Shared publicly  - 
I don't mind these GNOME forks so much.  But I would hope that they would take as mall portion of the developers and invest it in GNOME so since they have some dependencies on the internal components like GTK+.

Also, once you reach feature parity with GNOME 2, where will they go from there?  Will they just keep it in steady state or will they be essentially done?

Finally, what would be the attraction to new users, younger users who have grown up on touch interfaces like tablets and cell phones and are essentially presented with a soup'd up windows 95 look alike?  What about when laptops especially windows 8 enabled laptops have touch screens?  Will this design help them?  How can this interface work with these new hardware capabilities?

We aren't fucking around with design and changing interface just because.  Hardware is changing, new capabilities are required and sometimes a certain design has reached it's shelf life.
Keith Hill's profile photoDan Mashal's profile photoChetan Chauhan's profile photoSriram Ramkrishna (sri)'s profile photo
So why are you removing features then? Is that also because hardware capabilities are changing, so with new hardware changing font sizes etc no longer makes sense? Or what?

People don't want gnome 2 because it was a pinnacle of some GUI design. They want it because gnome 3 removed features they used, and made it harder to get their work done.

First the "shutdown/reboot" entry disappeared, now apparently the "log out" entry has been removed according to +David Miller. In the meantime, fundamental features like just starting a new terminal are still apparently not part of standard gnome 3.

+Sriram Ramkrishna, you're not usually in denial. But I wish some gnome3 people actually asked themselves why people do these forks to begin with. And no, it has absolutely nothing to do with "new hardware support". 
What's crazy to me is that the whole flap could have been avoided just with a few default extensions -- menu & taskbar alone would have gone a long way. It was a big shift, and it wasn't clear why.

Honestly, even though I'm now totally used to Gnome Shell and use it every day, I still don't understand why it's superior to the Gnome 2 model.
+Jason Brooks: bingo. The frippery extension alone makes gnome3 much more useful. That's one single extension. But because it's against how you're "supposed" to use gnome3, apparently it can't be merged as an option. 

+Sriram Ramkrishna, seriously. It's not "gnome 3" that people hate. It's the "ve know better, and ve know vhat you vant, and ve vill make you like it" part. 
+Ray Strode you're missing the point.
It's not "legacy mode" that people want.
It's "make this useful please" mode that people want.
+Ray Strode: can you get rid of the bottom panel? Does Exposé still work? (Bonus points if you can change it to be upper-right corner instead).

"Classic mode" in gnome3 has traditionally been a (very bad) joke. The frippery extension set has actually done a much better job, and it's only better because it's a "pick the parts you like" thing, not a whole new mode.

The whole "let people do what they want" really is important. I only enable the "Favorites in panel" part of frippery, for example, because I think the gnome3 "search for apps" mode works better than the application menu did.

But "let people do what they want" really seems to be against the whole gnome religion. I hope I'm wrong.
+Linus Torvalds Most of the people use extensions instead of forking, at least for the features you've mentioned. Extensions can be considered a sort of "soft fork", you modify the part you don't like, but take advantage of the development and enhancement in the other parts.

There are users who like vanilla GNOME 3, I am one of them. GNOME 3 help me a lot to focus in the task I'm doing (terminal, email, browsing and some office tasks from time to time ).

For those users who still like the traditional desktop, GNOME will make life easier with the new classic mode and they will support a set of extensions (at least a preview of this will be bundle with GNOME 3.8).
+Arjan van de Ven I was replying to Jason's very specific point about a menu & taskbar.  I realize some people have different reasons for disliking gnome 3. And I'm sure you realize a lot of people find it useful as it is today.  +Linus Torvalds I believe the various components are separate extensions combined together, so it should be possible to disable parts I think. The difference between this and fippery is mainly that this is supported by the same people who work on the shell, and that the extensions come bundled together in a special shell mode.  even in the classic mode the hot corner is still active to get to the expose-like overview.
Btw, there hasn't really been a "classic mode" before 3.8. there was a "fallback mode" that was a last ditch effort to get something on the screen if the user's video card wasn't up to snuff.  These days software gl is good enough that the fallback mode isn't as important, so it's getting dropped. We discussed this at he last GNOME summit, though, and decided just dropping fallback mode wasn't good enough, since some people weren't using it for fallback but for a closer to classic experience. that was the motivation for doing the supported classic mode extensions.
+Ray Strode even people I know that first liked it, are increasingly puzzled by the "improvements" in later versions.
Yes they are developers, so maybe not GNOME's target audience.
(after all the GNOME project has more than enough developers working on it so why would they care about developers ;-) )
+Jason Brooks see my share of Mattthias McClasen's 3.8 progress post.   You'll see that we have a classic mode that has all those things from GNOME 2 sans some minor things.  So we are listening and we have brought some of that back.

+Linus Torvalds didn't we go through this already, once before? :-)  I mean you have the ability to make the changes you want to make the desktop behave the way you want it.  There are some problems.  I launch terminals using alt-t because I can set some key bindings to do that.  It works great.  You can either use windows key + "ter" and you'll launch a termainl, that's giong to be a lot faster than even going to the short cut bar on the desktop itself.  I mean all the elements to making the desktop you want are in there.

What you're saying doesn't invalidate what I posted.  Hardware is going to bring in new capabilities.  Also, new users are going to expect things like multi-touch as part of the default experience.  What we had with GNOME 2 is not going to get us where we wanted.  Forget about the presentation, but underneath we moved to using popular, standard technologies and a method to be able to create something new, easily.

OK, yes, we took some features away because it didn't quite fit in the default design or didn't behave properly as well, or we felt it was superfluous.  Personally, I think we got it wrong on Nautilus because I took the trouble to read through a lot of reactions to that and realized that what we removed was used a lot.  I still advocate bringing those back, and maybe I can initiate a thoughtful debate about it.

That pinnacle set of features that you liked about GNOME 2 didn't come off the bat, it came from a combination of trial and error (cuz you know this is a volunteer project., we don't get a big budget to do GUI design, we have to do it the same way those fancy GUI designers figured it out, trial and error), research, and listening to the community.  That is still a winning combination that we are doing now.

Look, I'm well aware that we have an image problem.  Just look at the range of articles that haver been published.  It was clear we weren't communicating with the public on what we were doing as well as we could have..  We are trying to be better.  We had a long thread on foundation list about it.  The GNOME community isn't a single entity with one opinion, there are a lot of people in there with various opinions on where we should go and how we should do it.

I know that sometimes we come off as a bunch of ego-tistic folks, and I sincerely apologize for that.  Our community has our share of individuals who could communicate better.  But sometimes man, it's too much.  The vast number of emo posts, pettiness, hatred that I wade through in a week can really get to you.  More than most I go through all of those, and I respond to them as politely as possible.  So it's quite understandable when others lack tolerance or patience to react politely.  They've gone through this before and it's not pretty and so yes, its hard to engage or communicate what we are trying to do.  I can't speak to what the end state is going to be, but I will find it hard to believe that we won't be in a steady state that will at least make most people happy.  If not, well.. the forks will help.  I just hope that they will also give back to GNOME as well.
+Arjan van de Ven There's no question that some people don't like GNOME 3. I won't argue that. Everyone has their own reasons for choosing to use what they use.  I can tell you, anecdotally, that I know a lot of developers who do like GNOME 3 :-)  Though, of course, everyone has their own pet bugs and workflow issues they don't like. On the whole, I'd say it's improving. But, of course, not everyone is going to agree or disagree with me. To each their own, I guess.
+Sriram Ramkrishna: you're continuing to make excuses.

What part of "ve know vhat you vant, and you vill do it like we tell you to do it" did you not understand as being a problem?

The whole "set up a key combination to get a terminal by editing the preferences" is a perfect example. I use keyboards for writing, but I want that icon for adding a terminal.

Why cannot you just understand that people have different preferences? Why do you answer my complaints with "just do it the way I do it", when I just told you that that was the whole problem of the gnome3 mentality?

Do you really not see what the problem is? Even when you repeat it in your own "just do it this way" instead?

The extension model is wonderful, but it doesn't work, because the extensions are not part of the standard install. We did indeed already talk about why it doesn't work, you seemed to even agree with that.
And btw, +Sriram Ramkrishna, don't bother telling me about "mouse up to the corner to get favourites, right-click on the terminal icon, then select 'new window'".

Anybody who thinks that that is a good way to open a terminal (as opposed to "click on the terminal icon") is a f*cking moron. Don't tell me that this is all a "better user experience". 
+Linus Torvalds in the classic mode coming in gnome 3.8, the dash behavior switches to click-icon-spawns-new-terminal
+Sriram Ramkrishna My point is that if classic mode extensions had been available from the outset, these forks likely wouldn't have happened.  But I'm glad that these extensions will be available "out of the box" soon. 
+Ray Strode lovely. That actually sounds like gnome3 may be usable right out of the box.

And gnome3 has been improving. I complain, and I think extensions were badly done in some ways, and I think a lot of gnome people are in denial. But I'm actually back to gnome3 because with the right extensions it is more pleasant. 

Frippery + gnome-tweak-tool really make a huge difference. I don't understand why gnome3 hasn't just merged them.
+Bastien Nocera: that's a particularly stupid thing to say.

Here, let me explain it like you're mentally deficient, using mostly your own words:

   "Let's optimize for what your users want to do"

Doesn't that sound like a good thing? Can you understand that when I put it in (almost exactly) your own words?
+Linus Torvalds oh man, no I'm not advocating doing that way at all.  Yeah, they would be moron if you did that way.  I just have a keyboard shortcut to pull a terminal, or I said hit the windows key, type "te" and then return.  That's way way faster than doing that series of steps.  You should only do that ONCE when don't know the name of the app or something.  But if you know what you want, just follow the above steps.

So, what's stopping you right now from adding a terminal to short cut bar with the extensions that are currently available?

We seem to disagree on throwing all this stuff as part of the default experience instead of throwing it with extensions.  I know extensions are a little fucked up right now and we are trying to make that a better experience.  Some of that is an infrastructure thing, my opinion is that we need to make extension writers part of the release process.  That means, that we should have images ready for extension writers to port their software way way before the release happens, preferably before GUI freeze.  That should alleviate some parts of it.  Probably need to look at how Firefox manages it with their plugins.

As for the whole "you will do what we want, the way we want to", it's hard for me to address that.  The initial feedback was negative, but I have seen many haters turn into lovers after using the interface after awhile, going back to GNOME 2 was going back to a colorless world for them.  Maybe it isn't such a bad thing to ask people to at least try it and get used to using something with a new system to see if it works.  Perhaps as an experiment you should try to see if you can use the default experience for a month and see if it's better or not?
Maybe the biggest gap is a distro who customizes gnome3 to be by default "developer friendly"
+Jason Brooks Note +Finnbarr Murphy 's extensions were around from near the beginning. GNOME certainly could be better at messaging. Some, ike Sri, Allan, and Matthias are trying to improve that.  In some ways, though, big change is going to cause disruption and disruption is going to cause discontent.  We all have good intentions. it's just a big and hard problem.
FWIW I'd like to see panel-favorites extension in the set of classic-extensions (for the people who would like a non-activities mouse driven way to launch a new terminal instance). This cycle I might add an entry to tweak-tool to make it easier for people to set up the terminal shortcut. Not sure, I hope to get to FOSDEM to drink about it.
+Arjan van de Ven yeah, that would be pretty cool.. Colin Walters (who you should talk to BTW, he has some ideas of applications that seem similar to yours) was working on a build system that could do that.  The whole GNOME OS while somewhat misnamed, is about creating a reference image, but it could also be used to push out "developer friendly" version of GNOME 3.
GNOME certainly has an image problem, especially in regards to listening. We're quite good at telling people what we're doing, our main problem is explaining the why

Personally, what I believe we need is a forum where users can come and voice their opinions/thoughts/concerns/etc. As a result, I teamed up with Alex over at worldofgnome and launched (unofficial) forums - forums that I dream will someday become official. If you have thoughts/concerns/suggestions for GNOME I urge you to come check us out, make an account and start posting.

I for one would love to hear everyones ideas on how we can make GNOME better - in turn I can pass those ideas on to our developers & desiginers who I chat with on a regular basis. Perhaps someday if the forums thrive, they can even become official and we can convince current, past, and future GNOME designers, developers, etc to come post! 
+Sriram Ramkrishna: nothing is stopping me right now. I have the frippery "favorites in panel" extension, and it does the right thing. Really. Try it.

So why is that not available as a standard thing? Why does it break any time I do an upgrade? Why is it so hard to find (try to find it without knowing what you're already looking for on 

Same goes for font handling. There's an option for the huge fonts, but that's useless to anybody with good eye vision. You do realize that font size is important, don't you? Different people have different monitors and all that? Yes, I can get gnome-tweak-tool, but it's not in settings, so unless I already know to search for "tweak", there's no sane way to do that?

This is basic functionality. Why hide it? Why make it such a horrible user experience? Why - after several years - can't gnome people just admit to messing up in some areas?
+Sriram Ramkrishna Colin and I synced at last plumbers. What he's doing is kind of different ;-)

I'll be lauching my hobby OS hopefully in the next weeks... if I end up using GNOME3 as default GUI I'll have to make sure it's customized right ;-)
+Ray Strode it's not easy.. let me tell you.  Changing into a different direction also means fighting forces that want you keep with status quo.  Whether that means a new GUI design or opening up FOSS to women, those forces are at work and they are EMO.  People want Linux to be an exclusive club and some of the complaints about making GNOME user friendly is a direct attack on that.  I've learned a lot of things about people who run and use Linux by engaging with people across the spectrum.
+Sriram Ramkrishna I think a big part of the rub isn't so much the exclusive factor, but the perception that the gnome guys think "user friendly" and "developer friendly" are incompatible. 
While clearly those two are not exclusive (just look at an Apple OS), the perception that the gnome guys think this (not saying they actually do) seems to act like poison on many 'discussions'.
+Arjan van de Ven actually that sounds like a lot of fun, I would be interested in seeing the outcome.. might even help :-)

+Linus Torvalds See my previous comment about infrastructure and bringing extension writers into the GNOME  release process.  I think that is the right solution.  Put them on equal footing with regular development and we can resolve some of the issues with extensions breaking after reach release.  A lot of the breakage is likely just the version number needs updating in the json file.  But I agree with you on extensions, and more importantly Jasper and others do as well.  It's a problem that being looked at and hopefully addressed.  I will advocate making it part of the release process.

Regarding the font stuff, I'm clueless on that.  File a bug, and let me know I'll ask +Behdad Esfahbod , he deals with the font rendering stuff.  Have you tried looking at the "Universal Access Settings"?  In general, font should look good out of the box regardless of monitor etc.  I don't know why it was designed that way, it's worked for me.  I put in bugs for the things that bug me, and for the most part they get fixed.
+Ray Strode Totally. I do like Gnome Shell, and I use it now, and I don't think the forking was terrible or anything.

I really missed the bottom panel -- clicking there was my main way to cycle between apps, and to access my virtual desktops. I tried all sorts of extensions right at the beginning. I thought they looked ugly, and prettiness is a big part of why I switched from KDE long ago, so I focused on getting used to the changes.

I'm used to it now, and while I don't get how it's better, it may be prettier to have more junk pushed to the background. I think it may be driving me further into the browser -- with the tab bar as my new app panel, I think I use fewer native apps. I think I use my virtual desktops less, too.
+Arjan van de Ven well, the people I talk to don't know gnome folks from any other.  They lose credibility with me when they start arguing about GNOME making linux attractive to the general public.  Now of course, if you're talking about our kernel and other closely related communities of software professionals, well yes, I understand the argument.  But as you remarked about having a developer focused distro with the right cocktail of supported extensions that benefits software developer, I think that would be pretty awesome.
+Sriram Ramkrishna my plan is to have the UI up and running before the end of this weekend. Have not quite decided between xfce, mate or gnome3... right now the decision will depend on which one I can get to work and customize the easiest.
mate for sure is not all that well polished.
If GNOME offered its own APT repository and used /etc/apt/sources.list.d/gnome.list rather the same queue as Debian's distros then, that would be one big plus. That would mean more statically linked libraries rather than everything linked to system shared libraries, which would be one more big plus. I won't essay the reasons.... +Sriram Ramkrishna 
I hate the way OSX launches apps in general.. I never know what's running in general. Overall, in OSX things feel very crowded.  I also made teh mistake of making OSX act like GNOME 3 with the launchpad or whatever it is on the left top corner, of course it sort of acts like GNOME 3 but doesn't and then I confuse myself.  It's one of those things where ironically I'm fighting doing everything the OSX way because I want to do it the GNOME 3 way. :-)
+Arjan van de Ven if you use GNOME 3 I'll totally try it out.  Mate isn't a great choice.  Mate is dead already, and the other GNOME 3 based forks that are implementing GNOME 2 are going to work much better I think.  You should try Cinnamon?

XFCE also lacks polish.  It's a good DE, if you don't particularly care about DE experience and just want it to do whatever minimal set of things. I used to use it a lot when I wanted a DE to run under VNC.  But that whole use of GTK2 is just wrong.  I hope they will move to using GTK3 soon.  Looks like they have a plan here:
+Sudhir Khanger it's not a technical problem, it's a process problem.  There are probably a number of ways to solve it, but it requires working with extension writers.  Other solutions could be to take the top 15 extensions and test it before release to make sure that they work etc etc.
Sigh. Same ol', same ol'. I'm glad folks like +Linus Torvalds are still willing to repeat their concerns, I just wish someone would actually respond to them directly. It seems anytime someone says "why can't I add icons to the panel" there is some lengthy discussion about extensions, design philosophy, etc. Extensions just don't work, each update has proven that.

How about scanning the most used extensions and pulling them into the main release? Do we have numbers on that? The Frippery bits in particular would address a lot of the complaints. I don't personally like them mind you, I use hot keys for common tasks and the dock for the less common tasks, but enough people do that it should be considered.

Dropping font configuration is just ridiculous.
+Darren Hart As was mentioned earlier, 3.8 will have classic mode which will be a set of extensions managed by GNOME that are a lot like the fippery extensions.
The ICE process defaults to the DM's environment, so that meant more elaborate schemes for extensions, action icons, and icon bars. Otherwise, motif already solved it over localized UDP/TCP in X-Stations. +Darren Hart 
Yes, and I'm "cautiously optimistic". However, I don't believe people want "classic mode", they want the shell, and they want to be able to configure it. I haven't seen 3.8, so I will wait before passing judgement. However, my concern is that like with all the other attempts at such a mode with GNOME 3, it will be a shallow implementation that looks like stock GNOME 2, but has none of the configurability. Lipstick.

For example, the blog on this mode states that the clock is moved to where it was in GNOME 2. That demonstrates that the author completely missed the point. Nobody cares where GNOME2 distributions put the clock by default. They care about being able to put the clock wherever the hell they want. Middle, right, bottom panel, etc. GNOME 2 let them do that. Will GNOME 3.8 classic mode?

I for one really like the basic gnome shell experience. I imagine there will be aspects of classic mode that I would like. Will I have to choose between the stock experience and the classic mode?
+Sriram Ramkrishna What makes you say MATE is dead already? It is actively being developed. I can appreciate your opinion that gnome3 is great (it is, at some things) but please don't say some other project is dead because you disagree with they way they are doing things. Some of us aren't planning on upgrading (or don't even have the possibility to upgrade - ARM) our hardware.
+Steev Klimaszewski it's dead because 1) it uses gtk2 it's hard to maintain as it is 2) you can implement gnome 2 as one project already has using gnome 3 components.  I'm not trying to disparage Mate developers too much.  But it is a colossal effort, and you can extend gnome 3 + gnome-panel enough to replicate most of it.

It would be nice to have the mate developers brought into our community and work on the classic experience actually.  That would be a better effort than maintaining GNOME 2.
It may be "hard" to maintain but at least my same theme works when moving from mate 1.2 to 1.4 to 1.5 whereas my theme from gnome 3.2 broke in the 3.4 upgrade, followed by a break in the 3.6 upgrade, followed by a break in the 3.6.1 upgrade. (And no, i really don't like the adwaita theme. I use Linux on a multitude of machines, with resolutions ranging from 1920x1080, down to 1024x600 - even at 1080p the titlebars seem way oversized.) It may be a "colossal" effort in your opinion, but that is what the MATE developers want to work on. It might be nice to have them working on gnome3 but I appreciate a desktop that doesn't require JavaScript and jumping through hoops every release to make my desktop my own. I like when my desktop works after every upgrade, instead of having to find or make a new theme because the theme developer finally got disgusted enough with having to fix or completely redo their theme and just gave up on it.
And I thought XFCE decides not to go with gtk3 due to the increased memory usage. I haven't looked recently, but I know in the past that was the case.
Open source is great because you can scratch your own itch, just realize that not everyone's itch is caused by the same thing. 
Good on classic mode, then. My problem with Gnome Extensions is that they seem hard to install on a disconnected (for security reasons) system. The install-from-web hack is clever, but there are so very, very many use cases with Linux that facile cleverness doesn't always get the job done.
Actually the JavaScript bit is a rather moot point since nowadays polkit requires it for parsing it's rules (lolwut?)
+Steev Klimaszewski the thing is, they have no good story on apps. You might have the desktop OK but the apps will be static. Also you assume that issues with fonts and theming won't be fixed. They will be. It is a better to join with GNOME community work on classic and have a voice as someone doing the work rather than what they are doing now.
+Jonathan Abbey What would you like GNOME Shell Extensions to be? I'm in charge of the website, and I can make the proper changes if you give me a workflow that you'd want.
Ah, just so they could be downloaded easily and copied to another system for installation. We have internal mirrors of the relevant rpm/deb repositories, and it would be maximally nice if the Gnome Shell extension system could use something like that. 
The thing is +Sriram Ramkrishna people did try to have a voice before, why will they suddenly have one now? Are different people in charge of the projects? And I'm not entirely sure of what you mean by the apps are static - some of us don't need (well let's be honest, can't use) the fancy graphics and such. That it's taken from 3.0 to 3.8 before really addressing the issue with "classic" mode doesn't sit well, and fallback mode was absolutely laughable. It almost looked like it was designed to force you to not want to use it. So others found a place where they were wanted and accepted for wanting to be able to customize their systems. Not ostracized or made to feel like they didn't belong because they didn't see the graphical genius of others.
Are we not allowed to change our minds? The motivation for Classic Mode was dropping Fallback Mode -- if we do that, we need something to replace it.
Sure but don't expect people to come running back. You had your chance and blew it :) 
+Steev Klimaszewski You had a voice even from the very beginning, but we did not listen to those who just wanted to revert things back to what it was and ignore all our hard work.  We deserve as much consideration as GNOME does to its users.  Our labor should be respected as well.  I've written many bug reports that have been worked on, and they received the attention they did because I worked within the confines the design that we worked on.

When I mean by static apps,I mean that app writers are going to move on to GTK3/GNOME 3 libraries and that the apps that support GNOME 2 are going to fewer as time goes by.  The apps using gnome 3 are not going to look well integrated with GTK2.  That's why I say that time is against them.

The graphics and animation in GNOME 3 is fairly minimal.  There is animations are used for state transitions.  For instance, going from desktop to overview, or switching desktops.  We aren't doing anything on the level of compiz.  You should be able to use graphics card of 5 years back which is an eternity in computing.  At some point, if you want the desktop to move forward, you have to cede the lower end to projects like XFCE while you focus on modern hardware, trying to support every configuration from 10 years ago is not only difficult for GNOME, but difficult for distros as well.  That's a lot of hardware configurations to support.  Hanging on to old hardware of course has it's own consequences.  I do not know your circumstances that you must use hardware 6 years or older,  but the "fancy graphics" as you describe are all there to support usability, and are minimal as well as appropriate.
Fallback mode was dropped precisely because it wasn't good enough for our users and we wanted something better for people who require a  GNOME 2 stye workflow. It's your choice; you can use whatever you want and nobody is going to judge you for it. I don't think that any of the designers believes themselves to be a genius. This is a process that involves a lot of hard work on the part of a lot of people, and a lot of iteration. We aren't going around flaming people for doing what they want. We are too busy and it's not what we are about :)
+Steev Klimaszewski Things in life aren't usually so cut and dry.  You might decide not to use a product like say Apple mostly due to moral actions, but if the product is good you'll likely come back.  It's when one is perceived as immoral or dishonest that it's likely that people will not come back because of distrust.  If we are guilty of anything it would be that we have a poor external image.

If you take anything from this discussion, you should understand that we mean well.
+Sriram Ramkrishna

I hope this is not taken as unwelcome bashing but as constructive criticism. I personally still use Gnome3 because I feel it is the Linux desktop environment closest to the "Holy Grail", in terms of code distance.

I think the main "philosophical/social" component the Gnome3 development process gets wrong (or misunderstands) is decentralization and cherishing dissent.

Compare "extensions" to "Android apps". Which of the two is treated as a first grade citizen, never breaking across releases?

Expressing "dissent" via an extension adds value, constructively. If a platform starts cherishing such "dissent" and makes sure that such extensions/apps work out of the box in the next release, it takes off like a rocket.

Not a single large OSS project has gotten this right, with the possible exception of the Linux kernel where we treat binary compatibility as nonnegotiable.

Firefox almost got their extension model right - then started breaking them and the project fizzled (compared to what it could have become - it's by no means a failure).

Yes I know these projects are different, and that compatibility with decentralized "apps" needs care and is sometimes pain, that ABI compatibility sometimes complicates progress, but it's still an essential mechanism to express trust and gratitude to "apps" that came to our platform. It also guarantees that if we mess up some functionality (or if there's reasonable differences in taste), a decentralized, easily available solution is possible.

Breaking apps/extensions across releases, passive-aggressively centralizing is the primary platform killer/throttler in the OSS space.
This thread gives me a lot of hope for the future of GNOME. I've been tempted to switch but not anymore.
+Ingo Molnar: Extensions can change anything, there is no limit to what they can change, so there is also no ABI/API to safeguard. Any change in gnome-shell might break an extension. It takes some time to figure everything out, but what you said is known (read it before by the people who maintain the extension stuff).

Not aware of application breakage though.
By the way, one of the worst mistake made was by scholars that published Einstein's ramblings and that made people understand it, "specially." (i.e. "Swartz")
You don't speak for everybody, +Jasper St. Pierre, so your imagination is quite big beyond that you sounded like "A"-typical reverse engineer cyberbully that tries to alienate someone into suicide.
+Sriram Ramkrishna I find my self +1-ing nearly all of +Linus Torvalds's comments; he's described very clearly how most of us feel about GNOME 3. GNOME 2 was not amazing or sexy or overly desirable. But it was usable; it could be pushed and pulled and made to work as necessary. GNOME 3 lacks that flexibility. Things are missing. Things are difficult. It can't be customized in any meaningful way. It gets in the way. It inhibits productivity.

Developers don't undertake a fork and rewrite of a project as big as GNOME simply for fun and glory; it's big, it's complex, it's no fun. So the user first petitions the the core team to see reason. But the GNOME team notoriously dismisses these suggestions as whining. Finally, the sheer frustration of not being able to do their job drives them to what any developer would consider drastic measures and they fork a whole new desktop environment. No one wants that, but it's the only option left.
+Jonathan Ballard : also no clue what you're talking about. Clear that you're trying to troll and you seem upset. There was no reason to be so aggressive. Suggest going to a conference, hopefully in person you'll behave differently. FOSDEM is coming up for instance...
+Tyler Larson I think there is two sides to that.  Perhaps they are heroic developers trying to save their work flow, perhaps not.  Let's take a look at SolusOS's recent fork.  For their distro they forked a number of components in order to provide a GNOME 2 experience on GNOME 3 platform. 

Vincent Untz posted on planet calling them out on this:

There was no need to fork gnome-panel, they could have worked with GNOME and help provide classic experience on GNOME 3.  Adding new developers interested in providing the classic experience would have been advantageous to us.  We could have brought in fresh blood, they would have had a voice in GNOME as well.  I think Vincent did more than warranted, and yet those components were forked.  I'm going to trust Vincent on this and not the reasoning documented on their website.

Maybe you feel it's some kind of community outcry, I don't know.  These things are not easy to quantify and there are a lot of shrill people on the internet.  Happy people don't complain and we can't quantify happy people.

I will concede that perhaps we have not put our best foot forward and we could have communicated better.  I'm happy to keep rehashing and re-apologizing, if you like? :-)  If you want to read the threads regarding community management, feel free:

Alan Cox
+Sriram Ramkrishna people have tried working with Gnome and have now taken the logical step beyond. The real questionis can all the forks unify around one common working base which is controlled by pragmatists not fundamentists
+Sriram Ramkrishna: I think you may be a bit too optimistic about the whole "App writers will transition to gtk3" thing.

In +Subsurface, we're actually looking at the pain of moving away from gtk entirely, largely because gtk incompatibilities make gtk fairly unpalatable.

I'd actually like to use gtk3, because we use cairo for the actual "graphical" part (as opposed to the widgets), and gtk3 seems to have better cairo integration.

So why is gtk3 actually making us look at moving away entirely, despite ostensibly being "better"?

The reason is simple: compatibility. As an app writer, the last thing I want to do is waste my time changing the internal interfaces, only to actually lose compatibility. Yet that seems to be the gnome way. There is no sane way for us to write the app so that it works both with gtk2 (still used by most distributions, and both Windows and OSX builds) and with gtk3 (to look more modern on modern distros). It's an "all or nothing" thing. and the gtk people seem to have made it that way very much on purpose to help "encourage" people to upgrade.

As a result, we're actually looking at doing more work - changing the UI toolkit entirely - in order to not have to deal with gtk being incompatible with itself. This is despite having picked gtk as the interface originally, and despite the fact that we've gotten somewhat used to the gtk interfaces and have to learn Qt entirely from scratch.

Now, it's by no means a done deal, but it's seriously being talked about. And really, this is a project where gtk3 would be a good fit, since we already use cairo instead of gtkdrawable etc. But because we can't just leave gtk2 behind entirely (cross-platform and not interested in just dropping support for old distros), it's simply not an option.

I really don't understand gnome developers. Even within gtk2 there were all these totally unnecessary "let's drop support for old interfaces even if they are clearly just shim layers on top of new ones" which makes things painful when you want to have the same sources work across versions without unnecessary # ifdef'fery. And gtk3 seems to have been done explicitly with the "you have to rewrite your app" model.

For a library, that kind of "you can't even recompile, you have to rewrite things incompatibly" is basically a huge turn-off. And it really isn't just us. Look at all the non-gnome apps that simply aren't converting (for example, chromium and firefox both have the exact same issue: even if they might otherwise want to port, they can't just leave gtk2 behind, and with no sane way to do it while supporting both, it's a huge problem).
+Alan Cox I gave you an counter-example where they could have worked and we would have been happy, the gnome-panel maintainership was up for grabs and was offered.  So it's not always GNOME's fault.
+Linus Torvalds  We take the blame for that fully. GTK+3 builds and runs well on Windows and OS X, but we haven't released official binaries for several reasons. I've been on-and-off trying to make our developer experience on Windows suck much less, but it's a very off-and-on thing, and there's lots of roadblocks along the way.

You can try using some unofficial builds of GTK+3 for Windows from places like

There's going to be a GNOME Developers hackfest starting on Tuesday, and one of the things I plan to work on there is getting Windows and OS X support to be first-class citizens, where we have official binaries and official installers, and all our documentation and tooling works well across all three platforms. I think Colin really wants to approach that goal as well, where the output of OSTree will include Windows and OS X binaries for GTK+3 as well.
+Sriram Ramkrishna, I'd like to point out that +Subsurface really has a pretty small gtk interface footprint, so porting to gtk3 should be like half an hour of somebody's time. It's really not a huge project to begin with, and partly because neither +Dirk Hohndel nor me actually knew much gtk programming to begin with, we don't use hugely crazy interfaces.

So the porting part should be trivial. I could do it, send a patch to Dirk, and just be done any time.

But porting to gtk3 and still being able to compile with gtk2 doesn't look nearly as easy. Because we only use a small subset, we could probably add our own compat headers etc, and have just a handful of # ifdef's, and it wouldn't be totally horrible.

But unlike the "maybe half an hour" thing, it's much more work, much more testing, much more pain. And all this for basically no gain to us.  It's not like any of it actually improves the application itself, it just makes it harder to maintain. See why we don't want to do it?

+Jasper St. Pierre: so right now, the fact that Windows and OSX is not sanely supportable with gtk3 at all means that it's a total non-option to just do a one-way upgrade.

But even in the Linux world, the one-way upgrade doesn't work. There are people on Debian who don't just want gtk2, they want gtk-2.24, while we have been using things things from gtk-2.26 (ok, so some debian versions would like 2.20, but nobody really cares about people who stick to stuff quite that old).

So seriously - doing a one-way gtk3 upgrade is simply not reasonable. Not for a few years. 
+Jasper St. Pierre: just put another way: what we might find convenient (and perhaps other projects too) would be some kind of "here is a single compat header that people could use that introduces basic gtk3 interfaces and has wrappers for building under gtk2".

Then we could just take that header, include that instead of the current <gtk/gtk.h>, and things would work.

IOW, what we're looking for is the equivalent of the whole 

  # include <gdk/gdkkeysyms-compat.h>

kind of thing, which we already use for the 2.22 breakage. It wouldn't need to cover the whole gtk mess, but the basic "how to get from a         gtk_drawing_area to cairo", and all the basic widget changes. Things that should be trivial if you know gtk internals, but are just stupid and annoying differences that you can't really care about if you don't.
That's certainly an interesting idea, but the compatibility layer would be a lot more complex than keysyms-compat, since it's more than simple defines. We can't guarantee that all interfaces will continue to work in the new design, or maybe if they do work, they'll be slow. I'll certainly talk with other people about it.

The way you phrased it makes me curious about the specifics. Do you want something where after you port to GTK+3 and cairo, you add some header, and poof, it compiles with GTK+2? Obviously, you can't use any of the new widgets or things like that in GTK+3, without somebody backporting the widgets to GTK+2.
+Jasper St. Pierre: exactly. I'd love to port to gtk3 (and would be willing to limit my interfaces a bit), but have a way to still compile the end result for gtk2 for legacy platforms.

And no, I'd not need to use actual new widgets, and obviously I'm not actually doing that (since the project is coming from gtk2 in the first place). And I'm willing to modify the source code for this all, but I am not willing to do so if it means that I lose all the old gtk2 platforms.

Think of it in the sense of the "gtk2 to gtk3 migration guide":

where I'm prepared to do the "preparation" style steps, but am not willing to do any flag-day switch-over and leave gtk2 entirely behind.

(Side note: we have not even done any of the preparation, since there hasn't been any point. We could do more of that, and add the whole "GSEAL_ENABLE" thing etc. But without a way to stay with gtk2, it's kind of pointless to even prepare for gtk3, if you see what I mean)
+Jasper St. Pierre: one more comment: we really have nobody who really knows gtk at all in the project. So I just tried doing GSEAL_ENABLE, and I get tons of errors, and they are probably all trivially fixable, but when a "real" gtk developer sees an error like

   error: ‘GtkToggleButton’ has no member named ‘active’

he probably spends 2 seconds to fix it because he knows what the issue is, while for us it's a "uhhuh, need to google this".

Really, the whole project was done with devhelp and google. I'm not suggesting gtk people should actually look at the sources, because you might lose your lunch, but I am suggesting that it's probably how a lot of people and projects work. 

For some gtk developer, some small detail where the interface changed may seem like a tiny detail ("You shouldn't have done it that way to begin with, and it was always better to do Xyz instead"). But to an app writer that doesn't really care about gtk itself, but only cares about the end result, those small incompatibilities are surprisingly big deals.

So even the "stupid" interfaces are worth keeping around, because they may seem stupid and trivial to you, but to the app developer, not keeping them around is a big annoyance, and he/she/it doesn't care or understand why you think they are stupid. You know the library, you know what the internals are, but that's you. The app developer is different.
+Linus Torvalds This sort of feedback is exactly what we want. Since we live and breathe GTK+, it's difficult to put ourselves in an outsider's shoes. GSEAL_ENABLE is prepping you to make sure that you don't directly access public structures, to make sure that we don't break API/ABI in the future when adding new properties.

In this case, you probably had some code that did toggle_button->active that needs to be changed to gtk_toggle_button_get_active (toggle_button);

Want me to write some more words about this in the migration docs?
+Jasper St. Pierre: yes. So in this case, for example, you get absolutely no help from googling. The migration doc says "Use accessor functions instead of direct access", and I understand why it would say that, and what it means, but at the same time there's no way to find which accessor function to use.

So imagine somebody who isn't familiar with the internals seeing that error, and what should he do? He's just stumped.

IOW, I had no idea how to get from that error to the "Ahaa! I should obviously use gtk_toggle_button_get_active()", because googling the error message didn't find anything. It sounds entirely logical once you have that piece of information, but before you even know what the accessor functions are, it's not so logical ;)

Even just a few examples (so that you see the pattern of what the accessor function names are) would have made it obvious.

Just as an example, what's the accessor function for "widget->allocation.width"? Apparently I now have to do "gtk_widget_get_allocation()" (which doesn't return the allocation, instead I have to have an allocation and pass the address to it), and then I can do "allocation.width". It's a trivial change if you know that. But figuring it out if you don't have the background just makes you go "Grrr!".
+Jasper St. Pierre: that's fine.

But look at the example I have now: I have a GtkDrawingArea (that I connect the cairo surface to - remember, I need to work on gtk2). I used to do "widget->window" on it, and it JustWorked(tm). Now I'm looking at the gtk2 documentation for it all in devhelp, and what does it ask me to do? It talks about using widget->window and widget->allocation.width. Where do I find the accessor functions? They are totally undocumented, afaik.

It's not "gtk_drawing_area_get_window()", I tried that already.  Where are the accessor functions documented, when they aren't even documented for the types. 

Yes, I eventually found it in the GSEAL portion, and apparently I should have just used gtk_widget_get_window(), but that took a while to find. And then the same thing for a GtkBin 'child' etc etc. Some of them are easily guessable, but then when they aren't you have to walk the object hierarchy and search. Sure, I can do that, but I think you should try to document that whole "how to find the accessor function" thing.
Maybe I should talk about the rule more clearly. If you have a GtkWidget *widget; and you did widget->window, the accessor function is gtk_widget_get_window (widget). If you have a GtkBin *bin; and you did bin->child, you do gtk_bin_get_child (bin);

That should hold for most cases I can think of.
+Jasper St. Pierre: Yeah, and then it fails for gtk_window_allow_shrink(). Which doesn't exist, even though the field existed.  Because you're supposed to just mark the window resizeable instead. Fair enough, but again, that's not trivial to find when you start out with just accessing allow_shrink.

Anyway, I think it took me slightly more than half an hour to convert the (rather few) places that needed converting, and now I can compile with GSEAL_ENABLE. I'll send the patch to +Dirk Hohndel, but as an app developer, I wonder how worthwhile this was to the app. I think it was more worthwhile as an exercise to try to explain the pain to you guys than it necessarily was to +Subsurface itself.
It seems that allow_shrink was folded as part of gtk_window_set_policy. I don't know the history behind it.
Uhhuh. Trying to then compile or gtk3, I get "GtkComboBoxEntry" not existing as a type at all. And all our gtk_[hv]box_new() usage causes annoying warnings, with no sane way to shut them up (no, we are not using gtk_box_new(), since that doesn't exist in gtk2!)

And no gtk_label_set() or gtk_object_set(), or indeed GtkObject at all. We only needed it to set the "alignment" attribute on a renderer, there's probably some better way to do that, but again, getting alignment done on renderers is somewhat esoteric to start with if you don't know gtk, this is scary code. And I say that as one of the people who wrote the dang thing.

I remember this from the last time I looked at it, but I did it again just to give the details here. This is simply NotFun(tm) for very little gain.
+Jasper St. Pierre: that gtk_[hv]box_new() warning is a prime example of a warning that should not exist. You clearly have the wrapper function, and it's clearly trivial to maintain for you guys. There is clearly zero downside for just keeping it. What's the upside to ever deprecating that interface at all? None.

So all it does it make it painful to have code that should compile on either gtk2 or gtk3. So the gnome developers have actually done extra work for themselves to generate this warning, for actual negative value. 

So what's the advantage to code that is only there to actively drive away developers that would want to use your library? Seriously, that's all that deprecation annotation does. 
it should be noted that for gtk3, structures were made opaque precisely so they could be extended in a binary compatible way.  Also, sealing macros and accessor functions were added to gtk2 precisely so app developers would have a migration path. gtk3 is parallel installable with gtk2 for similar reasons.  In hindsight the structures should have always been opaque, but gtk3 was an opportunity to fix past mistakes.  Of course, having that kind of disruption isn't free for app developers, and that's unfortunate. It puts us in a way better footing going forward, though. And there has been efforts from various people (like Cosimo, Jakub, Jack Gandy, etc) to make gtk2 "fit in" as nicely as possible for the transition period.
+Linus Torvalds They're not just simple wrapper methods. They are separate subclasses. The rationale for removing GtkHBox/GtkVBox as separate classes is detailed in

Unfortunately, we can't switch the wrapper methods to return a GtkBox with a set orientation, as that would break compatibility with code that uses GTK_IS_HBOX / GTK_IS_VBOX. We considered adding new wrapper methods, but at the same time, it doesn't buy us much, as one can always use gtk_box_new.

The move to an orientation property helps us out for places where we want to change orientations on the fly. Lots of apps may want to do this, and we can't say "oh, object of type GtkHBox, please become a GtkVBox".

If you want to disable deprecation warnings, you can use -UGTK_DISABLE_DEPRECATED in your CFLAGS.

Things are a lot more nuanced and subtle than they may first appear. I'd love to put these rationales somewhere, rather than being lost to the G+ sands of time.

I'm working on gtk_object_set / g_object_set documentation for the migration guide right now.
I solved above canons before with object-oriented environments, and I don't mean stupid hyped two dimensional flat file "class" representations. I mean good old immersion, direction, component, noun, action, verb, natural language recognition as the superset of shell interfaces. Of course, it is not inherently friendly to those that demand magically clear writing.

We now have avatar-oriented environments and primitives to piece together, but there is no expert compiler suite outside its game wrapper porn.

Forget above, try to design RESTful UML components and it becomes quite self explanatory how stupid APIs have become.
the section "Compiling GTK+ Applications" in devhelp has some good info on how to tweak which deprecation warnings show up.  It's a little more flexible in gtk3 than gtk2 though, so it's worth reading both pages to see how things have evolved.
+Ray Strode: I understand the value of the abstraction of the accessor functions quite well. I have no problem with that part.

It's the "no actual idea of how to do it if you're an app developer, because the migration documentation tells you nothing, and the actual gtk2 documentation actually uses the non-wapped direct accesses and doesn't even mention the wrappers".

+Jasper St. Pierre: that's a particularly stupid and bad example, and whoever wrote that bug report is a moron.

It's trivial to wrap. Just do this:

   # define GtkVBox GtkBox
   # define GtkHbox GtkBox
   # define gtk_hbox_new(x,y) gtk_box_new_legacy(GTK_ORIENTATION_HORIZONTAL, x,y)
   # define gtk_vbox_new(x,y) ... same for GTK_ORIENTATION_VERTICAL 

and let the f*cking user just think his VBox vs Hbox code matters.

Because it does matter on gtk2. In gtk3, it's one unified type, and who cares?

See? As it is, now the app writer would have to do the above wrapping. You're making it harder - for no good reason - for applications to not care about the gtk version. 

In fact, right now gtk3 has actually done more work than necessary, and GtkHBox and GtkVBox are real types, and the wrappers are more involved than the above. Fine, that's good, and actually more complete than I think I would care about at an application level. Thank you.

But then you added the stupid warning. So now all that work somebody did to make it compatible is basically unusable, because that warning will pretty much guarantee that anybody who wants to use the compat helpers will just be annoyed as f*ck.

It's not one unified on GTK+3. The deprecations were added after GTK+3 came out, so we have to support the old types and the old system.

If you deal with compatibility concerns, you should probably add -UGTK_DISABLE_DEPRECATED your CFLAGS, which will disable the deprecated warnings.
If It is not intuitive, comprehensive, and coherent to any significant degree than it is deprecated and will be, pragmatically, obsoleted. There is nothing to understand.
My experience with young users is that they use touch devices when consuming content and use keyboard and mouse when creating content. I teach in a pretty rural area but I have heard no buzz about touch interfaces / win 8, etc. from my kids.

As for touch on laptops, maybe some of the kids will get them but I doubt the school will this decade. And while I will agree with people who say that students are attached to their phone and are text addicts, I think it is a mistake to think of young people as a monoculture.

If you ever wanted student data I could do a short survey, etc. I'm sure other teachers would do the same thing. I don't know what data you have about students but what you are saying just doesn't ring true with my interactions with young people.
+Eric Fitton Sure, Eric, maybe we can do a survey or something regarding this.   As you say, young people aren't a mono-culture so your experiences with kids could be different than others.  But it's good to see data points that challenge my perception of how these things play out.
Meanwhile, KDE continues respecting its user base and becomes more stable every new version.
+Sriram Ramkrishna Actually, the +XFCE  GTK3 porting is not really complete right now. And it boils down to a ressource issue. A CPU/memory issue ( but also a developer time issue. Look at the number of people in the core team. Right now that means something like less than ten people like +Nick Schermer or +Jannis Pohlmann.

There's noise about +XFCE  from time to time, for example when +Linus Torvalds announces he has switched to +XFCE or +Debian default desktop is changed to +XFCE (which never really happened, by the way). But besides this noise on the internet, nothing really changes on the developer front, there's still not much people giving time, fixing bug and adding features.
When I think about the OP and GNOME3 in general I think of xkcd #927. .  All I see is that Gnome tried to take Apple's approach of "You will do things this way and like it", which seems to work for Apple but it has not worked at all for Gnome.  The forks and public outcry is un-ingnorable evidence that Gnome needs to change direction a little.
The extensions should at the very least have a change of title until some changes to the core are made.  How can they be called 'extensions' if they are integral to the usability of the desktop experience or if they simply enable the customizations that users demand?
IMO you can't do the whole "Try it our way, press Win key + type 'ter' + enter to get a terminal. No really its awesome try it!..." thing in a half cocked way.  You either need to take the totalitarian approach like Apple and actively hinder customization or be completely indifferent. 
It seems to me that Gnome tries to walk and guide users down a given path, then completely changes path and then wonders why there's no one behind them when they turn around. This is apparent in both Gnome Shell and GTK.
I'm still waiting for a GUI that is actually fun to customize rather than a hindrance.. I'm using gnome shell at the moment, and love my setup, but I definitely wouldn't say it was fun to get here.
I think we can fix this. +Linus Torvalds just has to wear a wig and call himself "Grandma Torvalds", and instead of ignoring his (and other developers') user experience, the Gnome guys will try and improve that, instead of improving whatever they're improving for non-linus-grandmas.

[EDIT] To clarify: Linus is a sweet old grandma, and (using official Gnome nomenclature) she's just far too old and stupid to realise that she has to press so many buttons to open... her baking app. She would much rather just have an icon, or the option of having an icon. Even if she figured out that she could go to but the extension she uses keeps breaking every upgrade!

Also the new GTK interface has changed, and gran's just far too old and set in her silly ways to understand how to handle this change. Won't someone in the Gnome Community help poor grandma Linus?
+Linus Torvalds, I think this thread has changed direction, but to open a new terminal window, you could middle click on the icon. Personally I like the default behaviour but that might just be me.
+Pedro Araújo and by more stable you mean it survives more than one login before destroying all your settings and displaying the white screen of death? It's been a few months since I tried, but it seems to always end the same.
Reading these posts while using Cinnamon on Linux Mint.

I still laugh at the fact that the minimise window button got removed from Gnome 3, what a joke.
+Sriram Ramkrishna

Just three new Gnome annoyances I ran into yesterday, installing a brand new system:


it is not possible to turn Bluetooth off permanently. If I turn it off in Gnome system settings it gets re-enabled after the next login. Trying to remove the bluez package is impossible, as it removes gnome-shell as well. Don't tell me to file a bug: Google for "Gnome3 remove Bluetooth", it's been reported nonstop since 2011.

(Why do I need to disable Bluetooth? Due to security, also because it's never used, and because an enabled transmitter apparently drains the batteries of a BT heart rate monitor stored within BT range of the system.)


Multi-monitor support is not working adequately: with 4 gnome-terminals on 2 workspaces (one on each monitor - mixed resolution), the gnome-terminal focus is lost when switching between workspaces - it switches randomly between monitors as I switch workspaces via ctrl-alt-down/up. I'd expect focus to be preserved in the original workspace, regardless of what is done in the other workspace.


I had a little script that I wanted to run from the gnome menu or just simply run it from the file manager. Gnome would only offer to open it via "gedit", "Other Applications" had no "shell" offered. Running it from a terminal worked. After 15 minutes of trying to figure out the "Gnome way" I gave up and googled a gnome menu-editor. The settings only took effect after a full log out which was destructive - 15 apps were open and placed intelligently. Oh, there was no logout option so I rebooted the system. For a new menu item ...
+Linus Torvalds  first of all, sorry for my English )
BUT. I think the main problem isn't that G3 not useful from the box.
As for me, the problem is that people dont know how to make it prettier. It takes hours of googling to find out how to make G3 to do what you need.

No tutorials, no guides. Just few screeonshots on the main page and useless features description.

That is a real pain :)
+Linus Torvalds The Dash-to-Dock extension helps make GNOME Shell easier to use 'cause it turns its dash into a hideaway dock, which makes it easier to open apps than going into the Activities screen, and then the Applications screen in order to pull up GIMP or whatever, a hideaway dock just takes away some clicking and makes things easier on the user.

Also, making the Trash, Removable Media, and Computer icons show up on the desktop, and installing the Workspace Indicator makes it easier to switch between workspaces.

In addition, installing the Quicklists extension enables the Quicklists in GNOME Shell, which makes it easier on the user by being able to right-click on the app icon and select whatever action he needs to accomplish.

Played with GNOME Shell on a GX280 Optiplex running Lubuntu 12.10 at school, it's really not that bad an interface if you play with it enough and install the right extensions. A little on the heavy side, but so are Unity and Cinnamon so [shrugs].

KDE, as nice as it is, is like the Hummer H2 of the desktop environment universe, it's so heavy on resources.
+Linus Torvalds You should try Linux Mint and Cinnamon.  Brings back all the needed features and look and feel of the traditional desktop but builds on new technologies that look better.
+Ingo Molnar
For your bluetooth problem, maybe you should try adding a bug to the Gnome bug tracker? I didn't find any by searching google.
Really, I can't understand people that use this fuc*ing gnome-shell.
Totally nosense! Devs have removed ALL customization and don't want them. why?!?!
The shell it's ugly, uncomfortable, with a lots of "click + click + click" and even with the shortcut it's painful.
The menu it's terrible, panel too. And the extentions are not the cure at all of this.
The shell must be usable and the extentions must be "the sauce".
Gnome3 it's not more better that the shell. See nautilus.
And not, Cinnamon it's not better than gnome-shell.
If you start from bad idea and something ugly and wrong, this will always be ugly and wrong, even with makeup and flowers.

I've dropped gnome-shell to use Unity. It's not more better, I hate this all-in-one menu, but it's very much more usable than gnome-shell, and I can have my favorite icons on the dock without click  to show it.
And now I'm very happy that Unity will adopt QT/QML (, it's the right way to drop this mess!
Ugh.. damn thread got resurrected.. I'm going to ignore the folks that are belly aching.  You've clearly picked a desktop you like, and that's great.  I appreciate your opinion.  Can we move along now?

+Ingo Molnar Hey man..

1) You know that's a puzzler for me as well.  I find myself turning that thing off.  I usually forget about the bluetooth thing.  I'll ask around and see what the reasoning behind that.  That seems strange to me.

2)  That sounds like a bug.  Can you file a bug?  Even better, there is an image out with 3.8 that if you could test and see if that behavior still exists you could file a bug?  If not, I'll file one, but I prefer if you did since if they ask for information I don't want to do man in the middle communication.  I do have a two monitor setup with different resolution, I'll see if I can replicate it.

You can find the new image on under /pub/GNOME/misc/testing.  It's not a great image as it is kind of slow, hopefully we will get another one.

3)  That whole log off thing kind of sucks.  In the future, you can use gnome-session-quit I believe instead of rebooting.  It's not optimal.  My problem there was that I had some problems with Fedora crashing evolution-data-server, and gnome-session doesn't restart them for me, and so I have to sometimes log off and log back , but rebooting is just plain stupid.  I'm not sure about the problem with the mime type stuff.  I'll try to when I have time.
Hell, I totally fail to see how GNOME3 is not already optimised for spawning new terminal instances: WinKey →  t e r m → Ctrl+Enter
No extensions, no custom keybindings, no mouse interactions and pretty much easy to remember (Ctrl+click works in the same way).
+Sriram Ramkrishna

The Gnome reproduction image looks useful - I'll try it out and report all the bugs I can reproduce.
+Yves-Alexis Perez <OT> if that can cheer you up a bit  at La Sorbonne university (Paris 4) we were using XFCE (on top of  Freebsd) a couple of years ago (<2010) for an "introduction to CS" class (50 low end PCs) and our students loved it (way much that the former TWM) now we upgraded the pc and the new sysadmin (the freebsd guru left for other greener pastures) installed centos 5 (with gnome 2.x) and our students are  a bit less confortable with it but still ok.
For the record, they absolutely don't get why there's 3 types of fileselectors on a typical linux session: the firefox's one, the openoffice's and the gnome's one.  Each one with its own particular way of displaying your own filesystem.

And the first thing they ask is how to put back the left panel (à la explorer.exe) on the filemanager windows.  
Or you could just give up, like I did, and switch to something like LXDE where you get to keep more of your CPU and RAM...  Honestly, I use my computer to WORK.  Don't need eye candy.  Don't care if it "looks better."
Wow, bitching about opening a new terminal in Gnome? Heard of key mappings? 
+Sriram Ramkrishna You dislike GNOME, KDE, whatever, don't use it. shrug Just don't complain to others about it.

Just like if you, for example, prefer FreeBSD or Solaris to Linux. If you do, fine, that's your choice, but don't bash on the other OS.
+Sriram Ramkrishna

So I tried the following Gnome live image on a ThinkPad T500 with a second monitor attached:

and Gnome crashes upon bootup. The crash screen is not very helpful, it says I should contact the sysadmin - no other details provided.

Switching to a text console and logging in as root gave me a clue in dmesg, apparently gnome-shell segfaulted, while executing in

Which component on should I file the bug about the live image against?
Ingo, put it under gnome-shell. You can use the version as whatever number embedded in dusk image name. Thanks !
That's a packaging issue ☹  I assume it works for noone.
fwiw, the image works here if i do qemu-kvm -m 2048 -cdrom GNOME-3.7.90.iso
i'll put a new image up in a few days for 3.7.91 and we'll see how that goes.
Gnome forks exist in abundance because gnome isn't targeting laptop/desktop/netbook/ultrabooks, it's targeting tablets.

All the useful features that IRL users need are taken away, bit by bit.

"We aren't fucking around with design and changing interface just because."... Proof that people can and will convince themselves of anything when they feel backed into a corner.  The sheer level of critical reception to new gnome 'features', coupled with the numerous forks, has you on the back-foot.

The very worst thing about gnome right now is the useless application of often broken extensions.  The beauty of Linux, for me at least, when I first embraced it many years ago, was the package manager.  This made everything super simple and quick.  Gnome broke this, instead opting for a method whereby a user has to trawl endless pages of extensions and click a button to install ones they want.  These break with upgrades, and often hinder the user experience of the 'desktop' (if you can still call it that).

I personally feel that since the move to Gnome-shell, the Linux desktop has ground to a halt, with no real contender for a 'simple' DE that 'just works'.
+Sriram Ramkrishna We are already working with GNOME and Flashback developers to share more things as possible. But Classic mode and Flashback one cant satisfy all what MATE users want: the old nautilus interface (compact view, tree view sidebar, etc), the old control center, etc. But we like to work together on common things: use metacity, share applets, etc. And we have already removed a lot of duplicated  and/or deprecated applications and libraries, so now the MATE base consist of very few packages. I think all of this is useful for both MATE and GNOME projects.
+Stefano Karapetsas I should start another thread, the whole Mate/flashback/classic mode is probably worth another thread to discuss.  But I'm happy that you're working with everyone in the GNOME eco-system, I think that is pretty important.
+Olav Vitters  What exactly are you targeting then?  The desktop?  It certainly doesn't feel like it.  I don't even think the Gnome devs know their target audience anymore... They have some elusive goal that is unattainable.  They're trying to fix a problem that doesn't exist, whilst not fixing the problem which does exist... Hence why they're losing users in droves.  Here's a hint, people still want a desktop OS.  A useful, powerful, desktop OS.  Gnome is a broken toy OS, that tries to be a jack of all trades and master of none.  It treats its users like idiots, and hinders them when it should be helping them.
+Keith Hill 
IMO , xfwm4 + LXDE + pcmanfm + cairo-dock = best combo I've ever come across.
LXDE as usual is blazing fast and the real gem is pcmanfm which has all the features they've been culling from nautilus and is super snappy as the desktop app. It actually does a perfect compact view.
xfwm4 gives us all the compositor features we are so used to with gnome, without going overboard on the eyecandy.
Cairo-Dock is a great taskbar app (taskbars rule for productivity).
OK folks.  Closing out comments.  If people want I can create another post regarding forks if people want to argue about it.