Hacker News Reader

Comment by thristian | Hacker News Reader

I remember reading a computer-magazine columnist back in the mid-90s talking about how Windows 95's interface was too revolutionary and different, and that we should expect Microsoft to have a parallel "Windows Classic" release with the same superior technical underpinnings as Windows 95, but with the traditional Windows interface that their millions of users were familiar with.

Obviously that didn't happen, and I doubt that this "GNOME 3 with a GNOME 2 interface" will get very far either. GNOME 3 is pretty great for most people's workflow, and if you're particularly attached to some feature that's not present, you could try the GNOME 3 2D fallback mode (which looks pretty much just like GNOME 2), or XFCE, or LXDE. If those aren't conservative enough for you, there's still the various *box window managers, WindowMaker, pekwm, and dozens of others.

View on HN

I'd definitely disagree with that. While I'm sure there are some people who do like it, I think Gnome Shell is objectively flawed usability-wise. Just to list a couple obvious examples:

-The top bar is almost entirely wasted space; -Having the bar at the top violates Fitt's law access to a window's close buttons; -Again for Fitt's law, the top-right corner is wasted because that menu is rarely accessed by most, and there's nothing in the bottom-left corner; -The messaging notifier thing frequently fails to let me know there are new messages, the accordion effect makes it hard to click individual notifications, and the popup notifications take an obscene amount of space at the bottom of the screen and often overlap with something I'm reading; -There's no way for me to tell at a glance what programs I'm running, so I often forget that I already have a window with work open and my workspaces end up even more cluttered and disorganized.

I could go on but that's just off the top of my head. All of those are objectively bad design decisions when you approach G-S from a desktop use case. Clearly they're trying to shoehorn a touch UI into a desktop UI, but they're different use cases and when an unskilled design team tries to smash them together, you get the worst of both worlds.

I don't think people dislike G-S because it's revolutionary; I think people dislike it because it's genuinely--and perhaps irreversibly given the silly politics Gnome plays--flawed.

The Mint guys seem smart, so if they feel that a fork is necessary then it's probably the only way to do what they want given the constraints. They've shown a good track record of putting quality solutions out, so if they release Cinnamon at some point I'm sure it'll continue to be developed.

Edit: I also find it a bit funny that G-S was developed at all, because with a little customization you can get extremely similar functionality in Gnome Panel. Just install AWN, enable the Scale plugin in Compiz with a hot corner at the upper-left, move your panel to the top, and set a gconf setting to remove minimize and close buttons. There--you've just re-created 80% of G-S functionality and without having to spend years in development.

Interesting. I've never seen Fitt's law invoked to cover close buttons before. Note that serious users (who are the ones complaining the hardest about gnome-shell) never touch the close button: closing a window is a keyboard shortcut (generally Ctrl-w, though sometimes Alt-F4 if the app doesn't handle that) in my world. For a while I was running a custom metacity theme with no window controls or title bar at all, actually.

And I don't follow the "no at a glance view" of your work space argument. Hit the windows button and Ctrl-Alt-Up/Down and you can actually see all your windows, with no overlap. That's something that was never possible before, and I love it.

Obviously there's a lot of taste at work. Gnome 3 is a big change, and big changes break workflows and piss off experienced users. But I honestly think a lot of your "objectively flawed" argumentation is just nonsense. Smart people can like gnome-shell and use it very effectively. You just don't like it.

The point is that those corners are some of the most important screen real estate, taken advantage by other DEs. Take Windows, for example. Upper-right=close, upper-left=context menu, bottom-left=start menu, bottom-right=show desktop.

Compare to Gnome Shell: Upper-right=settings/account menu (rarely used IMHO), upper-left=activities, bottom-right=missed notifications, bottom-left=nothing.

Where Windows puts something useful in all 4 corners, G-S only has something useful in 2. Obviously one can argue about what is and what isn't useful, but from a purely objective viewpoint, one can definitely say that Gnome-Shell doesn't take full advantage of those extremely important corners.

As for my other points, for example the top bar: how can you deny that it's mostly wasted space? It's 80% unclickable black bar. That's not "nonsense", that's just how it is. Other DE's put something there: the window menu bar, an activities list, etc.

True that I can get an activities list by pressing super. But that's not "at a glance." That's "at a keypress, then a visual scan, then a click to go back to what I was doing." Most other DE's provide some sort of window overview at an immediate glance, ie., an always-visible taskbar.

I think my other points also stand. I don't doubt that some people do use G-S effectively, but that doesn't mean it isn't deeply flawed. If it weren't, then more people would like it.

Upper right in gnome-shell is where the shutdown and suspend options are. I use them every day. And the upper-left in windows (by which you really mean a maximized windows app -- windows puts nothing there, and I only occasionally have anything maximized) is the "context menu", which is near useless, being duplicated functionality with the existing window buttons. Note that Chrome doesn't use a context menu button, for example. I think arguing that windows puts something "useful" there is disingenuous. Windows merely puts "something" there.

I guess the top bar argument is valid, though note that there are a bunch of gnome-shell extensions that like to use the space. And the significantly thicker windows panel is likewise largely a waste of space (just a big bunch of icons duplicating what in gnome-shell would be the overview).

Looks: you hate gnome-shell. Good for you. You're not alone, lots of people do. I don't begrudge you your opinions, I'm just calling you out for trying to turn them into "facts" when they clearly aren't.

But you can't really compare the windows UI with gnome-shell, because they don't have the same workflow. For example, a "show desktop" action in gnome-shell is useless, because there is no good reason to interact with the desktop (it doesn't show icons by default).

Now, tell me, under Windows, how many times a day do you use a window's context menu, against, for example, the Start Menu? How many times do you expect a regular user to click on them? Furthermore, how many times can you activate a context menu from that position? Do you always use your windows maximized? You shouldn't optimize the corners of your screen for a windows button (because there might not be a window there). INSERT: the corners in gnome-shell always work the same way, independently of any window state.

Having something in every corner of the screen doesn't mean that it is universally useful.

Seems you missed all the design documents made for GNOME 3. Top bar is simplistic at this moment on purpose. There are more ideas behind it. However, just because you can put stuff somewhere, doesn't mean it is a good idea to do so.

The idea is to minimize distractions. Not to optimize for mouse clicks, hot corners, etc.

Defining your own criteria as the reason to call it deeply flawed it a bit easy IMO.

I'm judging it as it stands now, not as some document somewhere says it might be one day, maybe, if the developers get around to it. If they don't want people judging it, they shouldn't have released it until they felt it was ready to be judged.

I'm also judging it by criteria used by many UI experts. How DEs approach Fitt's Law is an important metric. The law exists for a reason, because UI experts found it came up again and again in real-life testing.

Defining your own criteria by saying "works for me, wontfix" is also a bit easy.

> The Mint guys seem smart, so if they feel that a fork is necessary then it's probably the only way to do what they want given the constraints.

These are the constraints for him, as far as I know (looking at the gnome-shell code):

1) extensions can't modify the original layout of the shell, because they are loaded after the shell elements are created. You can monkey-patch the shell, but only so much.

2) There is no easy way of destroying the panel (or other ui objects).

1) happens to be true because of the extensions constraint that they should have the ability to be turned on/off. That means that their entry point in the shell gets to be after the main ui objects are created. You can add new behaviors to them, and override them, except for their initialization (obviously).

The extensions system doesn't expose an API that allows for multiple entry points (for example, having extensions derive from an Extension class with Base and Restartless subclasses). This might be a design oversight, but for the most part it is sane, since it creates a base of code one can expect to be there, that any extension could depend on (again, simplifying inter-extension compatibility). Extensions written for gnome-shell won't be compatible with extensions written for cinnamon, which is extremely bad in my opinion. It means the developer community that's been growing around the platform might break into multiple directions, which can hurt users.

If there was a way of destroying the panel (or other main ui objects), though, there would be no need at all for this fork.

"Silly politics Gnome plays"

Care to explain?

Note: I'm a long time GNOME release-team member, GNOME sysadmin, GNOME bugmaster.

Essentially the entire "damn the torpedos, we're right and there's no going back now" approach Gnome seems to be taking. I've seen lots of people submit valid and thoughtful criticisms of Gnome Shell, and the only response Gnome seems interested in giving is "we have our grand vision and if you aren't submitting a finished patch that we may or may not accept then fuck off."

Maybe "politics" wasn't the best possible word choice but the idea is ego over engineering: a few guys at the top are making these decisions and more or less putting their fingers in their ears. At least that's the impression I've been getting.

We gave loads of test images before the release of GNOME 3.0. The "no going back", we've been making loads of changes based on feedback. I've listed some of these changes in the 3.2 release notes.

Wording like "fuck off" is completely unacceptable; I either warn or ban people (everyone, random person, developer, etc) for that.

"A few guys at the top": That is not how it works. If you're respected, people listen to you. That is how it works within GNOME.

Saying "valid and thoughtful criticism": valid maybe for you. But GNOME isn't about supporting every different idea. If an idea completely goes against the thought behind it (quite simply: focus on one task at a time), then yeah, use something else. Having an idea and going for that, to some that comes across as arrogant. But simply stated, GNOME is not right for everyone. That is what is meant with this.

But at the same time, there is an extension system in place and loads of code is in javascript, loads of stuff can be done via gobject introspection (meaning: access libraries easily via javascript).

GNOME 3.0 was the initial release. Further improved with 3.2, more design ideas will be implemented in 3.4, etc. Not saying 3.0 is bad (IMO, it was great), but eh.. there is a goal and it hasn't been achieved.

I've been around during GNOME 2.0, a lot of the feedback was the same. I think your impression is off.

Is there any supporting evidence for GNOME 3 being great for most people? My experience with it was so bad that I can't really imagine that being true. And it's much easier to find negative than positive comments about it on the net. Clearly the dislike isn't quite universal, but my impression is that it's the most hated "upgrade" ever.

Also, recommending the fallback mode as a solution is disingenous. At least performance doesn't suck as badly as with the real shell, and it's not totally alien to existing users. But it still has a number of totally ridiculous and gratuitous differences to both GNOME 2 and every existing UI I've ever seen. For example how could the way panels are configured make sense to anybody?

And if that's not enough, the word is that the fallback mode is going to get killed once performance without hardware acceleration is good enough. (And the bar for that has to be pretty damn low).

i hated gnome3 at first. i even stayed on gnome2 for quite a while to avoid the upgrade.

as it seemed inevitable, i bit the bullet and upgraded. linux is my only OS so i use gnome3 all the time. i have grown quite fond of it. what's best is that it hides so many of the distracting ui elements (either on purpose or simply because so many widgets just don't work) that used to pull me away from my work that i feel i'm actually more productive now. plus i love that my windows key launches the lens (is it called that?).

i think gnome3 is much better than unity.

That's exactly my experience too. I used unity on a netbook for a while and just couldn't make it work -- too many edge cases (even to the extent of getting my preferred side-by-side terminal and emacs to work cleanly). I stuck with it for a while, but never really came around.

I similarly tried gnome-shell when Fedora 15 was released and hated it, but stuck with it for the same reasons. And it paid off. I'm a convert now -- I don't like everything (app-based Alt-Tab will never work for someone with 12 terminals running, I really do want nautilus icons on my desktop, etc...) but the things I don't I know how to tune.

I'm honestly more productive now than I was on gnome2. It's a great environment, and still getting better.

not to mention the fact that i'm now hardwired to go to the left corner or "windows key" to do things, which is something i really like doing (and do intuitively now). i really like that one can just press the windows key and start typing.

regarding showing nautilus style icons on the desktop, you'll want to get the gnome-tweak-tool and then go to Desktop and enable the "have file manager handle the desktop" option. You'll have to restart, logout, or restart the desktop manager to see the change, but this should do it for you.

Just checking: when alt-tabbing, you know you can pause on the Terminal icon (or hit the down arrow) and switch between all your terminal windows that way, right? Granted, it's not as convenient as having all your terminals available from one keypress, but it's better than having to switch to the mouse.

Also, if you want Nautilus icons on the desktop, that option's available in gnome-tweak-tools (although I just found an extension that puts a menu of Nautilus bookmarks in the menu-bar).

Yes, I'm aware of that feature. And no, it's not good enough. I want to bounce between my shell and my emacs without having to deal with the 11 other shells and 3 other emacs windows present. The idea of app-based navigation is that you'll stay in that "app". I don't, my workspaces are collections of windows.

But the traditional navigation is still present (though still a little buggy -- the MRU order sometimes gets confused, and sometimes the graphical popup of each window gets the wrong location), it's just bound to a different key and can be fixed.

Maybe I'm in the minority, but I must have been using gnome the same way as the gnome developers, as the gnome3 interface basically gives me a setup I enjoy without the added configuration. I really like the dynamic virtual desktops, and the expose-style overview. Though I haven't written any myself, the framework for writing plugins looks nice as well.

Like the Windows anecdote, we have to account for the inherent confirmation bias. People who find the new environment adequate don't go online to rave about it and counter the negative opinions. The users giving the impression that this is "the most hated 'upgrade' ever", may still be the minority overall.

I think what you say about users who like/don't mind G3 not talking about it has some merit, at least as far as I can tell from personal experience.

We are a Linux shop and we are about 60/30/10% split between Linux, Windows 7, and OSX. Fedora is what most of us are using now, but we have some Ubuntu. Everyone that uses G3 at work likes/loves it. I personally switched off Ubuntu to Fedora because F16 have newer G3 than Ubuntu with PPAs had.

I guess I'm also in the minority in that I think Unity is decent, especially with all these new vertically-challenges 16:9 screens. This problem becomes much worse on netbooks which typically have only 600px vertical, 768 if you're lucky. In a day and age where we have stuff like the Galaxy Nexus and retina displays, why are desktop and laptop screens so far behind in pickle density? Alas, an argument for another time, but I say "long live 16:10". I love unity on my HP netbook (which also has a magnesium chassis and a 1366x768 screen and TWO mouse buttons.)

Going back to Gnome2-style WM feels like a big downgrade to me. I hate the ask bar. I also really like the "alt-back tick" to switch between different windows of the same app in G3. That said, it needs an option to auto-hide the top bar and PLEASE LET ME HIDE THE ACCESSIBILITY ICON. I know there is a package in F16 to do the latter, but installing a package to hide an icon is idiotic.

Since the true innovation of GNOME Shell isn't mentioned here anywhere the discussion is completely off. Surely the Shell is new and of course nobody was able to customize it on day one to his old habits. Even all these blog entries with 50-things-I-do-after-installing-XXXX did not help. The result was frustration.

GNOME 2 is customizable using entry points common to anyone familiar with *nix, there were scripts, CSS files, dialogs, options, configurations and more. All this thinking failed after installing GNOME Shell. You may call it offending. However UNITY showing 3D buttons looking like made with a pshop plugin from end of the last century is not an option.

So how do you customize GNOME Shell? Here is the new entry point: open firefox and goto to https://extensions.gnome.org/ Yes, you change your desktop environment from within your browser. Add, try out or disable shell plugins from your browser.

And the best thing all is powered by JavaScript. The language running on every computer having a browser is used to customize your DE. So stop whining and enjoy the desktop revolution. Getting curiuos? Find the location of a extension and open it in your favorite IDE, change a string variable, save it, press ALt-F2, enter 'r' press enter and voila it is there.

There is even a debugging tool, (Alt-F2, 'lg', enter). Now go and implement whatever you were missing in GNOME 2.

It is time to see how users would build their perfect desktop interface.

Huh, so extensions.gnome.org works now. The last few times I've looked for it, I've gotten a "permission denied" or other "no-such-website-here-yet" error. I only knew it existed because I happened across the "Gnome Shell Integration" plugin in Firefox's plugin list.

For what it's worth, when I first tried out GNOME 3, I customised it the same way I customised GNOME 2: gconf-editor (and a little later, dconf-editor).

While not a seperate product, you could run the old program manager and file manager from Windows 3.x by running 'progman' and 'winfile'.

> I doubt that this "GNOME 3 with a GNOME 2 interface" will get very far either

It's already the default in the current release of Linux Mint, they've just been hamstrung by the limitations of extensions in fully realizing their vision.

Reply on news.ycombinator.com