Problems With The GNOME Shell

[hubick]

I already posted my full ranty review at http://forums.fedoraforum.org/showthread.php?t=263609

Though I'm still wondering if there is an extension to get the workplace switcher back on the main panel?

The thing is, being that I'm basically wanting to customize the shell away from this new direction and back to what I had before (I *was* happy), I'm thinking I should just jump to KDE or XFCE if this direction is where Gnome thinks it's a good idea to be heading.

[kraftman]

Huh? There's already a GNOME3 PPA for Ubuntu natty which is quite popular, and don't forget that Fedora, also a very popular distro, is shipping with gnome-shell by default now.

PPA's not aimed at newbies. It was Ubuntu which helped the Gnome the most in gaining popularity.

I think the reason we some mudslinging going on here is because there are people who feel threatened by GNOME 3, fearing that their beloved KDE is going to be even further marginalized.

As far I can see after Gnome3 release KDE userbase is getting stronger and stronger.

[DanL]

The thing is, being that I'm basically wanting to customize the shell away from this new direction and back to what I had before (I *was* happy), I'm thinking I should just jump to KDE or XFCE if this direction is where Gnome thinks it's a good idea to be heading.

I kind of see xfce 5.0 heading in the gnome 2.x direction since there are other "light" DE's out there for increasingly obsolete hardware (lxde, fluxbox, unity-2d).

[agd5f]

mangobrain, AdamW, thanks for the hints! I'll try them out.

[movieman]

Those of you who are saying "real work" (as opposed to ivory tower happy sleep dream time play hour that those of us who is GS must be engaged in) are being both elitist and myopic.

Nope.

We're people who use Linux every day for real work rather than Facebook. Unity and Gnome 3 probably both work fine if you just want to open up a web browser and post a new funny kitten video to your friends (Unity does, at least). But they're god-awful for those of us who are used to having dozens of windows of a dozen or more applications spread across multiple workspaces.

Gnome 2 works for that. Neither Gnome 3 nor Unity do. This is a huge step backwards for those of us who actually use Linux to do something productive.

Now here is a concern of mine: the giant list of huge icons is a bad way to list applications. For one people generally know what they want to do but not the name of the application that does it (unless it's a primary app for them). The paradigm doesn't scale. Add fifty or sixty apps and it I simply a mess. A favorites section along with categories (but not as implemented) is a better solution.

And after calling us elitist, you apparently agree that the interface is broken. As for 'favorites', again it's fine if you use four or five applications, but it's a crappy idea if you use a dozen or more; you'll never have enough 'favorites' and hence the application you actually want to start is likely to have fallen off the list. Or if you do set it to two dozen 'favorites' your list will be so long that it will require ten seconds of scrolling to get to the application you want.

Straightforward hierachical menus work well. 'Activity' interfaces that require moving the mouse all over the screen, clicking multiple times and waiting for fancy animated effects every time do not. The fact that the fans of these changes have to resort to 'but I can press a key and then type the name of the application' clearly demonstrates just how poorly the graphical interface works.

[agd5f]

After a few tweaks from AdamW and mangobrain, I'm pretty pleased with gnome-shell. I think it will really start to shine when more shell extensions are available.

[Delgarde]

gnome-tweak-tools does not fix the fact that gnome 3 is meant for the tiniest netbooks. I've been using Fedora since the first version and after a few days of daily F15 usage I think gnome 3 was the biggest mistake ever made by Fedora. They should have at least waited for gnome 3.2 because 3.0 is raw and buggy, that is, if they are really determined to marginalize the big screen user.

How so? I have a few of the same issues identified in the article, but for the most part, Shell is great on a 24" screen. And on a netbook, it's no better (or worse) than it's predecessor in terms of efficient use of screen space - something Unity (the latest version, at least) does much better on.

All the people asserting Shell was designed for netbook use - I don't see where they're coming from...

[AdamW]

How so? I have a few of the same issues identified in the article, but for the most part, Shell is great on a 24" screen. And on a netbook, it's no better (or worse) than it's predecessor in terms of efficient use of screen space - something Unity (the latest version, at least) does much better on.

All the people asserting Shell was designed for netbook use - I don't see where they're coming from...

I think people mostly assume the giant icons are for netbook / tablet use cases, but it's not actually true.

It's valid to have the _perception_ that GNOME 3 is designed for netbooks or tablets or phones, or to suggest that its design may have the _effect_ of working best on netbooks or tablets or phones. Like you, I'd disagree - I'm quite happy running it on dual 20" screens - but it's a valid perception. It's definitely incorrect to say that it's _designed_ for such environments, though, because it isn't. From the Shell design wiki:

"Effectively works on contemporary hardware: the Shell will provide an excellent experience on touch-based devices and will scale down to small screen sizes. It has also been designed with wide-screen in mind"

note "will scale down to" small screen sizes, i.e., it isn't designed specifically to them.

On the giant icons:

"The avoidance of exclusive application categories and nested sub-menus is a distinct advantage of application launching in the shell compared with the GNOME 2 desktop. Users do not have to guess which category an application is in, and the motor control demands of the application picker are lower than those of menus. The application picker also utilises spatial memory, making it quick and easy to relocate applications."

You can agree or disagree with the concept there, but it's clear that the concept is not 'this is good for small screens'.

In general, if you read through https://live.gnome.org/GnomeShell/Design/ and sub-pages, it's pretty obvious no-one was designing exclusively with small screens in mind.

[hubick]

"The avoidance of exclusive application categories and nested sub-menus is a distinct advantage of application launching in the shell compared with the GNOME 2 desktop. Users do not have to guess which category an application is in, and the motor control demands of the application picker are lower than those of menus. The application picker also utilises spatial memory, making it quick and easy to relocate applications."

Hah!

1) Overwhelming my brain and forcing it to scan through the icons and text for every single one of the 100 or whatever app installed on my system is *not* faster than guiding me to it through a single level of categorical submenus, provided...

2) If users are having problems guessing which category an app is in, then maybe you should pick better categories?

3) Or better yet, put icons in *both* menus? ie, make the menu's based on tags, and tag an app icon with all appropriate labels.

4) How is having to first bring up an activities view, to do any of the things I used to be able to do directly, reducing motor control demands? And have these guys even tried using workspaces in their new shell?

5) I'm expected to be able to use "spatial memory" to remember where each and every app icon I want is located amongst all the 100 or whatever app icons are installed on my whole system, and this is better than before, because I'm too stupid to remember that "Firefox" is under the "Internet" menu? Really?

[liam]

I think people mostly assume the giant icons are for netbook / tablet use cases, but it's not actually true.

It's valid to have the _perception_ that GNOME 3 is designed for netbooks or tablets or phones, or to suggest that its design may have the _effect_ of working best on netbooks or tablets or phones. Like you, I'd disagree - I'm quite happy running it on dual 20" screens - but it's a valid perception. It's definitely incorrect to say that it's _designed_ for such environments, though, because it isn't. From the Shell design wiki:

"Effectively works on contemporary hardware: the Shell will provide an excellent experience on touch-based devices and will scale down to small screen sizes. It has also been designed with wide-screen in mind"

note "will scale down to" small screen sizes, i.e., it isn't designed specifically to them.

Touch was certainly a consideration in the design of the Shell (pages 4 and 34 of the original design doc) but I, frankly, thought the actual Shell would be a bit more... dynamic with regards to screen size. The exact same layout may not be optimal, and possibly cannot be, for vastly different use cases, sizes, devices, etc.
Though my monitors don't support such now, I plan on getting a couple high res monitors (think a couple megabytes or more), and I honestly wonder how well the limited number of ways one can manipulate windows will work (a simple tiling scheme would help here greatly), along with the already long distances my mouse has to travel if I am simply in research mode (i.e. mostly reading various docs/sites).
From personal experience I will say this: GS isn't designed with netbooks in mind. On my Acer Aspire One, the usable vertical space was simply too little. Several dialogs wouldn't even fit on the screen. This will be , or might already have been, fixed but it goes to show that netbook-sized screens were not foremost in the minds of Jon, Jakub and the rest.

On the giant icons:

"The avoidance of exclusive application categories and nested sub-menus is a distinct advantage of application launching in the shell compared with the GNOME 2 desktop. Users do not have to guess which category an application is in, and the motor control demands of the application picker are lower than those of menus. The application picker also utilises spatial memory, making it quick and easy to relocate applications."

You can agree or disagree with the concept there, but it's clear that the concept is not 'this is good for small screens'.

In general, if you read through https://live.gnome.org/GnomeShell/Design/ and sub-pages, it's pretty obvious no-one was designing exclusively with small screens in mind.

The giant icons are pretty obviously a bad design call. It's bad on phones, tablets, and it's bad on desktops. The reasoning on this point is so bad one could drive a chunnel digger through it and ne'er scrape the sides
Once you add a few more applications the giant list of icons just becomes unnavigable. The same thing happens on my Android and it is simply a massive mess. I don't know what the application is called that I am looking for, I just know I want to do X. On the desktop space I am more comfortable but the average non-oss user wouldn't be UNLESS they used the search feature and used the correct keyword (another reason to incorporate Tracker's more sophisticated stemming). But what about just using the mouse? Now you are forced to examine each of the icons and see if they adequately explain their apps functions (good luck figuring out what transmission does from it's icon...not that the icon is bad but it is simply a difficult idea to encapsulate in a small image).
I'm assuming you've been following the development fairly closely so I imagine you recall the issue of the application names in Overview from a few months back? If not, briefly the issue was brought up that a number of application names were simply too long for the space given to them as delimited by their icon in Overview. There was concern over ellipsizing, using more than one line if needed, I seem to recall abbreviations (???) being mentioned. Release was fast approaching so they needed to make a decision and giant icons was it (gnome.org seems to be down so I'm having problems finding the exact report I was talking about). You can tell by the reasoning given above that this wasn't a design decision that had a particularly strong basis in usability. They state that the lack of categories is a "distinct advantage" but that MIGHT be true in only one circumstance: they knew the name of the app they are looking for. If they know that much then I think they know what it does hence what category it is in (obviously fdo's categories aren't perfect but they can be dealt with and, perhaps, altered as needed to reduce ambiguity, besides, how many people were clamouring to the gnome mailing lists/bug tracker saying they couldn't find the application they were looking for b/c they couldn't figure out which category it was supposed to be in?). Realistically, if this average user knows the name of the app, then he probably uses it enough for it to be amongst favorites which makes it even quicker to find! As for "motor control", it is good they are thinking of those who don't have such fine motor control, but how are those people supposed to get into the application list to start with since the Application "button" is so relatively tiny? Again, that is clearly an ad hoc argument for giant icons.
Lastly, spatial memory. That is something written about, I believe it was Jon, and it was said that since spatial memory is really only useful for a few items it is not a scalable basis for design. Obviously there is something to that but even assuming the giant list of icons made use of spatial memory (it doesn't since the list requires you to scroll, something paging will help with, as you add/remove applications things move around, also there is simply the issue of scalability), spatial memory was expressly mentioned as being a bad idea for a general design principle. Clearly either I am recalling something incorrectly, or this was simply a bad, or at least inconsistent, argument.

I hope this doesn't give the impression that I am "anti-Shell". I think a change was desperately needed and incorporating some ideas from the new area of touch devices was a smart idea. Frankly, I just wish they had been more ambitious, but I realise time/manpower is limited and better to release something to hopefully move UX forward. I've been very happy with the level of polish GS has (amazing strides were made in the last few months), but somethings are clearly problematic and need to be dealt with.
http://www.youtube.com/watch?v=SrPgY...layer_embedded is the best category based menu I've seen. Something of similar to this would be an improvement and look nice as well

Best/Liam