secondary aspect of the menu, as a visible map of the functionality in an app. That discoverability is of course entirely absent from the HUD; the old menu is still there for now, but we’d like to replace it altogether not just supplement it.
Basically - they're looking for some other interface to replace that aspect of the menu, but in the meanwhile the menu is still there for discoverability purposes.
I quite like this it reminds me of Firefox Ubiquity project. Which should act as a warning to Ubuntu as that never gained any real traction beyond power-users.
The fact that Mark mentions that they one day wish to replace menus is odd. Menus work and are universally understood. This system could be great in a complimentary fashion, the blender example seems like a great use of the system.
I really hope that Ubuntu don't throw the baby out with the bath water in their attempt to innovate.
This is awesome. The current system doesn't work so well since menu items don't fit well into separate categories. In fact, nothing does. One example is the quite arbitrary division into 'File, Edit, View etc'. A discussion I hear quite often between people sitting together in front of word is -'Try that menu. Oh, that one then? No? Let me have a look'.Another example is the windows control panel, who didn't switch to flat view straight away? And grouping on the start menu into categories of programs? Largely gone. Sorting your music collection on disk? Gone. Structuring your HD in general to avoid searching? Gone. Categorizing websites? Gone.
This is just the last relic. As it is right now I'd rather google how to do something in Word than look through the menu, so this solution _has_ to be better!
3D modeling software such as Rhinoceros (http://www.rhino3d.com/) uses this model with great success and I'm sure there are others. The program has a GUI that is very modular and customizable but the command-line interface is the primary interface. It takes a short while to get used to but after you do it scales very well as your experience grows. I have thought for a while that Photoshop/Illustrator would do well with a CLI but this is a compelling case for a contextual but system-wide implementation of this concept. The closest we have is Quicksilver or Alfred and while they are indispensable, making this part of the OS would open up some interesting possibilities.
Emacs has always been at the fore-front of UI innovation :). It's just that nobody wants to admit it because it's slightly weird, not shiny and--unlike all the other UI efforts--aimed at efficiency rather than ease-of-use.
Another hyped super-feature which implementation quality likely will render it unusable. Dear Ubuntu developers! I respect you very much! Why aren't you respecting me? Why do you release seemingly untested stuff? Why your product is so unstable and buggy? Please stop making systems that hang when I lock my screen or fall apart when I switch to another window with Compiz enabled. I do not need your new fancy features. I need a system that just works. Thanks.
And the other take. Did you ever use mobile tablet device with multitouch? And after that are you really want to use that 70-ish WIMP interfaces? If you saw it one single time, when you manipulate your photos/videos "themselves", when you interact with your information and not with some buttons or menus or stuff like that, you'll never want to come back. Ones who demand full support for multitouch gestures across modern OS interfaces are not "future-oriented" people IMO. They are now-oriented. Next-generation interfaces are already here! They are the state-of-the-art for many users right today. And are you really want to impress us with tinkering around that 70s stuff? Than you guys stuck in 90s. Come on.
Whenever I see anyone write, "say hello to the future of the ______", my immediate reaction is to assume the author is an idiot. The only person or company who has ever been able to say that with any authority is Apple, and that's only because they have a track record of consistently creating the future of ______ for decades.
Considering this HUD is basically "let's take Quicksilver and make it for menus", I think my immediate reaction was correct again. The idea isn't completely terrible, but it's not "the future" and it's riddled with problems. Hope you're the type who likes keeping cheat sheets around, because unless you know the exact name of every single command in your application, you're going to be in trouble.
"We’ll resurrect the (boring) old ways of displaying the menu in 12.04..." This guy's hubris is astounding.
This is a great idea and as they note its really beneficial to the power user, BUT I can't see it replacing traditional menus for novice users (and potentially some intermediate users).
For novice users, discoverabiliy is extremely important. For any application new users want to learn what it does, and they aren't going to read the manual. Traditional menus allow users to browse available actions without having to know what they want to do. They don't know what they don't know, and well designed menus/UIs allow them to learn the application really quickly.
I also wonder about the older generation and their ability to pick up on this. I would be very curious how well less technically skilled 40+-year-olds pick up on the keyboard-based navigation over the traditional mouse-based navigation/interaction.
Is this actually solving a problem someone has stated they have?
This feels like an attempt at solving a problem that nobody has. In an effort to differentiate themselves, I think they've lost sight of what users actually want.
Here's what I want from ubuntu, as a (past) user: a solid distribution that gives me a reasonable default setup without a lot of fuss. I mean, I'm installing linux, you can probably already assume I sort of know what I'm doing. I want something that gets out of my way, not something constantly asking me to adapt and "rethink" how I go about using my computer. I know how to use my computer just fine, thank-you-very-much, leave me with my menus.
This is a beautiful and interesting concept, but isn't this really just a dressed up command line interface? This is a feature I would absolutely love to have, but I don't see it replacing the menu in the form demoed for the simple reason that this violates one of the core principles of user-centered design: visibility. If the user has no visual indication of what is possible within the system, how can we expect them to learn the interface?
Most readers of Hacker News are probably aware of the calculator feature in the Spotlight in Mac OSX, but if I had a dollar for every time I've blown someone's mind by showing it to them, I'd be counting money right now rather than writing this comment. The bottom line is, visibility is an extremely important design principle, because it informs the user what they can do within a system, so hiding possibilities is probably not a good idea.
I'm pretty excited about how this could make voice recognition a first class citizen on the desktop.
The simple voice tool baked into Windows* is sorta usable, but changing contexts to find things in menus or click buttons is a huge pain. You're constantly reminded that the interface you're using was originally designed for a keyboard and mouse. This has the potential to eliminate a lot of the pain associated with that.
* I don't know if Dragon improves the context-shift situation significantly. The last time I tried it was several years ago.
OS X has had this since Snow Leopard -- you can go to any application's help menu, and begin typing the name of a menu item, even if it's nested. Using the arrow keys, you can navigate down the list of results, and as you select each item, the menus expand to show you the location of the item with a large, blue arrow next to it. You can then either click the item next to said arrow, activating it -- or simply press enter.
Apple had this years ago; they just made a conscious decision to keep using menus.
What do I do when I forget the wording of a menu item? (or I guess a "HUD command")
Menus form a strong spatial association - I remember a lot of options by their placement, and only recall the exact wording when I see it. (Is it "re-open last tab", "reopen last tab", "re-open closed tab", or "reopen closed tab"?) Not saying that HUD is inherently bad because of this, but I'm curious how they deal with it? Am I missing something?
This doesn't seem like it would be very good for using with applications that you are familiar with. With context menus
you can remember the ALT + <key> combinations required to navigate to a feature quickly. With this you would have to type the name of the feature and then select it which might involve navigating a fair way down the list.
The only other option would be for all apps to assign keyboard shortcuts to everything
Also for applications that you are not familiar with you will have to guess what a particular feature is called and if it even exists in the first place. They would have to load all the apps up with synonyms and since Open Source developers can't agree on anything I'm not sure how many would even design their app around this.
Not to mention problems with users who have poor english or just poor spelling.
Having said all this, this feature is an excellent idea to supplement existing UI functionality. I've been integrating "feature search" type functions into some of my web apps for a while now and I'm surprised Apple and MS haven't done this more.
Looks to me like we are going full circle back to the command line with this. I mean, what's the difference between this and a command line shell with fuzzy matching?
I like Ubuntu being experimental but right now I'm having pains with Unity on my setup (multiple monitors and apps that don't quite fit with the UI). I'm seriously considering moving to a different distro while they sort the details out .
I was thinking the same, and found it particularly amusing that "design" types are only now finally realising the virtues of the command line. What'll the next Ubuntu UI revolution be? No X11, just a 'screen' session in a tty? (I'd actually use that, were it not for the fact that (a) too much of the Web is overly graphical to be browsed effectively in text mode and (b) screen sucks)
Moving to a different distro (and perhaps WM) is a good idea. I recently moved to Debian with Xfce and am enjoying the lack of gnomes.
Putting the keyboard back into computing is a strange direction to move considering the wild success of tablets and smartphones (namely of the Apple variant) that drop the keyboard entirely in favor of better touch interaction. An ideal interface has a low barrier between what you want to achieve and the way you achieve it, using a keyboard requires that written language always be between the two.
Ubuntu's current (11.10 oneiric) menu cleverness breaks things for me using a non-standard window manager. (The gnome-terminal menu appears even when set to hidden.) Solution: purge every package with appmenu in the name and reboot. I hope it remains as easy to opt out from the next batch of little-tested, slightly buggy innovations.
Another invention which will alienate even more casual users from GNU/Linux.
Users, even I as a progammer, do not like to have to re-learn and assimilate a new interface-model every few months or years. Ive learned a few times, why cant people stick to what is working already? The menu is super nice, the global menu, not so much but fine thats what some people nice. Now Ill explain to my girlfriend that there is the HUD menu? And after that youll invent something else something new, for what!? All that work and you just have a different kind of menu. Too bad, too bad.
Why not do some real user tests to find out where they are failing right now? Why users dont feel comfortable in Ubuntu? Yes, even those users, like me, and Ive used most distros for a decade. Its just something fishy about both Unity and Gnome2.
I once worked on a project with some friends that gave Windows this functionality. It basically scanned the menu of the current window, then gave a Launchy-like interface to choose any menu item. It worked pretty well, although some programs bypassed the Windows API so didn't work with our program.
We also gave it one other awesome ability, which I have yet to see in other such tools - a "search for window by name" feature. This searched your open windows in a Launchy-style interface, letting you switch to Chrome by doing "Ch<enter>" or to Visual Studio by doing "Vis<enter>". It also searched open tabs in all your browsers, letting you switch to gmail by typing "gm<enter>", etc.
The program was meant as a startup, way back many years ago when we didn't know any better :)
Strangely enough that core functionality (searching menu items) is kind of built into OSX. Along with ⌘-space for Spotlight, I use shift-⌘-? to search the menubar from keyboard. Its a great multi-tool as it hooks into the stuff you already know.
I think it would be better to offer the whole menu bar as an expanded tree in a vertical strip rendered to an overlay on the whole screen.
This would help the user get an at-a-glance view of all the menus' contents, sectioned logically by each top-level menu entry. It would also reduce menu navigation to mere up/down (and pgup/pgdown), and allow for some sort of an instant search (start typing letters) would also work. It would even work for touchscreens because we've already observed that scrolling up and down is one of the best and natural UI concepts that you can do with a touchscreen.
After two iterations of this I can se it being pretty good for my parents as they will be able to say "bookmark" or "back" to the voice recognition in the browser and not have to worry about the UI much at all.
That's another novel thing for Ubuntu, publishing a major change first in an LTS. Until now the philosophy was that an LTS was more stable, a non-LTS was more experimental. Sad that this policy was abandoned.
The "HUD" interface (silly name) looks like a good idea though, hope it is fast enough for netbooks and other small PCs. After using Unity for some time I abandoned it for XFCE because my netbook was just not fast enough.
This seems useful for exploring menu options when you don't know where things are, but not as useful for executing frequently used menu options. For exploring, it's great to be able to type "pref" and have all menu options for preferences show up. But if I know that the preferences I'm looking for are under Edit - Preferences, I'll just hit ALT-E, P and be done with it in less than a second.
I'm still not sold, I found the UL-Corner positioning greatly aggressive. I'd put it in the bottom personally but I'm used to the Unix 'bottom command line' habit.
Otherwise it's one step closer to a contemporary desktop command line, in the video the guy even typed 'undo' in inkscape, made me smile. I guess having this common iterative search of nested menus will greatly help users, as it helped many of us in the shell or emacs (or any system providing this idea)
I fail to see how this will apply to tablets and mobile phones, where you want people to use their finger as a mouse and not as a keyboard to type. (since Canonical said they were moving towards tablets and phones in the near future, I expect they follow the same design guidelines, but I could be wrong...)
I doubt this will completely eliminate other forms of interaction. It would be a mistake to remove keyboard shortcuts or buttons for common actions such as changing pages. The point is to not get into a ribbon type situation where there are a ton of buttons cluttering the window.
What a horrible idea. What happened to thinking about international users, for whom typing is a hassle? Learn from Japan, where Yahoo is more preferable over Google because you can find what you want rather than type it out...
The only thing I dont like about this is that it appears that you have to type for everything you're looking for.
With the example shown in inkscape - this is good if you dont know where something is -- but anyone who is proficient with a graphic(s) program knows that the more you can do with a left hand keyboard shortcut and a button which reduces the number of times you take your hand off the mouse the better.
What I would far prefer is if the HUD would allow me to assign arbitrary keyboard shortcuts (contextually) in any program.
This way - when Inkscape is active, I can use whatever shortcut I want to assign to "glow" and then that same binding can be set to a different command in a different program, say, GIMP - and I configure it all in one location.
I dont want to type for everything I want to do...