Shared publicly  - 
A bit of kicking and screaming notwithstanding, I have, in general, made peace with GNOME 3 and gotten on with my work. I had figured that they had bottomed out in the "making things worse" phase and had moved firmly back into the "let's make it actually useful again" phase of their development cycle.

Then I read about their "new approach to GNOME application design" and realized that they don't seem to be done breaking things yet. It is really nice to concede (in an after-publication addition) that "If you want to be able to view more than one window at once you will still be able to do so," but it makes it fairly clear that, whoever those users are that they have in mind, they work very differently than I do. I'm discouraged.
Henry Kleynhans's profile photoAndrew McNabb's profile photoDiego Elio Pettenò's profile photoPaul Jakma's profile photo
I wish I could see test cases behind their decisions. If there is any at all. Every now and then Gnome (and other) developers state how they did a test case or research and majority of those people decided we don't some option.
Unbelievable stuff. Maximizing windows by default? What the hell are they smoking?
I've used GNOME Shell 3.2, and I actually thought that it has a lot of potential. In fact, I'm looking forward to switching to it once my machines have good enough GPUs and rivers to match (I like to have a lot of windows open).

However, it seems like they might have tablet-itis, which has been afflicting all of the desktop vendors for the past few years. I do worry that they've been forgetting that these are primarily used on PCs and workstations.

Still, some of their design seems nice, and I'm still cautiously optimistic. At any rate, Shell's extensibility is still very high value. Indeed, a window tiling extension already exists for Shell. Think of the innovations that are possible with a well-integrated compositing desktop environment with a single extension system, not unlike what several popular web browsers have. Here an idea to whet your palate: controlling window tiles using trackpad gestures. That'd be epic.
"Cinnamon" created by the Linux Mint folks looks very interesting...I have it running in a VM to toy around with.
Speaking as someone who normally works with nine xterms spread over my display, I can say that I am very much outside of their target audience. And yes, I have seen the chop-your-terminal-window-up approach, and no thanks to that either. Sometimes I want to expand a window, for example to look at a larger-than-normal function, in which case overlapping the other windows is what I want to happen. I guess I should ignore Gnome when I finally give up on Unity...
+Paul McKenney gave tiling window managers a shot? Seems to me that working with a lot of terminals is something that kind of software manages really well.
I think it is pretty neat. And if you don't like the full screen by default you do realize we have an extension system to revert the behaviour if needed.

Overall, this shakeup is needed. Some of you just want to keep the old paradigms going and not change. The idea of a desktop project is to evolve with capabilities of the new hardware that gives you. Otherwise, the lowest common denominator is going to be this boring set of components that will only appeal to old linux hackers. If we concede that, the linux desktop will be dead within 10 years as newer users will not be excited by what looks like windows 95 from a bygone era.
+Paul McKenney - ha I think I beat you.. being a sysadmin I end up opening a terminal each task. GNOME doesn't deal with that very well does it? Because terminals are nebulous in terms of what they do. It also doesn't deal well with full screen. Clearly, we'll have to deal with terminals as an exception. (I think music players are an exception too) We are in fact looking at that and figuring out to make that work. A lot of us code at GNOME in the same manner. It is a set development model for Linux eco-system.
+Mladen Mijatov Keeping in mind that I want movable windows initially automatically placed in a tiled fashion, not a tiled straightjacket, yes, software could in principle do what I want. And Gnome used to. But if Gnome's new design paradigm is "one window at a time is enough for anyone!", I cannot realistically foresee a happy relationship with Gnome.
+Sriram Ramkrishna I am glad that I am not the only one who needs to look at several things at the same time! But it is not just Linux -- it is what I absolutely need to do if programming on a large system. Human beings have limited short-term memory, so we need to use display real estate to compensate. And if we are working on something that relates to several pieces of a large system, it only makes sense to be able to look at each of those pieces at the same time. Yes, software development is not the only use case, but it is an important one. To me, anyway! ;-)
Like most people in this tread I am outside the target audience of the new tablet-itis desktop (as +Andrew Clunis put it). According to ps I have 11 urxvt instances running across 7 wmii views (and a few other applications).

But, GNOME 3 and Unity are getting to a point that I wouldn't even give it to my parents to use. The last install I made for my folks got LXDE instead.
+Paul McKenney - Yep, and I think we're just in the initial state of evolving as we attack each use case. I'm particularly excited by the new web browser epiphany known as "Pages".

I agree that a lot of the time we use web browsers as online documentation that we use to refer to something. Certainly that is my use case. On those occasions, I don't want a full screen as I want to see both screens as the same time.

But know that we do hear your concerns. GNOME 3 didn't turn into a complete pile of goo as a lot of people expected. The extensions helped you revert behavior you didn't like and also add behaviors you didn't think you could do before. I expect a lot of interesting things once people get rid of their GNOME 2 fetishes and embrace and extend (in the good way!)
+Paul McKenney , you and me both. I have 4 LCD display at work (3840x2400) with a buttload of Eterms :)
Even on my limited 1920x1200 laptop, I still run 9 Eterms per screen.

I've been upgrading my laptop by doing a straight re-install of debian in the first time in 7 years (moving back from ubuntu), and I'm agonizing between e17, xfce, and lxde (coming from e16).
Gnome is not even going to happen, they lost me a long time ago. We are not their target users, they made that very clear.

For what it's worth, I'm so used to e16 pretty cool bindings and ways to interact with windows that I may only be able to ever be happy with e16 or e17 :)
I guess as long as we find a WM that we're still happy with, good enough for me, but granted that running network-manager without gnome, or even getting acpi/sleep to work can be a bit more work than it should :-/
+Marc MERLIN - things will be quite interesting when wayland comes into mainstream, mostly because the trend of distros being even more integrated with a particular desktop whether it is xfce, gnome, or kde is going to continue. That's my theory anyways.
+Sriram Ramkrishna you may very well be right. I'm not sure how much I'm looking forward to that, but I guess we'll see.
The thing that's the hardest is to sell something to people used to their key bindings. Getting a seasonned vim user to use Emacs or ae (alternate editor for recovery when only /bin is mounted in debian) is just going for a world of hurt.
Trying to get me to use a WM that does not offer what e16 does in window resizing and keys handling is just not going to work. I realize that not all key bindings can be the same, but they'd better be easy to change, or else I'm just not going to be interested in switching.
+Mladen Mijatov I use a tiling window manager. I have two monitors and 8 tiles/desktops (when I had one monitor, I had 15-20 desktops). Each desktop has 4-6 terminal windows, a browser or two (with gobs of tabs), one or two pdf files open, and xchat somewhere. This is how I can code and read docs at the same time, its how I multi-task between coding, testing, and reading email, and, ahem, goofing off, all at the same time. My physical desk is "enhanced" with scraps of paper for the over-flow stuff that doesn't fit on the available screen real-estate. I've been day-dreaming of a 3 or 4 monitor setup for many years now ... Gnome seems to be moving in the exact opposite direction: fewer, smaller screens. So.. as a co-founder of the Gnome foundation, I'm pretty unhappy about that.
+Marc MERLINE - it's quite possible to write extensions to change the behavior of the window manager. There are several that have added tiling features to mutter. Remember sawfish, which allowed to modify the behavior of the wm? You can do the same thing with extensions. The degree of how many mutter operations are available I don't know. But you can do those things. Key bindings of course, we have a mechanism.

For now I think I would recommend staying with E16/E17 until the full extent of the changes are realized. Then put some time into it and try to fix GNOME with extensions. Likely someone will already have come up with the idea and try to implement it.

Given time you can have your cake and eat it too :)
+Sriram Ramkrishna Just curious: with E16, I can do alt+middle button drag to resize the closest side of the window I'm in.
Do gnome key bindings allow that?
Can I double click or middle click on the title bar to roll the window up into a 12 pixel title bar only windows in gnome?
As for your comment that I spend a lot of my personal time I'd rather spend on programming I enjoy :) to maybe get gnome to do what enlightenment has done for me for 12 years or so, that's possible, but
a) I'm not sure I have the current programming knowledge to do this
b) I definitely don't have the time to spend on something that's already working just fine in E :)
c) More generally I have seen gnome farther and farther away from what would be a DM/WM I'd like, because "I'm not the target user", so it would take me more and more time to fix it to do what I already had and still want.

Gnome gets to target whatever users they chose, I find it sad that said target user is so far from the long time linux power user that I am, but as long as linux doesn't more or less force me to use gnome, I don't care too much.
For what it's worth, one of the reasons I did just dump ubuntu was because it was clear that they only tested a few WMs/DMs (and really only one) that they target. e16 does not even work at all in ubuntu for what it's worth (yes, I obviously fixed it back when I was using ubuntu), but that, the continuous network-manager breakages, unity, and the utter nonesense that is plymouth, was enough to make me switch back to debian.
I don't really want to go the same road with gnome, I already know it's unlikely to end well, so I won't waste my time or yours :)
+Linas Vepstas

Interesting. I use full screen terminals for coding and I think I'm maximally productive with them.

I use tabs and muscle memory to have things in the right place. I have up to 30 tabs open, spread out amongst 3-4 terminals.

I used a multi-window workflow in the (admittedly distant) past and found the other windows a distraction.

The most efficient flow is when I only see the thing I'm working on, that is when I'm able to write the best code.

I'm wondering whether you have compared your productivity under the two paradigms.
The work model that is presented by Unity just goes against my style and annoys me to no end in so many different ways. So yeah.
+Ingo - i do a hybrid, I try to keep the number of windows down to a minimum because I simply can't remember where everything is anymore. I do try to use tabs as much as possible. So in the end, my work model since switching has been less windows, more tabs.

I don't do a full screen, since I like using gvim. I could switch to another editor that allowed an execution window but I feel very lost without all the short cuts of vi. I should try to attempt it again.
+Ingo Molnar When doing actual coding, I often do use a single "window", which consists of a piece of paper that I write on with a pen. But it is not raw speed that I am looking for when taking this approach!
Yep, GNOME has admitted that they really just want to be like Macintosh 1.0. No yay. :-( Because people can't multitask, use complex tools, or would ever need to refer to information from two different applications side by side.
not only am I outside the target audience, I find myself unable to even use such a system for even a few minutes. I don't get a headache per se, but a certain amount of cognitive dissonance seems to set in!

Typical example: one of my devs has an issue and wants me to take a look, and if he has this new-fangled stuff I try for 2 minutes then tell him we'll do it on my machine and ask him to come over instead. Happens about once in a month or so.

Worse, this has started to affect my advocacy and ability to support the "F&F" (friends and family -- the non-tech people one helps out all the time) crowd. I find myself unable to work naturally enough to help them in anything that requires me to sit down at their computer.
I am currently on i3 which is a tiling window manager and have at least five named workspaces. In i3 you can switch workspaces independantly on each monitor. Named spaces have assigned windows to them so everything is neatly organizes automatically. :-)

This really helps build muscle memory and increase my productivity.
There is a big potential for a desktop that just feels like Microsoft Windows, so that people who have to use it could go to a Linux system and work there without any distractions. And then they should be able to go back and be productive. There would be still room for innovation. There are desktops that can be made to work like that, but it takes effort and knowledge. And then some subtle thing won't work, like to volume control or the keyboard switcher. Or something would break on upgrade, like GNOME's alacarte did, and there is nobody to fix it. Experimental UI doesn't belong to the default desktops of the popular distros. I'm still waiting when someone would realize it. It worked great for Mozilla Firefox. People can switch between MSIE and Firefox effortlessly both ways, and it's a good thing.
+Pavel Roskin - except that Microsoft and Apple are moving to the same paradigm. Microsoft now requires touch screens for windows 8. So in about 3-4 years, that windows 95 like desktop is no longer going to be interesting. In which case, us desktop guys are looking at a shrinking demographic. New users are certainly not going to want to go to what looks like an older OS. Plus, you're implicitly saying that OSS is not capable of writing a decent GUI. Is it then your position then that we should just copy whatever our competitors are doing? In a way, that is what we are doing, but we're ahead of them. We are already building a desktop that is going to take advantage of hardware trends today hopefully before windows 8 comes out and when apple comes out with their OSX/IOS hybrid.

The goal is to get free software to as many people as possible. Because, at least with GNOME unlike your traditional desktop you have the source you can modify it to how you want, we have extensions to change the behavior as needed. That also means that we need to get ahead of what our competitors are doing.

Anyway, I think I'll stop arguing for now. In the scheme of things, only GNOME apps are going to be seeing this behavior, you're not going to see gvim, emacs, GIMP having this behavior, it's only going to be certain apps that want to participate in the GNOME 3 way. So let's not be alarmist here. How many of you are going to be using Pages/Epiphany instead of Firefox or gedit instead of $EDITOR?
+Jonathan Corbet please don't post this kinda stuff for one to see first thing Monday morning, it ruins my entire week! This is seriously depressing :(

Once again it shows that the people doing the GNOME design have completely lost touch with their core users! They clearly are smoking something illegal. How is one support to work on code if you cannot have a (or a couple) of terminals next to your Emacs window(s)? This is utterly insane! Besides, code is written in 80 long lines, making the window full screen would just waste the bulk of the space.

I am still on GNOME 2 since Fedora 15+ still don't like my laptop, but since this idiocy is being forced upon us with the updates I was expecting to 'upgrade' eventually - hoping that the madness would have settled by then :(
+Sriram Ramkrishna where does Windows 8 require touch screens? Touch screens are yet another gimmic that are going nowhere in the desktop space, nobody wants to sit with their greasy fingers and run them over the screen where they are really trying to do work.

The problem is that this idiocy will ruin the GNOME apps, so mail clients etc will be rendered unusable. How am I going to compare a code snippet I received in an email against what I have in my editor or my terminal window, when the mail client is forced over the full display?

You guys keep talking about the new users, but you're clearly not realizing that the reason why most windows users don't like the new version of windows is that they are uncomfortable with the pointless moving everything around. This lets throw things up in the air every 24 months only caters to the 15 year old k3w1 k1d8 who are more interested in clicking on facebook than doing anything useful with the computers. For real users change needs to provide an improvement, and GNOME 3 (and Windows 7+ for most users) have proven to be nothing but the opposite!
Trying to recreate a tablet experience on a desktop pc is a dumb idea. There's a reason you use tablets and desktop pc's for different things. Tsk tsk tsk....
+Lars Bjerregaard exactly! I makes zero sense, and while the GNOME crowd clearly hasn't realized it, the Linux tablet desktop is called Android. For some reason I think it has been announced every year since 1998 or so that this is going to be the year of the Linux desktop .... fact of the matter is it isn't. The Linux desktop market is for technical users and technical users dislike being treated like this.
I don't mind if Unity is a choice for some people's desktop. But there was a reason we developed the previous desktop model. Canonical,GNOME and cohorts should let us continue to have that choice.
+Sriram Ramkrishna You seem to be basing all your arguments based on the premise that it's impossible to make progress without breaking things. I already showed to you why that argument is flawed.

The Linux kernel is orders of magnitude more complicated, and they manage to move forward without breaking things. Or at least, they acknowledge it's possible, and they try their best, mostly successfully.

Here you can see some of the most prominent Linux hackers talking about precisely that:

ELCE 2011: kernel panel on the importance of users

It seems GNOME has given up on the idea that stability is possible, and it's funny that's the main reason many people were using GNOME in the first place.
I agree with the tabletitis comments. It seems Gnome developers are focusing on that to the detriment of other things. Strange, since I imagine many developers use desktop systems with second monitors of at least 24" and maybe 30" size. I wouldn't give up my 30" monitor for anything. It is far too large and has too many pixels to run full screen applications. Tiling works great though.
Note that the applications themselves have to request to start maximized the first time [1] and if you unmaximize it the app will remember that for next time you start it.

1. So Inkscape can keep on starting in a useless and odd size made out of 50% chrome and not even displaying the whole document.
+Ingo Molnar Well, most of the time, I'm not writing code; I'm almost always debugging or measuring it or reading it. To read code, I usually read 2-3-4 files at the same time -- a few .c's .h's, in order to follow the flow. Often a pdf or website, if I'm coding in a language I don't know well (or library...) When fixing or measuring code, I usually have one/two windows for editing, one for watching a compile in progress, one for watching test results. If test results are graphed, then 1-2 graphs, and a window edit/control the graphing program.

If a job takes more than 6-12 hours to run, I usually launch 2--6 jobs, monitoring each in a different window & all the various subsystems, to make sure nothing crashed/hung. Often/usually, I have to log into a variety of different servers to boot/run these jobs, and so multiple ssh logins for each server.

I'm often working on 2-4 inter-related code bases at the same time -- maybe not the same day, but the same week. So, e.g. when I was doing hexagon code, I was hacking gcc, glibc, and kernel more or less at the same time; one desktop each, and a 4th desktop to boot and test. Any other short-term tasks (sysadmin, etc.) got their own desktop, so as not to disturb the workflow. Sometimes, I might not come back to a desktop for a few weeks; it was nice to see things where I left them; it made getting back in the groove much easier.

This is the combinatorial explosion of windows. The thought of doing the above with a few maximized windows boggles my mind.
Add a comment...