Clasen: GNOME 3.7 at the halfway mark

[Posted January 25, 2013 by corbet]

Matthias Clasen previews the changes in GNOME 3.8. "Allowing you to focus on your task and minimizing interruptions has been an important aspect of the GNOME 3 design from the start. So far, we just had a global switch to turn off notifications. The new Notification panel expands on this and allows fine-grained control over what applications get to annoy you, and how much."

to post comments

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 16:40 UTC (Fri) by Zizzle (guest, #67739) [Link] (96 responses)

This first half of the post looks like a poor OSX clone.

Second half actually looks usable. Wouldn't surprise me if the class mode becomes more popular than the shell.

Still it looks like Cinnamon is the more polished and usable desktop. Hopefully Fedora 19 defaults to it.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 16:55 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link] (15 responses)

Cinnamon is already in Fedora repository and it is possible for anyone to volunteer and maintain a Cinnamon Fedora Spin. It is very unlikely the default choices would be changed.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 8:33 UTC (Sat) by marcel.oliver (subscriber, #5441) [Link] (14 responses)

To bring this discussion on a more constructive track:

As a long-time Fedora user and current user of Gnome 3/Cinnamon I would very much welcome if Fedora would take a more active role toward Cinnamon. Here are some reasons and benefits:

Gnome 3/Gnome Shell is very much geared toward a particular vision regarding the work flow. Developers have made it clear in public that they do not wish to dilute their vision by trying to be everything to everybody. I think think is a valid strategy for a desktop project. As a distribution, however, Fedora should try to develop a broader vision toward the Linux Desktop which goes beyond just bundling every half viable project.
Without any intent to start a flamewar here, I believe I can say that the weaker parts of the Gnome 3/Gnome Shell model relate to the traditional "workstation" workflows involving big screens, possibly multiple screens, many simultaneous applications, multiple instances of a single application, and work models involving mixtures of command line and GUI interface components. In other words, "developer" or "technical" usage in some sense.
Cinnamon sits fairly light-weight on top of the Gnome 3 infrastructure, so it's a minimally invasive strategy to extend the reach of Gnome 3 beyond what Gnome Shell can offer.
Gnome 2 infrastructure/Mate are on the way out. Their value lies in working with old hardware where accelerated or software rendering of Gnome 3 is not working/too slow. They are clearly not a long-term strategy into the future. But more effort could be expanded into making sure that all valuable features find some equivalent in either Gnome Shell or Cinnamon.
Cinnamon is a small project, both in the size of code and in the number of developers, but it fills a much bigger and important niche. Having Fedora put some concerted effort toward its development would stabilize both projects with benefits much bigger than potential effort.

If this could be achieved, the question of what is the default would be essentially moot.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 12:48 UTC (Sat) by kigurai (guest, #85475) [Link] (6 responses)

I hear point two repeated over and over.
As a guy who works with development, with multiple terminals, IDE's and documentation open, I have no problem using Gnome-Shell. I actually think my workflow is better now than with the "traditional" desktop paradigm.

I realize that this is a personal taste, of course, but I see this "only for casual users" comment so often that it seems to have become some kind of universal truth. And it's not.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 10:30 UTC (Sun) by marcel.oliver (subscriber, #5441) [Link]

Please read my original comment again. I never claimed that Gnome Shell is not suited for technical or development work. I only said its weaknesses are more prevalent in that area - it's a different claim. I would think the reason is that such workflows tend to be more diverse and more complex, so that users have created their own working styles based on the specifics of their work and abilities, some of which map better and some of them map worse onto the Gnome shell model. I don't think anybody is losing face by simply acknowledging this.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 17:52 UTC (Tue) by whitemice (guest, #3748) [Link] (4 responses)

+1 Point two GNOME3-is-for-tablets is BOooo-ooo-ooo-gus! I use GNOME Shell on multiple large displays. It is great, fast, and stable. It is different, but once you do a little adaptation it is an improved and streamlined workflow to 2.x and taskbar mania.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 18:36 UTC (Tue) by marcel.oliver (subscriber, #5441) [Link] (3 responses)

Point two GNOME3-is-for-tablets is BOooo-ooo-ooo-gus!

Note: these are your words, not mine. I am also no-one to argue that Gnome cannot be used on multiple large displays. Peoples' workflows and requirements are very different, so who am I to tell you what works for you?

Here are some items which I found cumbersome with Gnome Shell and where I have not been able to come up with an equivalent workflow using native Gnome Shell techniques. Maybe you can enlighten me, it could be that I just did not discover the right tricks:

Using "minimize" as a quick, non-distracting way to say "I am done with this window, I don't think I want to look at it again but I don't want to close the application just now because I may have to revisit the case".
Dealing with applications (in my case scientific simulations in python) which bring up several near-identical windows that are hard to visually distinguish, but where the launch order is clearly defined and reflected in the task-bar order. Handling them in Gnome Shell drives me crazy, yet it's a breeze with the good old task bar.
Flipping windows on and off (I do this often for visual comparison of two near-identical graphs or images)
Selectively closing windows when a certain sub-task is done. On the taskbar, I can travel along and right-click-close all the windows which I know I won't need any longer without opening them again.

These are the ones that quickly come to mind. I also find that I travel longer distances with my mouse using Gnome Shell, but that is not a killer criterion for me. Probably I still have the "old" midset too deeply engrained due to 25 years of computing with traditional Unix deskop metaphors?

That Gnome Shell is "great" with external displays is plain false. Since Fedora 15 it has this bug which make Gnome Shell (Cinnamon works fine, though) a total no-go in my production setup. The fact that no-one has even looked at it makes me think that Gnome Shell does not have very many users who use external screens, at least not the way I do.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 30, 2013 10:46 UTC (Wed) by thisisme (guest, #83315) [Link]

Flipping windows on and off (I do this often for visual comparison of two near-identical graphs or images)

I'm relieved to hear I'm not the only one who does this, with graphs, lists and spreadsheets.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 30, 2013 15:12 UTC (Wed) by mgedmin (guest, #34497) [Link] (1 responses)

> Maybe you can enlighten me, it could be that I just did not discover the right tricks:

> Using "minimize" as a quick, non-distracting way to say "I am done with this window, I don't think I want to look at it again but I don't want to close the application just now because I may have to revisit the case".

I tend to middle-click the window title bar, which sends that window to the background.

You can also ask to have the minimize button back with GNOME Tweak Tool.

> Flipping windows on and off (I do this often for visual comparison of two near-identical graphs or images)

And a quick Alt-Tab or Alt-` doesn't work for this?

For images I prefer to do the comparison in my web browser: two images in adjacent tabs, fliping between them with Alt+number (or Ctrl+PgUp/PgDn, that's a bit slower).

> Selectively closing windows when a certain sub-task is done.

The overview is fine for this, if you can distinguish the windows visually. Which you already said you can't always, so disregard this part of my comment.

> That Gnome Shell is "great" with external displays is plain false.

Yes, multiple displays only work great if you position the desktops side-by-side. They don't work well if you position one of them above the other.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 31, 2013 15:01 UTC (Thu) by marcel.oliver (subscriber, #5441) [Link]

I tend to middle-click the window title bar, which sends that window to the background. You can also ask to have the minimize button back with GNOME Tweak Tool.

I knew about the GNOME Tweak Tool, thanks for telling about the middle mouse button (works in Cinnamon, too...). So I agree that this point is not a killer argument, it just makes me wonder why hiding it is considered an improvement.

And a quick Alt-Tab or Alt-` doesn't work for this? For images I prefer to do the comparison in my web browser: two images in adjacent tabs, fliping between them with Alt+number (or Ctrl+PgUp/PgDn, that's a bit slower).

What you suggest is rather clunky, at best a poor workaround. Especially using the browser is just another workflow indirection and does not help when the windows are actually application windows rather than views of a standard format file stored in the filesystem.

The overview is fine for this, if you can distinguish the windows visually. Which you already said you can't always, so disregard this part of my comment.

Here the overview is not too bad, but as you already realize, can be a problem when the window content is too similar. Another reason why the taskbar feels more natural is that windows which are logically related (thus opened at a similar time) are close on the taskbar, but typically scattered geometrically all over the overview pane. I think that's the primary reason why in Cinnamon, where I have both easily accessible, I very much favor the taskbar and use the overview more for "lost" windows where I don't have any mental connection any longer between position in task bar and window.

Yes, multiple displays only work great if you position the desktops side-by-side. They don't work well if you position one of them above the other.

That's a shame because here there there does not even seem a philosophical or "big vision" obstacle to just fixing it. Given the fact that it used to work great for years in old Gnome and continues to work in Cinnamon, it can't be that much of a fundamental problem to fix.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 16:58 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (3 responses)

Let's be more specific. Fedora volunteers are usually involved in just packaging components. Red Hat has a Fedora team but they are mostly involved in development that is directly connected to Fedora - web apps and other infrastructure within Fedora itself. Red Hat also has a desktop team which has worked with the larger GNOME development team to introduce the GNOME classic session and that is part of Fedora 19 and Red Hat is unlikely to invest development effort into Cinnamon because GNOME classic session is the focus for people who want a GNOME 2 like workflow.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 10:20 UTC (Sun) by marcel.oliver (subscriber, #5441) [Link] (2 responses)

This is very encouraging news - the first time I hear about it. It would be great if you could expand on the "Gnome classic session" or refer to a good resource about it.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 17:27 UTC (Sun) by marcel.oliver (subscriber, #5441) [Link] (1 responses)

Oops, I now realize that this is in fact in TFA. I did not initially read it to that point, shame on me. This is big news, I did not expect this at all.

It will be interesting to watch how it turns out. It might obsolete Cinnamon if it is done well. I am not a fan of forking everything if that can be avoided...

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 28, 2013 12:14 UTC (Mon) by Rehdon (guest, #45440) [Link]

Not to dampen your enthusiasm, but it doesn't seem likely that the classic mode will ever catch up with Cinnamon: I seriously doubt that they'd pour much effort into it because it might threaten their "vision" (not to speak of the "brand" ;) wrt Gnome Shell.

One could also remark that not such thing as a "classic mode" would have been conceived if it were not for the success of GNOME 2 and 3 forks (Mate, Cinnamon).

At this point in time, I'm pretty happy with Cinnamon as is: it's much more than a "classic mode", although among other things Nemo kept all the useful features that they threw away with Nautilus, and more stuff is coming with Cinnamon 1.8 (http://www.webupd8.org/2012/12/what-to-expect-in-linux-mi...).

Rehdon

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 23:59 UTC (Sat) by misc (subscriber, #73730) [Link] (1 responses)

People are free to work on it if they want. If no more peole work on it, it is likely because Cinnamon is unconvincing, you say you are a user of Cinnamon, what prevent you from helping them ?

I tested Cinnamon when it was in the updates-testing, and I was far from being impressed ( but it worked so I gave +1 karma ). There was clearly a lack of polish and small design errorss all over the place ( like non aligned button, non coherent options, or too much useless detail like the 16 different way of minimizing a windows, or the whole plugin applet in the control center that was just empty ), and I found that disturbing.

Then I started to look at the commits, the code is better than Mate ( who is mainly taking code from gnome nowadays ) or mint ( where the python code is not that great ), but there is still some weird stuff going on.

This one is clearly wrong :
https://github.com/linuxmint/Cinnamon/blob/61f89a61861c64...

Because that's just a no-op, but starting the python interpreter for nothing. What is fun is that another commit was done to workaround the same problem this file try to correct ( ie having .py not matched by the windows matcher ), on 8374cb9a3661cb3dc71dfd66b3916f5a5adddf5d .

The cleaner solution is to just rename the file, but well...

The whole C part of gnome-shell was forked and is left almost untouched since 1 year. And since they renamed every possible function, there is merge conflict from time to time ( and sometime, they get committed 287f77f771078a5bc2df026dfe2376bf8c469bba ). So it is not hard to see this approach will sooner or later be more costly to maintain, unless they decide to do things cleanly. Because for now, that's not really sitting "cleanly on gnome 3", that's more "duplicating half of gnome-shell".

And in the python code, there is still some weird stuff like subprocess call instead of proper library ( 0cbb29180eed3d389a22c7d1be35a24d5d931f16 ), hardcoded paths everywhere ( https://github.com/linuxmint/Cinnamon/blob/master/files/u... ), so no jhbuild or test as non root.

So there is lots of things to do, and you are right that they could benefit from more coders, but given they forked alacarte, nautilus and said they would fork the rest if needed, I am not sure if they really wish to collaborate and work to a consensus if they can just fork. yet people are free to help, I have seen fedora packagers sending patches so there is some help.

And to finish, as said in another comment, I also have a big screen ( 23" ), I use command line, a browser, evolution on my day job and at home, and I use gnome shell without problem.
Maybe you just didn't found how to use it, for example, the middle click trick ( middle click on the dash run a new instance of the application, quite handy and feel almost natural if you already use the middle click for cut and paste as any old X11 user ).

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 11:13 UTC (Sun) by marcel.oliver (subscriber, #5441) [Link]

People are free to work on it if they want. If no more peole work on it, it is likely because Cinnamon is unconvincing, you say you are a user of Cinnamon, what prevent you from helping them ?

Well, one can always push responsibility downstream. But the problem is not that there are not enough desktop environment options. What is lacking right now is a reasonably inclusive distribution policy toward the Fedora desktop which has a clear path into the future and does not lead to further fragmentation of the Linux desktop. If the "Gnome Classic Session" mentioned by Rahul Sundaram is what it might promise to be, that could be something worth focusing effort on. But just calling for effort without a strategy seems a waste.

As for me personally: I have given feedback to Cinnamon. (In fact, I don't think Cinnamon is perfect, but I think they are taking a sensible approach.) My own expertise is not in C/C++ programming, but then I don't think being able/having the time to contribute code should be a necessary condition for taking part in a discussion on features and strategy.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 1:31 UTC (Sun) by vonbrand (subscriber, #4458) [Link]

Fedora goes where, well, its members take it. If you want more Cinnamon involvement, participate and help out.

Disclaimer: I'm Fedora Ambassador, and so an active member of the Fedora community.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 16:55 UTC (Fri) by proski (guest, #104) [Link] (76 responses)

Fedora 19 feature submission deadline is January 29, and Cinnamon is not in the feature list yet. Moreover, there is no Cinnamon spin of Fedora 18. It's very unlikely that a desktop environment would become default before it's available as a spin for wide testing.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 17:24 UTC (Fri) by tjc (guest, #137) [Link] (3 responses)

Features/Cinnamon as Default Desktop

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 17:34 UTC (Fri) by bpepple (subscriber, #50705) [Link] (1 responses)

Any Fedora contributor can make a Feature page and submit it to FESCo, but having that feature become accepted appears *extremely* unlikely in this case.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 1:36 UTC (Sun) by vonbrand (subscriber, #4458) [Link]

Care to share the intimate insight that tells you so? Might it just be that it isn't being pushed because it just isn't up to snuff, or even just that there is very little real interest (you know, not just vocal whining on random websites, but real users and developers)?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 18:25 UTC (Fri) by Company (guest, #57006) [Link]

Fedora is so much fun these days, isn't it?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 17:38 UTC (Fri) by tetley80 (guest, #88691) [Link] (70 responses)

It's very unlikely that a desktop environment would become default before it's available as a spin for wide testing.

That's inconsistent. Where was the Gnome 3 spin of Fedora before it became the default desktop?

(the preview of the G3 shell in Gnome 2 doesn't count, as it was incomplete)

Using the testing-via-spin argument right now is also bogus as Cinnamon has had plenty of testing via the Linux Mint distribution.

Any arguments along the lines of "keeping Gnome 2 and 3 in parallel would have been too much of a job" (in order to have a G3 spin) would be also bogus. Fedora could have kept Gnome 2 as is, without putting any further work into it.

Arguments in the theme of "the changes for Gnome 3 were too big at the OS level" would simply demonstrate that a separate Gnome 3 spin was sorely needed to test out the changes in a wide manner and get feedback before making G3 the default.

The non-testing of Gnome 3 in Fedora is more consistent with the view that Fedora is simply a dumping ground for new, untested stuff. Fedora has "first" in its mission, but "first" doesn't mean "forced".

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 17:51 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link] (69 responses)

GNOME 3 was an upgrade to an already existing default. Cinnamon is being proposed as a new default. They aren't at the same level and while Linux Mint might have it, testing specifically the integration bits in Fedora is important and a number of things had to be patched even during the review of Cinnamon because Linux Mint and Fedora had different settings etc.

Keeping GNOME 2 as it is isn't really an option since it is dead upstream and no distribution has chosen to do that for precisely the same reason. Also there is zero "force" unless you consider any "default" as force since there are tons of options in http://spins.fedoraproject.org and one more is easily possible when someone volunteers to do the work.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 18:58 UTC (Fri) by tetley80 (guest, #88691) [Link] (68 responses)

Keeping GNOME 2 as it is isn't really an option since it is dead upstream

That's false reasoning from two points of view. (i) G2 is not dead upstream, as the MATE desktop is maintaining the G2 branch. (ii) Just because something is "dead upstream", it doesn't mean it's useless; you can always incorporate minor bug fixes, which is the job of a distribution.

GNOME 3 was an upgrade to an already existing default

Sorry, but that's a gross misinterpretation at best, and attempt at rewriting history at worst.

Labeling G3 as an "upgrade" is not in the same league as updating gcc 4.6 to 4.7, or going from Gnome 2.26 to 2.28, or going from kernel 3.4 to 3.5.

In comparison to Gnome 2, Gnome 3 is closer to a rewrite, even if we only restrict ourselves to looking at the UI changes. This clearly indicates that G3 was a brand new component. This new component got insufficient testing before it was forced upon the Fedora community.

If you can force a new component such as G3 to be the default, not applying the same standard to the Cinnamon desktop is disingenuous. This is further underlined by the fact that the "delta" between G3 and Cinnamon is much smaller than between G2 and G3.

Moreover, the Gnome 3 shell can be considered a regression from a UI perspective, with Cinnamon aiming to fix that regression (while still using the underlying Gnome 3 components).

Also there is zero "force" unless you consider any "default" as force since there are tons of options

If we use the above logic, nobody will mind if we suddenly switch over to the Clang compiler (instead of gcc), without any wide testing in Fedora. After all, if people don't like it they can create their own gcc spin. While we're at it, how about we swap the Linux kernel to use one of the BSD kernels? Nobody will mind, as after all, there are tons of options. People can always create a Linux-kernel spin of Fedora.

The point I'm making above is that Gnome 3 was an arbitrary choice of a default, without due testing of the proposed default by the wider Fedora community. It certainly wasn't a simple "upgrade".

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 19:15 UTC (Fri) by mjg59 (subscriber, #23239) [Link] (49 responses)

It clearly wasn't arbitrary. There are people paid to do GNOME development in Fedora. That's not the case for any of the other desktops, including GNOME 2.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 19:30 UTC (Fri) by tetley80 (guest, #88691) [Link] (48 responses)

That's not the case for any of the other desktops, including GNOME 2.

That's false. Red Hat is directly paying for Gnome 2 maintenance, via maintaining Gnome 2 in both RHEL 5 and RHEL 6.

It clearly wasn't arbitrary. There are people paid to do GNOME development in Fedora

Well then, the "paid employment" argument is awfully sounding like the main criterion for selecting Gnome 3 as the default in Fedora.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 19:44 UTC (Fri) by mjg59 (subscriber, #23239) [Link] (47 responses)

Nobody is paying for GNOME 2 development in Fedora. GNOME 2 is maintained in RHEL, but for versions older than the last GNOME 2 that Fedora shipped. But yeah, shockingly, it turns out that "There are people paid to look after this in the distribution" is a good criterion to use when deciding what should be the default in the distribution.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 21:21 UTC (Fri) by tetley80 (guest, #88691) [Link] (46 responses)

So then, is the "paid employment" a clearly codified and documented criterion that Fedora officially uses? If not, it indicates that non-technical reasons are being used underhandedly to make decisions on technical matters. Even if it was a documented criterion, wouldn't it amount to discrimination based on non-technical issues? Either way, this type of behavior is certainly not in the spirit of the Fedora community.

It also amounts to letting people loose on a major every-day UI component and putting a blind eye to what they're doing, just because they're getting paid for it.

It's all well and good to propose new things, but to implement major UI regressions is abusing the community's patience. This is yet more evidence towards the view that Fedora is simply an unstable testing ground for future RHEL releases.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 21:55 UTC (Fri) by mjg59 (subscriber, #23239) [Link] (38 responses)

What's non-technical about it?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 22:05 UTC (Fri) by tetley80 (guest, #88691) [Link] (37 responses)

The contribution of person X is more valuable than person Y, because person X happens to be employed by Z. Nevermind that it is possible for Y (or multiple instances of Y) to provide the same labor as X.

Technical reasons include: is software X more stable than software Y, or my favorite, does software X have more regressions than Y ?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 22:21 UTC (Fri) by mjg59 (subscriber, #23239) [Link] (36 responses)

No. The contribution of person X is more valuable than person Y because person X spends more time working on the code in question and has a direct incentive to fix bugs and integrate software in a timely manner.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 1:09 UTC (Sat) by ebiederm (subscriber, #35028) [Link] (35 responses)

Given that the regressions in the Gnome UI experience have not gotten better there seems to be a fundamental flaw in your argument.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 1:16 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (6 responses)

If you don't like the UI, just say so. Don't call them "regressions".

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 3:41 UTC (Sat) by hadrons123 (guest, #72126) [Link] (5 responses)

It is a 'regresssion', when you loose functionality.
Initially I was interested in gnome-shell and I am still using it but recently things started to disappear from components like nautilus.

Why do you think 'nemo' got accepted in fedora?
Becoz nautilus 3.6 took away the features, what people wanted.
This logic applies to cinnamon as well.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 4:34 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (3 responses)

Just to be clear, as long as there is someone willing to maintain a open source component, it can be in the Fedora repository. There is no overall authority deciding what should be 'accepted'. Why do you think there has been forks would be a more pertinent question.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 7:44 UTC (Sat) by hadrons123 (guest, #72126) [Link] (2 responses)

@rahul
Not all packages which are forked from existing components are accepted by FESCO. Would you accept a older forked gcc or linux-lts kernel accepted in fedora just becoz someone is interested in maintaining it?

Gnome components are forked so much becoz there is an audience for older versions and people are actively using it. New versions of nautilus should have more functions on top of old features, but instead calling it as unmaintainable and removing the features is not what users want. Now instead of nautilus 3.6, people are willing to stay with nautilus 3.4 with new name 'nemo'. Within a couple of years if the gnome project continue the tradition of removing the core features and thrust some feautres that the devs think is important, I am pretty sure there will complete fork of all the components.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 7:46 UTC (Sat) by mjg59 (subscriber, #23239) [Link]

FESCO doesn't perform package review.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 16:52 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link]

"Not all packages which are forked from existing components are accepted by FESCO"

You don't seem to understand the package review process. FESCo is not involved. All it takes is one package maintainer to review and approve and it is typically upto him or her to determine packaging quality and only that. We don't typically ask the question why at all as part of the review process. Ours concern are usually limited to maintenance and sustainability. They can appeal to Fedora packaging committee to determine whether it conflicts with the packaging guidelines and in this case, there was a request to FPC and they determined that Cinnamon didn't conflict. The relevant packaging guidelines are at

https://fedoraproject.org/wiki/Packaging:No_Bundled_Libra...

There is really nothing stopping an older version of GCC from going in and Fedora does that now and then for compatibility. Fedora repositories have a lot of forked components and they are accepted on a routine basis. If a forked version of GCC gets submitted for review, it takes just one maintainer to review and approve it unless someone objects and in that case, it is upto FPC and they might escalate all the way to FESCo or even Fedora Board but such instances are very rare.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 5:44 UTC (Sat) by mjg59 (subscriber, #23239) [Link]

Changing from GNOME 3 to Cinnamon would result in a loss of functionality, as did the change from GNOME 1 to GNOME 2. How are you figuring out which transitions are worse than others?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 5:41 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (27 responses)

Define "regression".

Regression

Posted Jan 26, 2013 14:43 UTC (Sat) by pboddie (guest, #50784) [Link] (22 responses)

When something that used to work no longer does so.

One can claim that the product is now different and that you can't compare the old product with the new one (that conveniently uses the old product's name), but to the person thinking that they're getting an update of the same thing, it's a step backwards.

This means that when we try and put a modern distribution in front of people, instead of explaining how it is mostly the same (and thus what they are used to themselves) but better, we now have to explain why it is different and things that used to work no longer do so.

Frequent blog posts by people pretending to be brand strategists and visionaries don't actually placate user concerns because the users are mostly concerned with whether stuff does what they need it to, not whether the "positioning" of the software is right according to current market trends, or whatever.

After all, no amount of blogging can make something work again or even work in the first place.

Regression

Posted Jan 26, 2013 15:38 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (21 responses)

Going from GNOME 1 to GNOME 2 meant that I could no longer set the tile background for launcher buttons. Would you class that as a regression?

Regression

Posted Jan 26, 2013 16:37 UTC (Sat) by sorokin (guest, #88478) [Link] (9 responses)

If someone complains about a feature missing it's a regression.

If a feature is removed and somebody used it, you alienate your user. If people in GNOME cared about users they will never make something like G3. If people in Fedora cared about users they will never put G3 as default only because G3 is made by the same people as G2.

That is the main difference between free and commercial software. If they alienate users they lose money. If GNOME people alienate users they don't care. They only thing they care about is a "brand presence".

Regression

Posted Jan 26, 2013 17:09 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (8 responses)

So KDE 4 is a regression when compared to KDE 3?

Regression

Posted Jan 26, 2013 18:10 UTC (Sat) by nix (subscriber, #2304) [Link] (7 responses)

To some, yes. That's why Trinity exists (and the semi-separate OpenSUSE KDE3 fork).

Regression

Posted Jan 26, 2013 18:33 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (6 responses)

Regression's a subjective term?

Regression

Posted Jan 26, 2013 19:19 UTC (Sat) by bronson (subscriber, #4806) [Link] (5 responses)

No, a regression is a regression. Whether anybody cares or not, that's subjective.

And, whenever you remove a feature from a large project, no matter how meaningless or crappy, you will find somebody who cares.

Regression

Posted Jan 26, 2013 20:54 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (4 responses)

Right. So "Regression" here isn't terribly meaningful - almost every update will change or remove some functionality in a way that annoys somebody. By this sense, changing from GNOME 3 to Cinnamon would be a regression. It's a term that doesn't add anything to the conversation, and people should stop using it.

Regression

Posted Jan 26, 2013 21:18 UTC (Sat) by khim (subscriber, #9252) [Link] (1 responses)

By this sense, changing from GNOME 3 to Cinnamon would be a regression.

No. If things is mostly the same but has different name people expect that some features will not be available. When something is billed as "pure upgrade" but it removes features - people become quite vocal. Is it fair? No. But that's life. Deal with it.

Regression

Posted Jan 26, 2013 21:29 UTC (Sat) by mjg59 (subscriber, #23239) [Link]

Someone defines the term, I explicitly say that I'm using that definition of the term, and then you tell me that I'm wrong? How about disagreeing with the definition, instead?

Regression

Posted Jan 27, 2013 16:16 UTC (Sun) by bronson (subscriber, #4806) [Link]

We can agree that cinnamon has fewer regressions, right? And that they are spending a lot of time reducing potential regressions that their users might experience?

Sounds like a meaningful conversation to me. Where's the confusion?

Regression

Posted Jan 28, 2013 22:51 UTC (Mon) by sorokin (guest, #88478) [Link]

> almost every update will change or remove some functionality in a way that annoys somebody

In GNOME -- yes. In other reasonable projects that is simple not true. I would say that for most projects that is not true.

Regression

Posted Jan 26, 2013 16:56 UTC (Sat) by pboddie (guest, #50784) [Link] (10 responses)

Did it still have launcher buttons?

You can make the case that the lack of a gold exhaust pipe on a hypothetical electric car - let's call it the Ford Luxury - is not a regression since its non-electric predecessor - also called the Ford Luxury - needed such a thing, whereas no-one is really going to miss it on the electric model. The argument in such cases is that technology has made something obsolete and thus the need to choose between, say, gold and steel has been eliminated.

The problem is that the view (or excuse) that technology has made something obsolete is brought out far too often. GNOME 2 won't let you do something that GNOME 1 did? It's because it's all better, that's why! When the developers famously closed bugs against GNOME 1 because GNOME 2 was new and different (http://www.jwz.org/doc/cadt.html), they illustrated precisely this phenomenon.

People want the benefit of brand recognition but also the benefit of not having their current product compared to the previous one, even though having the same name on the product is inevitably going to invite such comparisons. If you want to enjoy the former benefit, you have to relinquish the latter.

Regression

Posted Jan 26, 2013 17:10 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (9 responses)

Yes, it still had launcher buttons.

Regression

Posted Jan 27, 2013 15:44 UTC (Sun) by pboddie (guest, #50784) [Link] (8 responses)

Did anyone miss the removed feature? If not, maybe it's not really a regression. If they did ("My launcher buttons look awful now - why did they change this?") then one can discuss whether it's a minor regression or not.

Perhaps launcher buttons couldn't be transparent before and the background setting was a workaround that was made obsolete, just like the absence of an exhaust pipe on a hypothetical car whose predecessor required one. Since the result is almost completely superior, there's little reason to complain about it, but that almost certainly cannot be said for many consequences of these big product upgrades.

Regression

Posted Jan 27, 2013 16:27 UTC (Sun) by mjg59 (subscriber, #23239) [Link] (7 responses)

So we've gone from a regression being something that no longer works to something that no longer works and which nobody really misses, maybe. And it's probably only a minor one. Which sounds awfully like they're subjective, and then we're just back to "I don't like these changes". Why not just say that you don't like the changes, rather than pretending that there's been a truly objective analysis?

Regression

Posted Jan 27, 2013 17:52 UTC (Sun) by pboddie (guest, #50784) [Link]

No, I conceded that if a feature goes away and nobody notices, you don't have to take a position on it. It's still a regression, but nobody really cares, and you can argue that in the grand scheme of things, no significant feature has fallen off the feature list.

I also noted that if a feature goes away because it is obsolete - to take an ancient example that once applied to some desktop systems, you don't have to allocate memory manually to a process because the system now does that for you - you can ignore the regression because nothing that anyone was doing before that they can no longer do now (say, allocate a process size of N) is anything they still need to be able to do (because the system will give the process N if it asks for it). Since there is no longer any benefit in even being able to do those obsolete things, there's no general functional regression (you can still run that process).

None of this has anything to do with whether I "like the changes". The issue I have is the way that people deny the experiences of the users by playing games with definitions of what the system was and is, as if the users were supposed to care more about the brand gymnastics than the features actually being delivered.

Regression

Posted Jan 28, 2013 14:04 UTC (Mon) by nye (subscriber, #51576) [Link] (5 responses)

Yes Matthew. You're a very smart person and you can successfully outmanoeuvre people by playing semantic games without ever having to lower yourself to making meaningful points.

Clearly, everything you say is therefore correct. It must be wonderful to know that you are so much better at everything than anyone else. I hope you enjoy your future career in law, and wish you all the best.

Regression

Posted Jan 28, 2013 16:05 UTC (Mon) by mjg59 (subscriber, #23239) [Link] (4 responses)

The meaningful point is that "This UI has regressions and therefore should never have been made the default" is a false claim to a measurable change in quality and an implication that, as a result, the choice was irrational. People should stick to "It breaks my workflow" or "I find this change objectionable for some other reason", which are arguments that can lead to actual discussions about whether the improvements to other people's workflow outweigh that, or whether there would be some way to tweak the behaviour such that a specific usecase becomes practical without compromising any other design decisions. Bringing up "regression" shuts that opportunity off, because we all know that regressions are unarguably bad and need to be fixed immediately. People should say what they mean rather than throw around terms they can't even meaningfully define.

Regression

Posted Jan 28, 2013 16:17 UTC (Mon) by patrick_g (subscriber, #44470) [Link] (3 responses)

> People should say what they mean rather than throw around terms they can't even meaningfully define.

With Gnome 2 I was able to control my CPU governors with a graphical applet.
With Gnone 3 there is no applet and I can't control my CPU governors through the GUI.

For me it's simply a regression.

Regression

Posted Jan 28, 2013 16:40 UTC (Mon) by mjg59 (subscriber, #23239) [Link]

There's no "For me". A regression is a regression. With GNOME 1 I used to be able to set the background to launcher tiles. With GNOME 2 I couldn't. Is this a regression? When defining regression as "Something that used to be possible is no longer possible", yes. Do I care? No. Did anybody? Probably, but it was 10 years ago and they've probably got over it.

Regression

Posted Jan 28, 2013 17:02 UTC (Mon) by pizza (subscriber, #46) [Link] (1 responses)

>With Gnome 2 I was able to control my CPU governors with a graphical applet.
>With Gnone 3 there is no applet and I can't control my CPU governors through the GUI.

Perhaps there isn't an "applet" but there is an extension, available through the extensions.gnome.org website.

Regression

Posted Jan 28, 2013 18:52 UTC (Mon) by patrick_g (subscriber, #44470) [Link]

> Perhaps there isn't an "applet" but there is an extension, available through the extensions.gnome.org website.

You're right. The extension is here : https://extensions.gnome.org/extension/444/cpu-freq/
It didn't exist six months ago and I'm glad someone took the pain to create it. I retract my comment about this regression :)

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 20:31 UTC (Sat) by jubal (subscriber, #67202) [Link] (3 responses)

Change that breaks user workflow.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 21:38 UTC (Sun) by pizza (subscriber, #46) [Link] (2 responses)

> Change that breaks user workflow.

Here's the problem with that.

Pretty much any change (short of a crash fix) will qualify as breaking someone's workflow -- ie "it's different than what I'm used to"

It's been said many times over in this thread; nobody's forcing you to update anything. Your three-year-old Fedora/OSX/Windows/DOS/whatever installation works as well (and identically) today as the day it was released. Heck, go with something like RHEL and it'll be continually supported for a decade.

But don't complain about it lacking $randomfeature, because in the real world nothing is independent.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 28, 2013 7:21 UTC (Mon) by ebiederm (subscriber, #35028) [Link] (1 responses)

The command line in unix has decades of updates and even several complete reimplementations of various commands without breaking a persons muscle memory or workflow.

It is a cop out to say that you can't have that in other UIs.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 28, 2013 7:40 UTC (Mon) by mjg59 (subscriber, #23239) [Link]

Oh come on. Try doing anything vaguely advanced with GNU utilities and then use the same commandline on BSD. Try using ifconfig syntax with ip. tar has had several incompatible changes in option symantics.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 22:14 UTC (Fri) by raven667 (subscriber, #5198) [Link] (6 responses)

That's not valid logical argument. Taking in all the relevant information including the size and availability of the maintainership of a code base are reasons for making distribution decisions. It is not "underhanded" to make rational decisions nor is it necessary to thow up the bureaucratic smoke screen about "documented criterion" as if decision making is about some mechanistic following of a pre-written rule book. You can take all of this as just more evidence to service your prejudices or you can try to better understand the world around you and why things happen, the choice is yours.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 22:50 UTC (Fri) by tetley80 (guest, #88691) [Link] (5 responses)

nor is it necessary to thow up the bureaucratic smoke screen about "documented criterion" as if decision making is about some mechanistic following of a pre-written rule book.

I can understand the need for flexibility, but the use of a "documented criterion" cannot be simply dismissed as "bureaucratic". If one does not have good and well thought basis for making decisions, then one gets arbitrary decisions which can lead to strife, such as UI regressions. If "documented criterions" did not work, nations wouldn't have laws and constitutions.

There is obviously a trade-off between codification and flexibility. However, in this case it is my strong opinion that too much flexibility in the UI area has provided a disservice to the Fedora community.

You can take all of this as just more evidence to service your prejudices or you can try to better understand the world around you and why things happen, the choice is yours.

A nice condescending comment. I do not wish for this discussion to degenerate into a flame fest, so I will not entertain this further.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 23:16 UTC (Fri) by mjg59 (subscriber, #23239) [Link] (3 responses)

"UI regressions"

You keep using that phrase as if it means something other than "I don't like this UI", but I don't know what.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 30, 2013 0:04 UTC (Wed) by bojan (subscriber, #14302) [Link] (2 responses)

I think you've participated in enough discussions here on LWN, including with myself, that actually explained what that meant. And I see you keep repeating this "question" as if it was never answered.

You know, stuff like this. Changing workspaces took one mouse click before and one view change (using GUI). Now it takes several, including a lot more mouse travel. Desktop visibility is zero, which was not the case before. The amount of pixels that change when working remotely using VNC has been significantly increased, which makes things even slower over poor links. It is practicality impossible to move items on the panel or define new panels using techniques available for at least two decades. Etc.

The _measurable_ stuff.

As I pointed out numerous times, Gnome 3 overview is essentially an implementation of RFC1925(6). That in itself is a regression, because everything is one step further away.

Unfortunately, nobody in Gnome development team is brave enough to acknowledge any of these facts. Instead, we are getting an Ultralite Gnome Classic, which just looks a bit like "classic", but it can't do most of the stuff that was possible before.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 30, 2013 8:49 UTC (Wed) by ovitters (guest, #27950) [Link]

"brave enough to ack that"

Learned loaded question in another article :)

Clasen: GNOME 3.7 at the halfway mark

Posted Feb 2, 2013 3:46 UTC (Sat) by fandingo (guest, #67019) [Link]

https://extensions.gnome.org/extension/464/workspacebar/ took two minutes to find and two clicks to install. That seems to completely alleviate your workspace problem.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 20:40 UTC (Sat) by raven667 (subscriber, #5198) [Link]

I may be condescending but while I don't care for GNOME 3 Shell either I think that stating so plainly is sufficient, one doesn't need to go fishing for procedural reasons to reinforce ones opinion. I don't think you can successfully argue that having available, dedicated resources for maintaining a software project shouldn't be a criteria for a distribution accepting it and making it part of the default, whether this is written in a procedural manual or not.

If you don't like GNOME 3 Shell just stand up and say so, no need to beat around the bush.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 19:16 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link] (16 responses)

Your reasoning doesn't make sense. You claim that GNOME 2 is not dead upstream because there is a fork but the presence of the fork doesn't say anything about GNOME 2 itself and the fork didn't exist when Fedora switched to GNOME 3. GNOME 3 is certainly an upgrade to GNOME 2 in the same way that KDE 4 is an upgrade to KDE 3 and Fedora 18 installer is an upgrade to Fedora 17 installer. You might not like that upgrade but arguing that they shouldn't be considered an upgrade despite sharing a lot of the same components and put up by the same project is just silly. There was a GNOME 3 feature proposal btw which was accepted by Fedora engineering steering committee.

If you want to propose a switch by default to Clang, it will go through the same feature proposal as Cinnamon and must have a very robust rationale for doing so. So I am afraid I don't see your point at all. Nobody is using any force to make you install Fedora or even if you chose Fedora to force you to install GNOME.

As we move forward, we continue to make more choices available and in fact, one of the accepted features for Fedora 19 is to bring in Enlightenment 17. I am working on that.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 19:44 UTC (Fri) by tetley80 (guest, #88691) [Link] (15 responses)

You claim that GNOME 2 is not dead upstream because there is a fork but the presence of the fork doesn't say anything about GNOME 2 itself

For all intents and purposes, MATE is Gnome 2. Furthermore, Red Hat is still maintaining Gnome 2 in RHEL 6. As such, Gnome 2 is far from dead.

the fork didn't exist when Fedora switched to GNOME 3

You are of course correct in the above observation. However, this is boiling down to a chicken and egg argument (pardon the unintended pun). MATE probably wouldn't exist if Gnome 3 didn't have such massive UI changes. If Gnome 3 was more evolutionary, or had built in options to let people choose which new UI elements to use, we wouldn't be having this discussion.

There was a GNOME 3 feature proposal btw which was accepted by Fedora engineering steering committee.

The Fedora review process is hence broken, given the amount of fall out that the UI changes in Gnome 3 have generated. It strongly suggests the engineering steering committee has a very cavalier attitude when it comes to UI regressions.

Nobody is using any force to make you install Fedora

This is consistent with a cop out. You have responsibility to the Fedora community, and instead of admitting that the Gnome 3 shell UI is a point of pain for many in the Fedora community, you are in effect telling them to go away.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 20:42 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link] (7 responses)

GNOME 2 *upstream* is dead and GNOME 3 was that natural upgrade path for Fedora just like the upgrade from KDE 3 to KDE 4 or any other new upstream release. MATE and Cinnamon are more recent projects and they are available in the repo if you prefer them.

"You have responsibility to the Fedora community, and instead of admitting that the Gnome 3 shell UI is a point of pain for many in the Fedora community, you are in effect telling them to go away"

Now, you are being intentionally dishonest. I told you explicitly that I am working on integrating Enlightenment and I helped review Cinnamon as well. So no, I am not telling them to go away but use whatever they prefer and I am working on making those choices available.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 21:59 UTC (Fri) by tetley80 (guest, #88691) [Link] (6 responses)

use whatever they prefer and I am working on making those choices available.

So on one hand we have a major UI regression in a component that has worked fine for years, and on the other hand we're being told that this regression doesn't matter because of the many choices we have?

Okay, let's follow this through. Gnome 2 used to be a first class citizen in Fedora. It has been replaced (not "upgraded") by a completely different user interface, known as the Gnome 3 shell. Packages such as Cinnamon are closer to the Gnome 2 UI, yet they're second class citizens (ie. I need to manually install them in F17/F18, or explicitly do a network install in F18). Yes, technically the choices are there, but you need to dig to find them.

Summary: "You liked ABC? Sorry, you can't have it anymore. Here's DEF instead. Oh, you don't like that? Well, we have GHI which is kind of like ABC, but you'll have to find it yourself. Have fun!"

If Fedora is serious about "making choices available", Cinnamon should be elevated to be one of the main options (in the installer) for the user interface, while also being present on the _default_ installation media. None of this separate "spin" nonsense.

While this might be a workable solution, it's also a type of bug fix. It would be far more productive to prevent these kind of UI interface regressions in the future.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 22:12 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link] (5 responses)

You can easily install MATE which you claimed is equivalent to GNOME 2 so "ABC" is available just fine and no digging requires since is well advertised and widely documented.

sudo yum install @mate-desktop

The "default" download is a live image cannot hold all the different desktop environments due to size constraints and this is the reason we have multiple options at http://spins.fedoraproject.org. I am not sure what more you want honestly. You seem to be arguing for the sake of it.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 23:03 UTC (Fri) by tetley80 (guest, #88691) [Link] (4 responses)

The "default" download is a live image cannot hold all the different desktop environments due to size constraints and this is the reason we have multiple options at ...

The discussion started off about Cinnamon being the default UI for Fedora, so let's not fuzzify the issue by stating that the MATE desktop takes up too much space.

Given all the pain about the Gnome 3 UI, Cinnamon provides a more traditional user interface, while still using Gnome 3 components. Cinnamon should be a first class citizen in Fedora (along with Gnome 3), and be on the default install media, not relegated to a separate spin. In contrast to MATE, Cinnamon requires relatively little space. It can be considered as an addon to Gnome 3, not a replacement like MATE.

This discussion is starting to go around in circles, so I'll leave it here.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 1:22 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (3 responses)

I bought up MATE because you claimed that the equivalent doesn't exist and Fedora isn't providing the choices when it is demonstrably clear that there are multiple alternatives to suit everyone's personal preferences. I don't see why a single yum install cinnamon/ gnome-classic-session in Fedora 19 or the gui equivalent of checking a tick-box is that hard for anyone. You seem to want to insert your personal preference into the default image as an addition and claim that even a different image with your personal preference as a default isn't sufficient and I don't think you have provided any realistic justification for that.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 13:06 UTC (Sun) by deepfire (guest, #26138) [Link] (2 responses)

> You seem to want to insert your personal preference

Sorry for this bit of judgementality, but I used to regard you as
one of the more reasonable fellows.

The "personal" preference sticker you slap on tetley80's words
is actually being attached to the collective perception of an awfully
significant audience.

Tread carefully -- the project you represent can handle only so much self-inflicted alienation.

Listen.. Can you hear it? The sound of mindshare going away?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 15:06 UTC (Sun) by ovitters (guest, #27950) [Link]

IMO, http://en.wikipedia.org/wiki/The_pot_calling_the_kettle_b...

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 15:06 UTC (Sun) by rahulsundaram (subscriber, #21946) [Link]

I just represent myself and "personal preference" is a very neutral word. Like I said, you might not like something and that's fine but claiming that choices aren't available unless it is in the default image is hardly reasonable.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 0:53 UTC (Sat) by misc (subscriber, #73730) [Link] (6 responses)

> Furthermore, Red Hat is still maintaining Gnome 2 in RHEL 6.

So on one comment, you say that the fact that people are paid to work on a software should not be taken in account, and yet, you now use the same exact argument to say that since people are paid to keep gnome 2 aline, then it is not dead and so should be considered.

Can I recommend a little bit of self consistency ?

And frankly, for people wanting gnome 2, there is RHEL, SLES ( around 50$ per year ), or for the one who cannot afford this either due to volume or to any reason, there is Centos, Scientific Linux, etc.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 17:06 UTC (Sat) by pboddie (guest, #50784) [Link] (5 responses)

Self-consistency doesn't come into it here. By noting that people are paid by Red Hat to maintain GNOME 2, it indulges and then demolishes the argument that the existence of paid maintainers must be the factor that determines inclusion in Fedora (subject to argument about the relationship between Fedora and RHEL).

Of course, Red Hat may be eager to stop having to maintain GNOME 2, and the real reason may be that the existence of external paid developers is what really decides whether something is included in any of the company's products - a matter of not having to bear all the load internally - plus the level of enthusiasm for GNOME 3 amongst internal developers compared to that for GNOME 2, but I'd also hope that the users of the company's products might also get some say (other than voting with their feet, of course).

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 20:49 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (4 responses)

It doesn't demolish anything really. Fedora is not a Red Hat product and Red Hat maintains an older version of GNOME 2 than what was included in Fedora when it was replaced by GNOME 3. Red Hat is willing to put in that additional resources to maintain a commercial product but those same resources are not available for Fedora. It isn't about external paid developers as such since majority of the GNOME developers are from Red Hat in the first place but paid development is part of how you evaluate sustainability of the project. Maintaining a older version has a significant cost and that is not the focus of Fedora. All of this should really be obvious I think.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 17:09 UTC (Tue) by hadrons123 (guest, #72126) [Link] (3 responses)

Fedora is not a Red Hat product

Is it so?

That's an interesting piece of information to know from you.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 17:56 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

If you have ever read the original announcement of Fedora or ever visited the about section in http://fedoraproject.org or heck if you just see the domain name, you would have known this about 8 or so years back.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 30, 2013 15:14 UTC (Wed) by hadrons123 (guest, #72126) [Link] (1 responses)

I do know that fedora is a red hat sponsored community product, but you are asking us to believe that red hat has no role in what packages gets in and out of fedora, specifically for example cinnamon or systemd. Fedora Cinnamon package review was waiting for more than 6 months citing various reasons and finally it was uploaded. I do appreciate and thank your effort in getting cinnamon in to the official repos, but you have to admit that there was so much hesitation from other people in Fedora camp.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 30, 2013 15:48 UTC (Wed) by rahulsundaram (subscriber, #21946) [Link]

You have shifted your points but yes, Red Hat has zero role in Cinnamon or systemd getting into Fedora. You don't have to take my word for it. The review process is entirely public.

https://bugzilla.redhat.com/show_bug.cgi?id=771252

It was submitted by a volunteer, Leigh Scott and I took up the review voluntarily.The major opposition came from another volunteer, Christoph Wickert who was concerned about sustainability and violation of packaging guidelines regarding forks and we discussed it with the packaging committee and it was eventually approved by another volunteer, Dan Mashal. The delay was partly my fault since I took up the review and didn't finish it up quickly since I was focusing on release specific tasks at that time and we also had to get the packaging committee approval and the reviews themselves took time.

systemd was submitted by me and approved by another volunteer.

https://bugzilla.redhat.com/show_bug.cgi?id=598299

It became the default after approval by FESCo (a fully elected body btw) which did delay it by one release despite having a number of Red Hat employees in it. So if you are looking for favoritism you have picked the wrong examples. Are there other packages submitted by Red Hat employees as part of the job? yes but the important point is that the process is not designed to be any different regardless of who you work for.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 19:22 UTC (Fri) by drag (guest, #31333) [Link]

> That's false reasoning from two points of view. (i) G2 is not dead upstream, as the MATE desktop is maintaining the G2 branch.

G2 didn't '_die_' upstream. Gnome released a new version of the desktop. The Gnome 2 developers became Gnome 3 developers. All the GTK libs and everything else that applications need to be backwards compatible still exists and have had new stable releases.

Mate is a fork of Gnome 2 and didn't exist until much more recently then Fedora following the upgrade path away from Gnome 2. It's a _NEW_ project.

http://fedoraproject.org/wiki/Features/Gnome3

Targeted release: Fedora 15
Last updated: 2010-04-04
Percentage of completion: 100%

http://en.wikipedia.org/wiki/MATE_(desktop_environment)

Initial release August 19, 2011; 16 months ago

And it didn't make it into Fedora until the Fedora 18.

If you want to get all weepy over this stuff then that's fine, but at least don't just make random crap up.

> Labeling G3 as an "upgrade" is not in the same league as updating gcc 4.6 to 4.7, or going from Gnome 2.26 to 2.28, or going from kernel 3.4 to 3.5.

It's in the same league as going from 2.4 kernel to 2.6 kernel.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 17:46 UTC (Fri) by mjg59 (subscriber, #23239) [Link]

I don't think that's got anything to do with it. On the other hand, the fact that Cinnamon doesn't have several people paid to do full time work on Cinnamon, with a special focus on Fedora, probably does.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 17:53 UTC (Fri) by tjc (guest, #137) [Link]

> Second half actually looks usable.

The second half probably looks a lot like RHEL7.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 18:19 UTC (Fri) by Kit (guest, #55925) [Link] (1 responses)

> This first half of the post looks like a poor OSX clone.

GNOME's design has pretty much always* been heavily inspired by Apple's OSes (or more so a caricature of them). GNOME Shell seems to be an obvious exception to that design...

* I hear early GNOME 1 was much more Windows like, while early KDE was more Apple OS like... but I didn't use either till late GNOME 1 / KDE 2.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 20:27 UTC (Fri) by sorpigal (subscriber, #36106) [Link]

If by "always" you mean "Since 2.0" then maybe, otherwise no. Prior to 2.x there wasn't a lot of UI design coherence and it seemed more inspired by CDE than anything else, except maybe individual whim.

Early KDE was a lot more like Windows than MacOS. Try comparing some KDE 1.x screenshots with Win9x, for example, and you'll see a lot of resemblance.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 16:49 UTC (Fri) by robert_s (subscriber, #42402) [Link] (27 responses)

Tangentially relevant, but I can't *stand* those on|off toggle switches that seem to be proliferating everywhere nowadays. They are horrendously ambiguous. Why would I slide the switch _towards_ the word "on" to turn it off? Surely the word "on" being exposed implies that the slider is over, and therefore in, the "off" position?

I can't be the only person that sees a problem with these.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 16:59 UTC (Fri) by proski (guest, #104) [Link] (1 responses)

They annoy me too. They take much more screen space than checkboxes and need to be translated.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 15:57 UTC (Tue) by mgedmin (guest, #34497) [Link]

> They take much more screen space than checkboxes

That's a good thing, actually.

The new switches bugged me too, but I got used to them after a week or so. The clearly different background color when the widget is in the "ON" position helped.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 17:08 UTC (Fri) by epa (subscriber, #39769) [Link] (3 responses)

It seems like a good thing to replace checkboxes with these sliders when displaying on a tablet or touchscreen interface. However, that means making a global change so that every checkbox in every application is replaced by the new control. It doesn't mean a mishmash of different checkbox-style controls depending on when the program was written and in which language or toolkit.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 17:18 UTC (Fri) by mpr22 (subscriber, #60784) [Link] (2 responses)

Skeuomorphic binary sliders suck on touch displays too.

binary sliders

Posted Jan 25, 2013 18:56 UTC (Fri) by mlinksva (guest, #38268) [Link]

I don't understand them on physical devices, let alone on-screen configuration. Or rather, I have to think a few seconds each time I encounter one. Much prefer switches/buttons/checkboxes. But I'm probably missing something important.

(Binary sliders minor complaint; overall good work GNOME people.)

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 20:19 UTC (Fri) by nix (subscriber, #2304) [Link]

For extra fun, make the sense of the label the opposite of what normal humans would use to refer to it. A good example is what Amazon did to the Kindle around the time of the Kindle Touch: "Wireless on" suddenly turned into "Airplane mode off", which both needs to be translated even into British English and is also precisely the wrong sense, since nearly all the time when people are turning their wireless off they're doing it to double the battery lifetime of their Kindles and not because they're getting onto an aircraft.

But then it got controlled by a stupid skeuomorphic binary slider, so sliding the slider towards the word 'On' turns the wireless on again (though they call it 'turning airplane mode off'), and sliding it towards the word 'Off' turns the wireless off (though they call it 'turning airplane mode on').

Making simple things complicated...

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 17:40 UTC (Fri) by tjc (guest, #137) [Link] (9 responses)

I think so too. The label actually makes things worse, since one doesn't know if it indicates current state, or potential state as a result of "flipping" the switch. Every time I use one of those things I have to stop and think about it. The problem is that the label in the wrong place: if it were directly above the switch, it would make sense, but then of course it would take up more room.

Light switches in buildings are label-less, and this doesn't seem to confuse too many people. Perhaps toggle widgets should be the same, and follow convention: left off, right on.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 8:43 UTC (Sat) by aleXXX (subscriber, #2742) [Link] (1 responses)

When switching a light switch, you see whether the light is currently on or not, so switching will always change that, so no label is needed.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 12:42 UTC (Sat) by nijhof (subscriber, #4034) [Link]

When switching a light switch, you see whether the light is currently on or not, so switching will always change that, so no label is needed.

For added fun maybe one could add a multi-way switch ("hotelschakelaar") to the GUI, where effectively the on or off status is an XOR of all the switches..

(certainly in my house I knever know which switch will turn which light on or off: most of the time there are three switches together in some random order compared to the topology of the rooms, half of which are multi-way)

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 19:02 UTC (Sat) by bjencks (subscriber, #80303) [Link] (6 responses)

Most normal (with a toggle sticking out, not the flat rocker-style) light switches I've seen have "on" and "off" printed on them. "On" is below the toggle, so visible when the switch is on, and "off" is above and visible when the switch is off.

"Three-way" switches (the ones mentioned below that XOR multiple switches) are unmarked. That's actually usually my first cue that it's a three-way switch rather than on-off.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 16:46 UTC (Sun) by nix (subscriber, #2304) [Link] (5 responses)

Curious. Outside the theatre (where they have high-tension lighting) and bathrooms (where lights are pullcord-controlled by regulatory fiat), I've never seen any light switches other than the rocker-style ones. They never ever have labels, or any indication as to which light they control, unless someone has stuck a sticker on them indicating which is which (and sometimes, especially for inside switches controlling outside lights, the light they control is not even visible: once I even saw a switch controlling a light mounted on the side of a different house!)

I suspect the nature of light switches is one of those things heavily controlled by historical contingency and national regulation. You cannot say 'normal' about things like light switches: different regions have different normals.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 12:18 UTC (Tue) by sorpigal (subscriber, #36106) [Link] (4 responses)

In the USA a typical light switch has on and off printed on it (and done in raised beveling so that a blind person could figure out the distinction if necessary). Most everyday light switches follow this pattern and I think it's probably where the GNOME3 style checkbox switch concept gets its skeuomorphic inspiration.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 12:54 UTC (Tue) by cortana (subscriber, #24596) [Link]

I think it is more likely that the inspiration was the iPhone user interface.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 14:46 UTC (Tue) by micka (subscriber, #38720) [Link]

On the other side, I have not yet seen a light switch with anything printed on it (wall switch or switches on the power chord).

My parent would not know the meaning of "on" or "off", anyway.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 16:25 UTC (Tue) by nix (subscriber, #2304) [Link] (1 responses)

That's a scary-looking light switch, both too easy to catch something on and tear, and quite hard to flip when you have no hands free (I routinely turn lights on and off by just leaning on them with my chin).

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 19:05 UTC (Tue) by foom (subscriber, #14868) [Link]

Not really. Complaining about the usability of something that's been in wide use for decades from a sketch of it is pretty funny.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 18:52 UTC (Fri) by nim-nim (subscriber, #34454) [Link] (6 responses)

They are an i18n trainwreck too

Most languages do not use two/three-letter words for on/off. But the widget assumes it is the case (on/off is the only bit that makes it non-ambiguous; every time I see it I'm reminded of All the UI disasters where designers had to include a huge "press here" arrow to get users to notice their hidden-in-chrome button)

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 1:22 UTC (Sat) by misc (subscriber, #73730) [Link] (5 responses)

On my version of gnome ( 3.6 on F18 ), there is nothing to translate. Either the widget is greyed, with a O to mean "unchecked", or this is blue, with a I, to mean checked. I am not sure of that's colorblind friendly.

So i guess the "on/off" is just a locale variation of the gtkSwitch widget ?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 16:00 UTC (Sun) by mathstuf (subscriber, #69389) [Link] (3 responses)

I can never remember which of I and O is on and which is off. I remember that the O looks like a closed circuit, but I can't remember if there's a "but its backwards" part to it as well. To be fair, the I can be rationalized as a crossbar linking a switch, so "it closes the circuit" isn't foolproof. Anyone have a trick that works for them?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 27, 2013 16:22 UTC (Sun) by bronson (subscriber, #4806) [Link] (2 responses)

Binary: 1 is on, 0 is off?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 28, 2013 22:48 UTC (Mon) by mathstuf (subscriber, #69389) [Link] (1 responses)

That's probably how I *should* remember it, but I encountered the symbols before I knew binary, so it wasn't a trick I would have come up with. Maybe I'll remember it now that I've commented here about it, but historically my problem with mnemonics (and similar tricks) is that I can't remember the mnemonic itself, defeating the purpose. Usually, it takes some rather obscure connection for it to "click" and be second nature.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 16:09 UTC (Tue) by mgedmin (guest, #34497) [Link]

I believe these come from the IEC standard ("Graphical symbols for use on equipment"). And yes, they were inspired by binary 0 and 1, if you trust Wikipedia.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 28, 2013 16:32 UTC (Mon) by nim-nim (subscriber, #34454) [Link]

The "on/off" is the original English widget, that GNOME people found so cool-intuitive-and-self-explanatory they fostered it on everyone. It's based on some physical switches common on a few Anglo-Saxon locales, that were never used anywhere else because their English-only nature. (physical should have raised warning bells, software is not hardware, most physical analogies have a terrible history)

Of course during the implementation process they found the whole concept was untranslatable. Instead of revisiting the original decision, they just dumped the text labels on most locales

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 19:03 UTC (Fri) by bluss (guest, #47454) [Link] (3 responses)

It has already been fixed too (idea-wise): http://www.chrisnorstrom.com/2012/11/invention-multiple-c...

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 25, 2013 19:26 UTC (Fri) by dashesy (guest, #74652) [Link] (2 responses)

That sounds very neat, nice design

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 0:11 UTC (Sat) by proski (guest, #104) [Link]

I think it's still an overkill where a checkbox would suffice. But it could be a nice replacement for the radio buttons.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 26, 2013 8:54 UTC (Sat) by aleXXX (subscriber, #2742) [Link]

And a reply to that:
http://scriptkiddie-chronicles.tumblr.com/post/3505542606...

Alex

Classic mode questions

Posted Jan 26, 2013 22:13 UTC (Sat) by bojan (subscriber, #14302) [Link] (49 responses)

Looking at the classic mode screenshot, there are a few questions:

- where is the workspace switcher (i.e. visibility is still zero)?
- can it be customised (can components be moved from panel to panel or from place to place on the panel)?
- will it be usable on a computer that has no 3D acceleration?

Looking at the state of extensions, the answer to the first one is probably "there won't be a fully functional one". If nothing changed in the way components are laid out in Gnome 3, the answer to the second one is probably "this cannot be done". Don't really know how good emulated 3D will be by the time 3.8 hits released distros. Last time I checked (F-17) it was too slow and was burning CPU like there was no tomorrow on my VMware VM.

Classic mode questions

Posted Jan 26, 2013 22:55 UTC (Sat) by drago01 (subscriber, #50715) [Link] (43 responses)

1. Don't know the answer to that yet probably there will be one (you have 4 fixed workspaces though).

2. No.

3. No ... but "a computer that has no 3D acceleration" in 2013?! Even my phone has a GPU that can handle 3D acceleration at a rather large resolution.

> was too slow and was burning CPU like there was no tomorrow on my VMware VM.

VMware does support 3D acceleration just install the driver.

Classic mode questions

Posted Jan 27, 2013 1:19 UTC (Sun) by cortana (subscriber, #24596) [Link] (5 responses)

> 3. No ... but "a computer that has no 3D acceleration" in 2013?! Even my phone has a GPU that can handle 3D acceleration at a rather large resolution.

My VirtualBox VM doesn't. Once fallback mode is gone I'm regretfully going to have to find a new desktop environment.

Classic mode questions

Posted Jan 27, 2013 4:45 UTC (Sun) by Cyberax (✭ supporter ✭, #52523) [Link] (4 responses)

1) It actually does.
2) Fedora can use purely software LLVMPipe for OpenGL.

Classic mode questions

Posted Jan 27, 2013 13:32 UTC (Sun) by cortana (subscriber, #24596) [Link] (3 responses)

> 1) It actually does.

Not IME. <https://www.virtualbox.org/ticket/9581> <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=651936> <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=651936>

> 2) Fedora can use purely software LLVMPipe for OpenGL.

Good for them. If only it worked in Debian <http://www.youtube.com/watch?v=synr-avC0WA>.

Classic mode questions

Posted Jan 27, 2013 15:42 UTC (Sun) by Company (guest, #57006) [Link] (2 responses)

Maybe you shouldn't run a useless distro.

Classic mode questions

Posted Jan 27, 2013 17:00 UTC (Sun) by cortana (subscriber, #24596) [Link]

Hardly the kind of comment we want on LWN. Please take this attitude elsewhere!

Classic mode questions

Posted Jan 27, 2013 17:08 UTC (Sun) by nix (subscriber, #2304) [Link]

Such a distro would include, uh, Red Hat and Fedora, since both prefer QEMU as their virtualization engine, and nothing QEMU supports for display output (not SPICE, yet, and certainly not VNC) supports virtualizing 3D.

So, currently, Fedora when run atop Fedora is a platform that doesn't support 3D, even if the hardware does.

Classic mode questions

Posted Jan 27, 2013 8:43 UTC (Sun) by bojan (subscriber, #14302) [Link] (2 responses)

> VMware does support 3D acceleration just install the driver.

Installed as part of default Fedora. Still, slow. Still, high CPU usage.

Classic mode questions

Posted Jan 27, 2013 9:36 UTC (Sun) by drago01 (subscriber, #50715) [Link] (1 responses)

> Installed as part of default Fedora. Still, slow. Still, high CPU usage.

Then there is a bug somewhere (or a configuration issue) ... this is not the expected behavior.

Classic mode questions

Posted Jan 27, 2013 23:25 UTC (Sun) by bojan (subscriber, #14302) [Link]

There is no configuration. All automatically generated.

Classic mode questions

Posted Jan 27, 2013 8:54 UTC (Sun) by bojan (subscriber, #14302) [Link] (14 responses)

> 2. No.

See, that in itself, is a major regression.

A perfectly fine, working system of arranging things has been replaced by chance (e.g. whoever loads last gets to be the leftmost, for instance). Also, users cannot use the usual drag/drop to organise their desktop.

Which serious desktop environment doesn't support even the simplest of customisations in 2013? Even the brain dead Windows 8 has the old desktop hidden away somewhere...

Instead of working on Gnome Classic in terms of "here is a set of extensions that will give you look XYZ", the developers should be thinking about "here is a set of tools to make you desktop the way you want it". I know - not the "philosophy" of Gnome...

Classic mode questions

Posted Jan 27, 2013 9:39 UTC (Sun) by drago01 (subscriber, #50715) [Link] (13 responses)

> Instead of working on Gnome Classic in terms of "here is a set of extensions that will give you look XYZ", the developers should be thinking about "here is a set of tools to make you desktop the way you want it". I know - not the "philosophy" of Gnome...

There is a limit on what you can do with configuration. "the way you want" can be very different for many people. Drag and Drop to reorder some elements does only help with one aspect. So in GNOME 3 there is a powerful extension system that lets you change / edit almost anything ... so it allows way more changes that GNOME 2 did to achieve the "the way you want" (for users that for whatever reason don't like the defaults and/or have specific unique needs).

Classic mode questions

Posted Jan 27, 2013 11:59 UTC (Sun) by bojan (subscriber, #14302) [Link] (12 responses)

> there is a powerful extension system

Yes, so powerful it _cannot_ help you with reordering of elements, for instance. It also requires fiddling with source code, as it has been pointed out numerous times.

I have an idea. Why don't we just have autoexec.bat and config.sys. That'll fix it. :-)

Classic mode questions

Posted Jan 27, 2013 12:03 UTC (Sun) by drago01 (subscriber, #50715) [Link] (11 responses)

> Yes, so powerful it _cannot_ help you with reordering of elements, for instance.

That's plain wrong.
You can write extensions to reorder whatever you want to reorder.

> It also requires fiddling with source code, as it has been pointed out numerous times.

It does not require "fiddling with source code" ... writing code is enough. Or get someone to do that for you.

> I have an idea. Why don't we just have autoexec.bat and config.sys. That'll fix it. :-)

The extension system allows that as well. You can write an extension to read those files and do actions based on their contents. ;)

Classic mode questions

Posted Jan 27, 2013 12:25 UTC (Sun) by paulj (subscriber, #341) [Link] (6 responses)

Out of the subset of desktops users who already have the programming skills to learn a JavaScript extension system for their window manager in any reasonable amount of time, many will not have the time or inclination.

Then there's the remaining set, who even if they had the inclination, couldn't do it in a reasonable amount of time. Further, some significant subset of those will not have the inclination.

Really, it's 2013, and we can't re-order icons with the mouse by default? It's approaching 20 years now since the days of complex config files for window managers were seen as a good thing. And at least back then they were primarily declarative - which didn't require programming skill (things like m4-preprocessed config file systems excepted).

Classic mode questions

Posted Jan 29, 2013 0:15 UTC (Tue) by khim (subscriber, #9252) [Link] (5 responses)

Really, it's 2013, and we can't re-order icons with the mouse by default?

Exactly. It's 2013 and that's why we don't reorder icons with mouse. MacOS had the ability to reorder program icons in 1984. MS Office 97 had the ability to drag-n-drop menu items in 1997. But you can not rearrange program icons in iOS 6.1 and menu MS Office 2013!

I'm still not sure if it's good trend or not, but it's not like GNOME is the only program which moves in this direction.

Classic mode questions

Posted Jan 29, 2013 8:32 UTC (Tue) by paulj (subscriber, #341) [Link] (4 responses)

To be clear, we're discussing launcher short-cut icons, e.g. on the panel.

In both Android and iOS you have complete freedom to add, re-arrange and remove the short-cut icons on the default / "home" / front screen.

Classic mode questions

Posted Jan 29, 2013 16:22 UTC (Tue) by mgedmin (guest, #34497) [Link] (3 responses)

Likewise, in GNOME Shell I can re-arrange and remove the short-cut icons on the dock.

What was your complaint again?

Classic mode questions

Posted Jan 29, 2013 16:30 UTC (Tue) by paulj (subscriber, #341) [Link] (2 responses)

What part of the word "panel" was confusing to you?

Classic mode questions

Posted Jan 29, 2013 18:31 UTC (Tue) by mgedmin (guest, #34497) [Link] (1 responses)

Sorry, I lost track of the context and forgot you were talking about the new classic mode, rather than stock GNOME Shell (which doesn't have launchers on a panel).

Classic mode questions

Posted Jan 30, 2013 7:49 UTC (Wed) by paulj (subscriber, #341) [Link]

The panel in GNOME shell also is lacking an ability to add/remove/re-organise launchers with the mouse. It's not incapable of supporting launchers though...

Classic mode questions

Posted Jan 27, 2013 13:29 UTC (Sun) by bojan (subscriber, #14302) [Link] (3 responses)

You obviously don't even understand the problem. Two extensions cannot arbitrate who's going to be where without knowing about each other or having a central location for config.

The silly suggestion that users should ask someone to write code when they want to move something on their desktop is truly amusing. It does remind of config.sys and autoexec.bat days and ordering of drivers, so they can all fit.

Classic mode questions

Posted Jan 27, 2013 15:05 UTC (Sun) by drago01 (subscriber, #50715) [Link] (2 responses)

> You obviously don't even understand the problem. Two extensions cannot arbitrate who's going to be where without knowing about each other or having a central location for config.

A extension can allow reordering using DnD and it can as well write a config file (or use gsettings) to store them. It can also listen for other extensions that get enabled (after it got loaded) and reorder them as well.

Before claiming that others "obviously" don't understand anything please go and read the code you are talking about.

Classic mode questions

Posted Jan 27, 2013 22:27 UTC (Sun) by bojan (subscriber, #14302) [Link] (1 responses)

So, extension A says "I want to be leftmost". Extension B says "I want to be leftmost". Who wins without central configuration? Who decides who's wishes are more important?

Please, look at some of the extensions and how they clash with each other in this way. Without predefined, central configuration, which decides on what goes where with finality, it is not possible to have deterministic behaviour.

Classic mode questions

Posted Jan 27, 2013 23:32 UTC (Sun) by bojan (subscriber, #14302) [Link]

But, coming back to the original point, which was about (basic) ability to customise. If I remember correctly, the editor of this site used to run a vertical panel, for instance. He cannot do this any more (if what you're saying is true).

So, the Gnome Classic is more like Gnome Classic Ultralite. Just looks like it, but really doesn't do the same thing.

Classic mode questions

Posted Jan 27, 2013 10:51 UTC (Sun) by paulj (subscriber, #341) [Link] (16 responses)

My computer has 3D acceleration, but using it increases the mean-time-between-hard-lockups from "so infrequent I don't notice", to "in the order of a day". (Nouveau on NV44). Cinnamon doesn't help me here, as it's still based on mutter, and using 3D.

I guess the GNOME answer is that I'm supposed to buy a new graphics card, or computer.

Classic mode questions

Posted Jan 27, 2013 11:59 UTC (Sun) by drago01 (subscriber, #50715) [Link] (15 responses)

So when you buy a car and the third gear does not work will you pretend that it does not exist or try to get it fixed?

Your graphic driver has a bug that causes your system to look up the obvious solution here is to get that debugged and fixed rather then running away from it and pretending that the system cannot accelerate 3D. It can and you paid for a GPU that does that ... so why not use it?

Classic mode questions

Posted Jan 27, 2013 12:12 UTC (Sun) by paulj (subscriber, #341) [Link] (14 responses)

I looked in the phone book for "graphics driver garage", but there don't seem to be any near me.

Also, why are you assuming I chose that GPU? Had it been my choice, I would indeed have gotten something with better drivers. It wasn't my choice though. I'm sure I'm not unique in that. Not everyone has the ability to choose to buy graphics cards that have well-supported 3D drivers - either because of their position, or possibly because of their knowledge.

Sadly, a significant number of computers do not have well-supported 3D under Linux. Sadly, this means GNOME Shell (and derivatives) are not going to give users a nice experience on a significant numbers of computers.

That's reality.

Classic mode questions

Posted Jan 27, 2013 12:47 UTC (Sun) by drago01 (subscriber, #50715) [Link] (13 responses)

> I looked in the phone book for "graphics driver garage", but there don't seem to be any near me.

You have searched for the wrong thing ... what you want is a bug tracker to report your problem (and data to help to get it fixed). In your case you have another driver to chose from.

> Also, why are you assuming I chose that GPU?

I have just said that you paid for it. Maybe that is not true either but someone has payed for it ... why do you just accept that it is useless rather then try to get it to work?

> Sadly, a significant number of computers do not have well-supported 3D under Linux.

Citation needed. That might have been true in the past but 3D is no longer considered just "nice to have" and driver have improved significantly because of that.

Also the comment I have replied to stated "does not have 3D acceleration" ... finding such a system in 2013 is a challenge ;) (if you exclude obscure server chipsets).

Classic mode questions

Posted Jan 27, 2013 13:26 UTC (Sun) by paulj (subscriber, #341) [Link] (12 responses)

Yes, I know about bug trackers, I have used them. However, you mentioned it was like getting a car fixed. So where do I go where I can pay someone to fix GNOME-shell induced graphics problems? I don't have any confidence that just leaving a bug report in some tracker somewhere, in the hope some hacker I have no contractual relationship with will fix it, will lead to this issue being fixed. No more than I expect my car to be fixed if I just park it near a car mechanic.

As for which computers don't have good 3D support under Linux, install Fedora on any NVidia system. I don't have a citation to hand, however it's surely not controversial that NVidia graphics constitute a sizeable set of modern computers?

Classic mode questions

Posted Jan 27, 2013 13:52 UTC (Sun) by drago01 (subscriber, #50715) [Link] (11 responses)

> Yes, I know about bug trackers, I have used them. However, you mentioned it was like getting a car fixed.

My point was you won't just accept that it is broken but try to get it fixed.

> So where do I go where I can pay someone to fix GNOME-shell induced graphics problems?

Find an engineer that has the required knowledge and pay him for fixing it. You can search in IRC channels or mailing list for people with the required skills.

> I don't have any confidence that just leaving a bug report in some tracker somewhere, in the hope some hacker I have no contractual relationship with will fix it, will lead to this issue being fixed.

If you don't even tell the developers about your bug chances are that it indeed never gets fixed unless someone fixes it by accident or someone else hits it and reports it.

> As for which computers don't have good 3D support under Linux, install Fedora on any NVidia system.

Done that and works fine.

> I don't have a citation to hand, however it's surely not controversial that NVidia graphics constitute a sizeable set of modern computers?

No the controversial part is that they don't have working 3D acceleration. Just because you hit a bug with your specific GPU / configuration does not mean that NVIDIA cards do not work at all. Also there is the binary driver which works as well.

Classic mode questions

Posted Jan 27, 2013 14:11 UTC (Sun) by paulj (subscriber, #341) [Link] (10 responses)

My desktop wasn't broken before. It broke with GNOME's window-manager adding 3D as a requirement!

The problems with nouveau's 3D support and GNOME-Shell are not exactly there because of a lack of my bug-report. Look in the bug trackers yourself I'd love to see nouveau support improve.

I have in the past donated money to or bought things from wish-lists for various projects and people. Ideally, I'd pay the employer of the majority of GNOME and Linux graphics hackers, but they don't (yet?) have any offering of support for GNOME shell, afaik. Been less able to do this the last few years, for personal financial reasons.

I guess there's lessons here for me about what I should rely on.

Classic mode questions

Posted Jan 27, 2013 16:11 UTC (Sun) by drago01 (subscriber, #50715) [Link] (9 responses)

> My desktop wasn't broken before. It broke with GNOME's window-manager adding 3D as a requirement!

You can't expect developers to not utilise the hardware that is available because there is a bug on your system. Apparently it either always has been broken (and you just did not notice it) or it coincidently broke at the same time you upgraded to GNOME3 (by getting the broken driver at the same upgrade).

> The problems with nouveau's 3D support and GNOME-Shell are not exactly there because of a lack of my bug-report. Look in the bug trackers yourself I'd love to see nouveau support improve.

[citation needed]. Which bugs? Sure there are bugs but stop claiming that GNOME3 does not work on nouveau just because it hits a particular bug on your system. Just tested it again (NVA0 GPU) works just fine.

Classic mode questions

Posted Jan 27, 2013 16:32 UTC (Sun) by paulj (subscriber, #341) [Link]

My original comment was about the mean-time-between-lockups. Yes, it works initially. Now stick with the Nouveau drivers...

Classic mode questions

Posted Jan 27, 2013 16:38 UTC (Sun) by paulj (subscriber, #341) [Link] (7 responses)

Also, you keep interacting with me by blaming me as being the cause of why I (and evidently others) are blocked from using GNOME shell because of problems related to its 3D dependency.

Note that I'm not blaming any one, I'm just giving factual statements as to what led me to make the decisions I did. I suggest you reflect on the utility of reacting to such by blaming me.

Classic mode questions

Posted Jan 27, 2013 18:08 UTC (Sun) by drago01 (subscriber, #50715) [Link] (6 responses)

I am not blaming you for anything I just replied to your statement "I guess the GNOME answer is that I'm supposed to buy a new graphics card, or computer." saying that buying new hardware is not necessary fixing the driver is sufficient.

Classic mode questions

Posted Jan 28, 2013 14:36 UTC (Mon) by nye (subscriber, #51576) [Link] (5 responses)

>saying that buying new hardware is not necessary fixing the driver is sufficient.

Your 'sufficient' solution is orders of magnitude more difficult than the 'not necessary' solution. Fixing this requires either:
a) Finding and employing a team of sufficiently skilled people for sufficient time to solve the problem - let's say financing ten person-years, for a rough order of magnitude (I would hope it's somewhat less than ten, but one is clearly unrealistic, given that it entails catching up with a constantly moving target from a position quite a long way behind)
b) Waiting until either somebody else does this, or
c) Waiting until enough volunteer time goes in to provide the equivalent result.

Some combination of b and c is the most probable outcome, but we're still looking at a few years yet, at least. In comparison, buying new hardware is simple and relatively cheap.

That's not to say that it's the *best* solution, since clearly we'd all like to have working high-quality drivers for all the hardware we can buy, but saying "oh, x isn't necessary because you can just do y" is silly when x is a far easier option than y.

Classic mode questions

Posted Jan 28, 2013 18:44 UTC (Mon) by drago01 (subscriber, #50715) [Link] (4 responses)

> Your 'sufficient' solution is orders of magnitude more difficult than the 'not necessary' solution.

Depends on the bug in question ... we don't have any details nor a bug report with details.

> a) Finding and employing a team of sufficiently skilled people for sufficient time to solve the problem - let's say financing ten person-years, for a rough order of magnitude..

No idea where you are getting this numbers from but if fixing it costs more then buying new hardware then sure go ahead and buy new hardware. To go back to the car analogy if fixing the car costs more then buying a new one I'd rather go and buy a new one.

Classic mode questions

Posted Jan 28, 2013 21:34 UTC (Mon) by paulj (subscriber, #341) [Link] (3 responses)

There is little point in me filing a bug, as I have little information other than "{graphics,machine} locked up hard, reboot needed". I didn't notice any particular pattern to when it would happen. Any bug I'd file would have a low S/N ratio. I doubt I am only one seeing this with Nouveau, e.g., just from the last few months:

https://bugzilla.redhat.com/buglist.cgi?list_id=1055201&...

I don't think it is terribly controversial to say that 3D with Nouveau is less than super-stable. The driver is also blacklisted by Google Chromium for WebGL, for whatever that's worth (they seem to have a high bar though).

My solution for now is to use a desktop that exercises only the more mature, 2D parts of the graphics driver. I may give a GNOME shell / mutter based environment (i.e. Cinnamon) another go when I upgrade to F18, as it has a blacklist and (I gather) will fallback to soft-rendering (or can be made to).

Classic mode questions

Posted Jan 29, 2013 11:32 UTC (Tue) by drago01 (subscriber, #50715) [Link]

You can set the environment variable LIBGL_ALWAYS_SOFTWARE to 1 to force fallback to software rendering.

Classic mode questions

Posted Jan 29, 2013 14:18 UTC (Tue) by ovitters (guest, #27950) [Link]

Nouveau indeed is sometimes *really* unstable. It does not lock up hard though. E.g. numlock and so on works. Wait 30 seconds or so and it recovers. Wonder if you experience the same or not.

I noticed it is related to system resources somehow. Meaning: if I don't open too many tabs in Firefox, it is way more stable.

Classic mode questions

Posted Jan 29, 2013 16:57 UTC (Tue) by Thanatopsis (guest, #14019) [Link]

I'm a big Open Source advocate and whenever possible, I use and support such software. Sometimes it just doesn't work. I've encountered issues with the Nouveau driver in the past, particularly running programs under WINE. Try using the Nvidia driver. Fixed the problems I was encountering.

Classic mode questions

Posted Jan 27, 2013 13:23 UTC (Sun) by deepfire (guest, #26138) [Link] (1 responses)

> > - where is the workspace switcher (i.e. visibility is still zero)?
> Don't know the answer to that yet probably there will be one
> (you have 4 fixed workspaces though).

How soon the 4 fixed workspace limit is going to be lifted? (I mean, is 4 a magic number or what?)

What about the 2D grid workspace layout?

Classic mode questions

Posted Jan 27, 2013 13:53 UTC (Sun) by drago01 (subscriber, #50715) [Link]

> How soon the 4 fixed workspace limit is going to be lifted? (I mean, is 4 a magic number or what?)

That is a config option. Has been since 3.6 (where you can also enable fixed workspaces using a config option).

Classic mode questions

Posted Jan 27, 2013 0:46 UTC (Sun) by misc (subscriber, #73730) [Link] (4 responses)

Bastien Nocera told me that the plan is t reduce automatically and gracefully the most resources consuming effects if needed.

Given that I can use gnome-shell with vnc/spice, I see no reason to not be able to do it with vmware virt system.

Classic mode questions

Posted Jan 27, 2013 8:45 UTC (Sun) by bojan (subscriber, #14302) [Link] (3 responses)

> I see no reason to not be able to do it with vmware virt system.

I cannot speak for the reasons, because I really don't understand enough to comment. However, when I tried Gnome Shell/mutter in my VMware session under F-17, that was noticeably slower than fallback/metacity. It was also consuming a lot more CPU.

Classic mode questions

Posted Jan 27, 2013 21:34 UTC (Sun) by smoogen (subscriber, #97) [Link] (1 responses)

Well it is going to use the CPU to emulate the effects versus the non-existant GPU so your CPU/memory usage will be much higher. I am not sure that the items Bastien was talking about are in F17, F18 or something being worked for F19.

Classic mode questions

Posted Jan 27, 2013 22:33 UTC (Sun) by bojan (subscriber, #14302) [Link]

> Well it is going to use the CPU to emulate the effects versus the non-existant GPU so your CPU/memory usage will be much higher.

Which brings us to the real point: what is the purpose of using 3D rendering here? Is it to display (more or less) useless animations and endless screen changes (i.e. overview) or is it so that _appropriate_ hardware does the job of rendering? With Gnome 3, it appears to be the former.

Classic mode questions

Posted Jan 27, 2013 23:43 UTC (Sun) by bojan (subscriber, #14302) [Link]

> However, when I tried Gnome Shell/mutter in my VMware session under F-17, that was noticeably slower than fallback/metacity.

To be fair, I just tried with F-18. It seemed faster than F-17 in this scenario, but all Windows were distorted (half width, distorted colours), so not sure whether that was a proper test.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 28, 2013 11:01 UTC (Mon) by heijo (guest, #88363) [Link] (3 responses)

Is the halfway mark the point where only half of the features remain?

That seems quite bad, I hoped we were already down to a tenth at most.

At this pace, it's going to take a few more years before we can finally have the ideal, perfect, completely featureless, desktop.

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 28, 2013 11:45 UTC (Mon) by drago01 (subscriber, #50715) [Link] (2 responses)

No it is the halfway mark between the 3.6 and 3.8 release ...

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 28, 2013 23:34 UTC (Mon) by mathstuf (subscriber, #69389) [Link] (1 responses)

Then wouldn't "3.8 at the halfway mark" have been a better title? Or maybe "3.7: the halfway mark"? Or "3.7 is the halfway mark"?

Clasen: GNOME 3.7 at the halfway mark

Posted Jan 29, 2013 10:41 UTC (Tue) by ovitters (guest, #27950) [Link]

3.8 has stable releases, so half way through 3.8 could also mean half way through the stable cycle