Ubuntu 11.04 will remove the notification area (systray) | Hacker News Reader

Ubuntu 11.04 will remove the notification area (systray)

design.canonical.com

63 pointskilian64 comments

Fantastic, I am just sad that KDE didn't do this first. It has been known for years that the system tray as it is today is a POS. It is being abused every way possible and causes a complete mess for users. The one I _really_ hate is when I click the CLOSE WINDOW button and the application hides in the system tray while still running. I am glad someone is trying to make things right.

I never got that argument really. It can become messy. So can the desktop, or your home directory. But do you prevent users from storing a file where they want it because of that? Windows world went too far with any program and their dog putting an updater, or status icon in the systray - but in Ubuntu you have the choice of not putting new stuff there. For example, I've got an update notifier, power, network, dropbox and volume in the notification area now. How is this abuse? Why should I resign from having those icons there?

They've already done the experiment with removing update notifications from the notification area (names point at the irony here). Instead we got a system approved pop-under, 200+ people complaining in the bugreport and Mark saying something like: That's how it's going to be, now be quiet. It was the same for window button placement lately. And that's a reason I'm upgrading to Debian this time instead of 10.04.

The abuse comes from random applications that like to sit in your system tray. Looking right now I just notice Akregator, an application I launched probably a month ago is sitting in my system tray. Pidgin also has something there. I can't tell Gnome not to allow those apps in the tray.

You're not convincing... If you're not using Akregator, why don't you close it? Pidgin has a setting in (translating into English - text might be different) Tools -> Preferences -> Show icon in the notification area - choose "Never". It could be argued that the default Pidgin behaviour (enabling the icon) is good, because many people expect only the icon, not the whole communicator to be visible while they work.

I shouldn't have to set a setting for the close button to close an application. I hate applications that do this becasue they defy my expectations when I use them. Pidgin may just happen to have a 'behave normally' setting. But not all applications will, and certainly they won't all have the same way to set it. I find that, in general, it's a better use of my time to avoid applications that do this, than try to figure out each one individually and make it behave the way I want.

Isn't the close button issue completely unrelated to the notification area issue? They don't depend on each other. Rhythmbox has the notification icon, but closes on 'X'-click.

When I click the 'x'-button on rhythmbox, it doesn't quit, it stays open in the notification area. Ubuntu 10.04

"but do you prevent users from storing a file where they want it because of that? "

There's some new "pad" thing that does exactly that.

but do you prevent users from storing a file where they want it because of that?

Then again, if most people aren't that concerned with the file hierarchy, then why foist it on them?

There's some new "pad" thing that does exactly that.

We'll see how this unfolds. I suspect that the UI of the iPhone OS has conditioned people to think of files and items as being "in" an app. I suspect that Apple will implement some way of sharing or transferring files between apps.

Fantastic, I am just sad that KDE didn't do this first.

Actually, KDE did it first: the application indicators specification started as what was basically a port of KDE's KStatusNotifier.

Isn't that a bit cocky? "We, Ubuntu, the biggest most important distro out there, are doing away with the Notification tray, so if you want our millions of users to use it, you'd better get with it." What about the other distros? Are they going to have to follow suit after all the applications rewrite their code? I can understand this coming from Gnome, perhaps, but Ubuntu doesn't own Gnome. Why not have a wrapper for the notification area that turns it into a menu, for a last resort?

That's ok. Mark Shuttleworth wants to be Jobs mk II. This is his text from the ubuntu bug report on removing the update notification from the notification area.

"But, we're not afraid of making bold moves. Ubuntu itself was a bold move, and has attracted a fair amount of criticism for its very existence, but that didn't stop us. If we want to transform the Linux desktop from where it is today, to something that Apple will feel obliged to emulate in parts, we are going to have to make bold moves and big changes, and those will cause distress. If we're right, the result will be fantastic, and the changes will be embraced by other distributions and upstream. If we're wrong, they won't."

Unfortunately many people get the feeling they're trying a lot of things without enough thought put into it - and when people report problems as bugs, Mark steps in and says no. They're simply becoming something like Apple of Linux world.

Well, maybe to backtrack a little, I don't have a problem with that Steve Jobs pioneering attitude. You can disrupt the market if you'd like. But to have that immediate effect on the developers of applications, making them change their UI fairly significantly just for Ubuntu, I don't think it's their place. If they really wanted to be bold like that, they should make the changes in the apps themselves.

Forking an indeterminate number of applications? Now that would be cocky.

Ubuntu needs to be a bit cocky if they're serious about improving desktop Linux. They won't get anywhere if they're too afraid of stepping on toes.

Agreed. Ubuntu is no less cocky than Apple. I see Ubuntu as the "Apple" of the Linux world. And they have to be cocky if they want to make any breakthroughs or any real innovations.

As another example, I have been running 10.04 beta for a while. At first I disagreed with their choice of moving all the window controls to the left, but now I am used it and actually like it better. I think it is more "efficient" as it usually minimizes my mouse movement. What is the point of this rambling? Ubuntu did something they thought was right, but wasn't popular initially. Eventually many people realized it is a better design choice. Apple (and GNOME in the Linux desktop world) have taken the same approach before. Sometimes it backfires but often it works well.

>I think it is more "efficient" as it usually minimizes my mouse movement.

Are you left handed? I tend to keep my cursor on the right presumably as an extension of the desktop metaphor - reaching across to the right means a far larger movement.

However, I can see theoretically that if you have your application launch button ("Start", "K" or what have you) on the left that you'd want to push other things that way. However I'd rather move the "pin" and "menu" buttons to the R of the menu bar.

I trust if KDE ever went that way they'd make it an option.

> Are you left handed?

No. The reason it is more efficient is because the mouse pointer usually hovers more around the left and upper edges of the windows. That is where all the most frequent menus live are and most often used toolbar buttons. You can think of it as the logical origin of the window. For example the 'File->Save As..." menu, the "File->Close" menu, the "Back & Home" buttons on the browser and so on.

It would be an interesting HCI project to run a background statistics gatherer that would record mouse coordinates relative the window in focus.

The only time I usually move the mouse to right is to scroll (but I personally use the mouse wheel for scrolling) and resize the window. But most of the time it hovers on the left side.

Also, just because you hold the mouse in the right hand doesn't say anything where the cursor is on the screen. You would have to consider where the cursor already is when you want to close/minimize/maximize the window and how far you would have to "travel" with it.

Another 2 things to consider:

1) Mouse travel might seem like a very small change, but the action of closing a window is very frequent. So a small change multiplied many times can add to quite a bit.

2) This ends up working even better for laptops since a trackpad already offers a fairly small working surface compared to the surface available for a desktop mouse. The shorter the distance the cursor moves, the less trackpad "strokes" one has to perform.

EDIT: formatting & syntax

>It would be an interesting HCI project to run a background statistics gatherer that would record mouse coordinates relative the window in focus.

It would indeed.

My mouse cursor definitely lives at teh right about 2/10 in from the right screen edge.

I do use shortcuts to access menus (as well as mouse, I'm fickle) but would see your comments as reasons to invert the menubar menu order and arrange them from the right rather than reason to move other stuff left.

But I'm probably an anomaly. In MS Windows (which I rarely use) my Vista install has the taskbar [vertically] on the right.

HN's reply button often annoys me because it is left aligned whilst as I'm a L-R text reader the proper position to me would be the right as on completion that is where my gaze falls.

KDE dialogs usually have Cancel/OK on right and help and other auxiliaries on the left. For example the systemsettings dialog has "apply" on the right, which inline with my above thinking seems the correct position.

Physical mouse travel for someone with all left-aligned action buttons could still be greater than someone using a "fast" mouse with action inputs spread across the screen.

I can't imagine that you wouldn't be able to manually add the notification area. It probably just won't be enabled or installed by default. As a last resort, you can just enable it, but for developers who want their programs to be easy to use, they'll have to make sense for the many users who won't have notification areas.

Or we can just not care about Ubuntu and develop for all the other platforms...which all have system trays

Design by committee, and the "tread softly" model of UI design is why everyone else is so completely behind Apple on the UI front. I for one am glad that Ubuntu is taking a stand against it.

There was a link posted here a day or two ago about the secret to Apple's success - they don't do focus groups, they don't have big surveys about what will and won't work. They just do, and when you have capable, knowledgeable people at the helm people will get behind you.

I see this as the same thing.

> "Why not have a wrapper for the notification area that turns it into a menu, for a last resort?"

This doesn't work, it has never worked. Just look at Windows: "hey devs, we've deprecated this horrifying monstrosity that should never have seen the light of day. For the sake of compatibility we've preserved some functionality, but please switch over to the new better thing soon"

Developers: "meh"

But what exactly does Apple do that disrupts the scene? It's their scene to disrupt, I don't really have the same problem with that. Ubuntu is just one distro, they don't own Gnome. Now developers are forced to pick sides. Maybe that's the main point I should have made.

Maybe I'm biased on this - but Linux has done nothing on the UI front for years. Even discounting gigantic embarrassments of programmer-UI like the Gimp, both Gnome and KDE have really, really bad UI. There's no consistency, nor intelligent design except perhaps "Windows is doing it, so it must be right".

Let's be honest, desktop Linux UI right now is really just a (bad) clone of Windows 2000.

So Gnome isn't going to fix their UI on their own - but yet that's what your users are interacting with every single day. I for one am glad that Ubuntu is fixing what the Gnome team has demonstrated for years that they cannot/will not.

Here's the way I see it: either we continue to sit and hope in vain that Gnome will step up and do something great with their UI... or somebody else takes charge at the cost of fracturing the community. For the sake of desktop Linux, I choose the latter.

Ubuntu doesn't own Gnome

I thought anyone was allowed to fork any free software, and was encouraged to do so if they don't like the existing direction?

Well I didn't say they didn't have every right to do it. I just think that it would be more considerate of them to negotiate with the other players. When I say that they don't "own" Gnome, I mean that they're not in change (yet, at least) of the Gnome ecosystem as a whole. Making a move like this is going to cause developers to have to pick sides between distros.

If they really want to fork Gnome, just be honest and say, "Gnome sucks, we're forking it and calling it "Gnomebuntu".

What I don't understand is why control of window behavior, including hiding/minimizing, is up to the programmer and not the user. Goes for every OS/shell I've seen so far.

upvoted. This is a very valid question - the system tray imho is just a moniker for a 16x16 icon. I would want the ability to either minimize any app to a normal toolbar or dock to a system tray.

> I would want the ability to either minimize any app to a normal toolbar or dock to a system tray.

Use alltray: [quote] Dock any application in the tray

With AllTray you can dock any application without a native tray icon into the system tray. It works well with GNOME, KDE, XFCE 4, Fluxbox, and WindowMaker. [/quote]

PS.

I prefer to use switch-or-exec plugin in Sawfish at hot-key: I will press hotkey, Sawfish will switch to application desktop area and then bring application window. If application is not running, then Sawfish will run it. Configuration is very simple: regexp for window title or window properties to find application window and command line to execute when application window is not found.

For example, I can press "Alt-Shift-e" to switch to editor window to enter few notes. If editor is not found, then new editor will be run. When I will finish, I will press "Alt-Shift-b" to switch back to browser, "A-S-m" for mail, "A-S-i" for IDE, "A-S-t" for terminal, and so on.

"Configuration is very simple: regexp for..."

I did a double take right there.

I'm surprised one of the legions of Apple fans hasn't answered yet. I can guess what a fan would say though: "it's not up to users because computers are already too complicated and have too many configuration options and users just want something simple that works and they need the programmers/designers to pick the "right" way and just implement it.

Coming from a long history of Debian and black/fluxbox, I've always appreciated Ubuntu for their package selection and default hardware setup. They've hooked up a lot of stuff I hate managing by hand (like printing and video drivers) automatically. I still run openbox on it, though, which happily insulates me from Shuttleworth's reinvention of the stock UI.

Watching the debate over new window control placement and removing the systray reminds me it's nice to be a little off the beaten path. :)

Oy, but I like being able to minimize my rhythmbox, tweetdeck and pidgin to the notification area. :(

I'll suspend judgement till I actually use whatever they come up with though.

How about being able to minimize ANY application and dragging it to the system tray. This way you wont have to wait for each app to implement this, reduce code bloat, and the rest of us wont have to wonder:

  * Why isn't this app in the taskbar?

  * What does close window do for this app?

  * What does minimize do for this app?

  * What does file quit do?

I implemented something very similar (albeit slightly less slick) for the GNOME panel. It works nicely.

http://blog.kevinmehall.net/2010/bringing_pin_tab_to_wnck

This is literally my first time hearing about pin tab in chrome. Behind the curve entirely. Thank you.

Yes, that would be very nice.

Nice idea. The top-right icons (at least, in XP) could be (from right to left): x to close, window to maximise, line to minimise to taskbar, and a dot (in a square) to minimise to the system tray.

Applications could then act differently based on where they were minimised - e.g., icon animations to convey a message, can't steal window focus, can show little alerts/messages, etc.

KDE allows this on a per application or per window basis "skip taskbar" and "skip pager". Sadly I don't think you can have alt-tab overlook such skipped pages.

That's because you've happened to grow accustomed to the misdesigned concept of "minimizing applications to tray". The fact that it's been part of graphical user interfaces for some time and mostly gone unquestioned doesn't mean it's good design.

Huh? How can something that improves usability be bad design? I've grown accustomed to it because it helps me.

It improves usability for you, perhaps largely because you've internalized the mental model that it instills ("there are two places to minimize applications to: one is the regular task list, and the other is this tray thing"). That doesn't mean it doesn't ruin usability for lots of other people, especially those new to computing.

You mistake effect for cause. I internalized that mental model because that UI paradigm improved usability. To clarify, the improvement in usability stems from being able to not clutter my task bar with long-lived apps that I'm not actively switching back and forth from (like music players and contact lists of IMs). In the absence of statistics supporting your case, I'm assuming it's the same for most other people who use that feature. As for new users, no one stays new forever, and it seems counter-productive to abandon increased functionality because it might confuse what will by definition always be a minority segment of users (or because some group of designers thinks so).

They already have solutions for this: Rhythmbox will be integrated into the volume widget and pidgin and tweetdeck/other twitter apps in the MeMenu. it's really far more elegant, I think.

Huh. I think this is the first time I've considered the possibility that there are people who actually like those notification icons. I always figured they were just the product of egotistical developers who each think their own application deserves my attention more than the others.

Personally, I hate when applications like rhythmbox crowd the notification area. There is no event involving rhythmbox that I will ever want to be notified of. "I'm still playing that song. Still playing. Still playing. OK, now it's over." Gee, thanks. I figure if an application has nothing new to tell me, it should stay over in the taskbar with all the other ones.

But indeed, a notification icon can do things that a taskbar entry can't. Notably, the context menu in the taskbar is defined by the taskbar, not the application, so things that take one step with a notification icon take two or more via the taskbar.

So now I think the problem isn't the notification area, it's the taskbar. Maybe if the taskbar worked better, developers wouldn't pollute the notification area so much.

Stuff in the notification area also takes up less space. If I have a few things running all the time, I don't have to devote half of the task bar to them.

I find it interesting how Ubuntu is modifying the actual GUI, when I would have thought such a task would fall to the GNOME developers.

As a kubuntu user, I've always hoped I'm getting a vanilla KDE with ubuntu's packaging; I want all KDE programs to work out of the box. Is KDE similarly altered by canonical? This move seems like it can only further fragment Linux.

If they weren't putting changes and policies in place to make the software more consistent and friendly, then wouldn't Ubuntu just be Debian unstable with a different logo and release schedule?

You've got some big assumptions of agreement there with the whole consistent and friendly thing there.

Some many reasons have built up over the years for the educated user to not use Ubuntu:

-- The pulse audio fiasco

-- The network manager that couldn't manage anything but vanilla networks (I hear that's changed now, but I'm not about to sacrifice any more nights of my life to Ubuntu for finding out)

-- Massive and arbitrary changes to the windowing buttons without so much as a nod to keeping the actual community involved

-- Brown, then purple

-- Core application that deletes your whole drive if you don't supply parameters. Now that's pretty bad...but the clincher is that when questioned about it, the senior Ubuntu developer gave us a most excellent entry to add to the encyclopedia of bad ideas

-- And now, no system tray, "we don't like it so we throw it out" haha

I have a feeling the revolution will not be Ubuntu-ized.

this reeks of armchair commentary,

ubuntus network manager has been first class for ~2 years, I have less problems on ubuntu that I do with my mac in regards to broadband dongles, its audio has also been long fixed

the aesthetic changes (purple to brown) are a large improvement that everyone asked for, the window buttons really dont matter that much (and they can be moved back)

the core application was very much a niche issue for those messing around inside a bootloader, its not like opening notepad ran rm -rf /

and as for the system tray, im not particularly fussed either way, but I am glad that someone is worrying and doing something about ui issues instead of sticking to the status quo

The most annoying feature of GNOME Ubuntu for me, after the look&feel, is the managers.

I use a LAN; I don't want a network manager. I use a desktop; I don't want a power manager. I have speakers; I don't want a sound manager.

If there's a way to hide them, it's not simple. You can't just uninstall them, because they've taken over the duties of actually managing the resources as well, in addition to just showing me information I don't want.

If by armchair commentary you mean 14 years of using and administering every major Linux distro, including a couple years of mostly Ubuntu, and having something to say about it, then you're right on the button.

If you can't figure out how to change the desktop background, you probably should not be using linux of any kind... :)

What a genius comment. Zing, whammo, ur so funneh haha

hah. Same to you :)

This decision is the perfect illustration of why Ubuntu has a very long way to go before becoming a true user-friendly operating system. You just cannot remove the feature this way. You just can't. This forces dramatically new user experience down every user's throat. Honestly, how many people do you think will like being treated this way?

If you want to remove the feature, deprecate it. Turn it off in all new installs, leave it as is in all upgrades, and optionally notify the user about the deprecation.

I agree that we need to change the current UI paradigms. I am not a fan of "if its not broken, don't fix it".

However, I've written a small applet that uses notifications and I don't quite understand how things will change when what he describes happens. I have not upgraded it because the notification mechanism in karmic sucks.(I will update the code when I finally install Lucid)

Does anybody know? Will applets disappear completely?

I can't really wrap my mind around the idea that adding more menus, probably multiple hierarchies of menus, is a better solution here. I wish they had included a mock-up because I can't really visualize how this would work. Are people going to dig into menus to find stuff? It seems like both Windows & OSX are moving towards less menu digging -- not more.

I understand the need to upset people by forcing change in order to make improvements in the long run. However, I will still list my annoyances, although I'm prepared to reserve final judgement for now.

I'm forever getting mixed up between the status menu and the messaging menu. The only thing I really use them for is Empathy, which now seems to be split between two menus. And to bring up the contact list it now takes two clicks (or more often three because I clicked the wrong menu) instead of one.

I don't like to just minimize Empathy or Rhythmbox since that gets in the way of my "working set". As a programmer working on distributed systems I often have perhaps up to a dozen windows in a workspace in each of eight workspaces. "System" stuff gets in the way of my window management. I'd like this stuff to be somewhere distinctly different.

> No competent designer, sitting down to design an operating system from scratch, would say to themselves “I know, let’s have two completely inconsistent ways to hide windows”.

I disagree with this. I want two different ways, because they have two different purposes. One is my for the working set: stuff that I'm doing. The other is for stuff that's always there, like a new email notification, an IM contact list, background music and the clock. I don't see a problem here.

I think that a notification bubble worked just fine for notifications. You'd have the icon pointed out to you, and then it would be there in the background using minimal space until you attended to it. That's about as unobtrusive as it gets. Grabbing clicks has never been a problem for me, but not being able to see under it has. I'd prefer to be able to close the notification bubble rather than being forced to wait; being able to click through is pointless if I can't properly see through.

Instead I now have an annoying popup telling me that I should update appearing in a separate window that I have to close only for it to pop up again. I understand that this is intended to be improved, but at the moment it seems to keep demanding attention and interrupting me and then it forces me to wait while it loads.

I suppose that the problem is that as an experienced user I know what the notification icon for "updates available" meant and I want it to stay there as a reminder. However, I only want to be interrupted once. At the moment (on Karmic) I get interrupted repeatedly and the only way to stop this is to remove the notification entirely.

I agree that the UI should be consistent. I would prefer if any suitable application had the capability to either minimize to the system tray or to the task bar based on a user setting with a sensible default. The close button should always close it (or if not, the behaviour should still be consistent across all applications).

I understand that some of my complaints are intended to be addressed already, so I look forward to my experience getting a little better, at least. I thank Matthew for the blog post - at least I know that these things are being worked on.

> No competent designer, sitting down to design an operating system from scratch, would say to themselves “I know, let’s have two completely inconsistent ways to hide windows”.

heh, this quote reminds me of one of the most fustrating things about osx, theres like 4 ways to hide a windows

1. minimise to dock 2. "close" but dont quit 3. hide 4. quit

I think I might be missing one as well, and they all work completely differently when you want to being the windows back.

You can also "roll up" a window, IIRC.

The problem with having different menus for different categories is that it optimizes the use case where a person has Bittorrent, IM, music and 10 different things in the background versus a person who only has one.

The person with only one background service will see unnecessary cluttering with different kinds of menus. Take a look at the forums, a significantly large number of them ask about how to remove the messaging menu (because they dont use it)

In addition, I have a clipboard manager in the system-tray. Which menu does that come under ?

I agree with most of what you said, but I disagree that forcing change to drive long-term improvements is a good idea.

There seems to be a trend to making UIs less configurable and more "consistent" with the app designer's presumptions about how a UI should work. But does this justify sweeping away well-established UI conventions for something new and untested?

If you make the new UI a configurable option, one of three things will happen:

1. People will gradually discover the new UI option, try it out, and if it genuinely does improve their workflow, keep using it in the long term. In this scenario, your improvement will gradually displace the old convention, and you will eventually reach a point at which old convention can be retired.

2. Some people will adopt the new design, but many will continue to prefer the old convention. Here, there is no objective superiority of one solution to the other, and the best direction is to let it remain a configurable option.

3. Adoption of the new feature will be negligible. This might be because the established convention is in fact superior, but it might also be attributable to user simply having entrenched habits. Either way, it implies that the new solution does not have enough of an advantage over the old to overcome what users are accustomed to, and that the established convention is good enough. In this case, you should acknowledge that the change was unnecessary and revert back to the old convention.

This approach allows for UIs to evolve to suit actual usage preferences, and avoids the pitfalls of constantly changing everything in pursuit of an abstract and prescriptive notion of consistency.

Reply on news.ycombinator.com