A User's First Look at GNOME 2.0

Gentu writes: "OSNews has just published a review of the Gnome 2.0 desktop environment and its verdict is not so positive. The author feels that the new version is limited in many ways and with a UI not well designed."

Comments

Some valid things, and a lot of not-so valid

[JanneM]

Some of the issues he brings up seems valid. That said, I run Gnome2 and I don't recognize many of the problems he brings up.

First, for me, Gnome2 is far faster than Gnome1.4. This goes for most individual applications, as well as the desktop overall.

Lack of options: Well, yes and no. There has been a serious attempt at providing sensible defaults for a lot of stuff, and hide away rare and/or strange options into the gconf system. While some people like being able to tweak their desktops to hell and back, for many users it is just plain confusing to have as ridiculously many options everywhere as Gnome1 had. Note that for those serious about tweaking, gconf is there for your time-wasting pleasure. :)

Gedit: I've tried repeatedly, but I am unable to duplicate the marking thing he talks about.

Galeon has continued to work flawlessly for me, as have all other Gnome1 apps I have. he mentions that he does not have a Gnome1 installation; that may be an explanation as to why Gnome1 apps do not work...

As for 'scattered settings' - huh? I get all settings neatly in the 'Desktop Preferences' menu. That certainly includes things like xscreensaver settings and pretty much everything else he gripes about in this area. I do not have a 'Desktop theme', as he seems to have, but just the 'theme' option - as it should be.

I get the feeling there is something rather wrong with the reviewers setup; something like an incomplete install, or a mix of older and newer packages or something like it.

/Janne

Ah, memories

[Gryphon]

Reading this review of Gnome reminded me of the days (about a year and half ago now) when I was still trying to use Linux and Gnome (somtimes KDE) as a desktop machine.

Nothing coherent about the UI design, hunting around to find configuration panels, getting messages that tell the user to download this package or that package (which leads directly to Dependency Hell).

These days, I use Mac OS X. Sure, it's UI isn't perfect. And I know, it's an apples to oranges comparison, Free Software to commercial. But man, do I ever enjoy using a coherent desktop with one place to change settings (System Preferences). No fuss! No muss! I'm far more productive.

And my Linux server continues to hum away in the basement, quietly powering my website.

Life is good.

Re:Talk about laugh

[satanami69]

Waimea ~1.5 seconds
FVWM ~4.0 seconds
Gnome 2.0 ~25 seconds
KDE 3.0 ~1 minute

hell, all i usez it for it to open mozilla anyway. I'll take waimea.

Quick Analysis

[Leimy]

GNOME-1.4: Still hard to figure out when you first sit down on it.... I personally had trouble changing an Emacs icon to use Xemacs and ran around looking for a "property list" for it... I think you have to manually edit some text file is what someone said... I stopped using GNOME immediately.... That's no way to do a GUI IMO.

KDE3.x: Slow... very slow. Too many virtual functions need to have code relocated at runtime. Luckilly This site is addressing some fundamental linking issues with C++ [among other things] on GNU tools. In fact the GNU tools are starting to be built with some of these optimizations now as was evident on my RedHat box at work. FreeBSD needs to try to do the same since its my main development platform [luckilly its a dual Athlon MP 1600 so *nothing* seems slow there :)]. There are also a few UI issues like the Author of this article suggested but I must say that people want a snappier [speedwise] desktop and don't want the power of an industrial strength server just to run their desktop. [note: I love KDE... I have committed code to KDE... this is as objective as I get :)]

I spend most of my time on Mac OS X. The concept of being able to run the Microsoft Office Suite [which I actually don't yet on my Mac] on a Unix environment with 75% or more of my favorite tools either in place or on their way is very attractive. Let's face it nothing does DOC like Word [thank god!] and for compatibility purposes with all of my coworkers there just isn't a real substitute for everything it does. We use the revision control built into Word and other things so please don't offer Abiword, StarOffice, OpenOffice or KWord as alternatives. You can suggest till your blue in the face but you can't make my company change its stance on what tools must be used.... Its a fight not worth fighting based on my experiences with the alternatives out there. [I write a lot of stuff in LaTeX now... then I cut n' paste to Word when I have to... Time consuming and stupid yes but I don't have Word for OS X yet... :)].

I never got around to experiencing BeOS first hand but I heard it was a thing of beauty... There has been a fair amount of talk about adding the BeOS file system to OpenDarwin's CVS but I don't think anyone has committed the time to it yet.

Advice to KDE: Please please please don't get too bloaty... [application duplication seems to be a bit of a problem there... Why does the standard source distribution have to include these things anyway?] I love IOSlaves and KParts and think they are uber-cool but the end user doesn't give a shit about any of that because it doesn't directly enhance their experience... just gives the developer a woody.

Advice to GNOME: As a developer I do not agree with C as the tool for doing Object Oriented Code... especially when the manner in which things are being wrapped up is very C++ like. GTKmm has a long way to go before it can do what Qt can last I checked so I think that if you code for GNOME and want full access you must use C [correct me if I am wrong please... its been a while and I want to be as fair as possible]. I have to agree with some of the Author's UI comments if his experience was authentic and correctly reflects the actual situation. I still think GNOME is prettier than the KDE defaults but there are very good things coming in that respect it seems from what I have been able to follow on the mailing lists. [again I am unfortunately biased due to my KDE involvement].

Advice for OS X... yes.. sometimes you just have to realize that indeed your shit can also stink. The only major boo-boo I remember was the iTunes installer clobbering some linux partitions... That was naughty but obviously not a test case for Apple 'in-house' or it would have been caught. Live and learn! I understand some people have trouble with the lookupd for OS X dropping out on them from time to time [though I haven't seen this myself yet.] but that's not really a UI comment is it? Hmm, I guess keep doing what you're doing and maybe think about allowing users to pick schemes other than Aqua or Graphite in the appearances menu. Don't rush it though... I love the quality thus far and can deal with a minimal set of choices when it keeps the UI simple and straighforward [yes I still use the single button mouse on OS X because its actually possible to do so due to a good UI design around simplicity.]

I'd invite comments and criticism if I didn't know already that I was in for it.... so go ahead and get your shot in... I don't care - its only slashdot :)

Dave

Re:Gnome and KDE are more or less the same these d

[Avakado]

I'll try to be a bit more constructive in my criticism this time:

Aqua's keyboard navigability: It's well known that keyboard shortcuts will improve your efficiency when using a GUI. Every single part of a GUI should be accessible via the keyboard, so that experienced users can be as effective as possible, using these. These shortcuts must act consistently throughout the entire GUI, and properly marked (like underlining the character that is part of a shortcut in menu items). Moving from one widget to the next, scrolling, opening menus, starting applications et.c. should all be possible via the keyboard. Text widgets would also benefit from having more shortcut keys, like ^U for "kill line", ^W for "erase word" et.c. In many of the applications of MacOS X, most of this functionality is non-existant.

Multiple desktops: it's obviousely an advantage to be able to have multiple workspaces running at once. Users not wanting this feature can easily refrain from using it or disable it (or not enable it). Aqua does not provide this feature at all.

Configurable look (themes): if you for some reason can't stand the default look of Aqua, or want any other color than blue or gray, you are out of luck. As far as I've been able to tell, there's no way to change the appearance of GUI widgets (beyond the colors blue and gray), as opposed to virtually all open source widget sets I've seen. You might argue that themeability slows down the GUI, but that can easily be resolved by providing a binary interface (i.e. styles are dynamically loadable libraries) like KDE does (and Mosfet Liquid and Keramik use).

Scriptability: You mention AppleScript, and claims it is like having shellscript for GUI. No it isn't: you are bound to use that specific language. They could easily have supplied a network protocol (like KDE's DCOP) or any other more generic interface. Since they didn't, everything has to go to this dreadful language. Any experienced programmer would instantly fear "an easy-to-use, approachable, English-like language".

Stupid messages: "You need to click here to continue" (why not just friggin' do whatever needs to be done, instead of requiring user interaction at every possible step?), "An error has occured" (did you ever hear about strerror()?) and similar. While many of these aren't severly obstructive, they are nevertheless very annoying signs of sloppy programming and interface design.

Widget usefulness: in certain applications (most notably the QuickTime player), completely untraditional widgets are used for the sake of visual appearance. Many of these widgets seem like they're designed to be handled with a physical hand, and not with a pointing device and keyboard (like knobs and switches).

sawfish 2.0

[Ender Ryan]

My experience so far has been a mixed bag. I like some of the simplification, they've tried (with only some success) to use saner defaults and make everything easier to configure and use.

The panels work ok, but there are some serious quirks. If you make a panel with no menu, and you remove the hide buttons... you can no longer configure that panel. Only way I found to fix it is to add a menu to another panel, drag it to the menuless panel, then you can use the menu for configuring the panel. A pretty large oversight if you ask me.

The other extremely annoying panel problem is... on logout/login, the panel completely forgets the order of the launchers on it! If that's not a huge oversight, I don't know what is.

Now for my biggest gripe. Sawfish 2.0. Someone was smoking some serious crack. I don't mean to be mean, but it has absolutely been destroyed. It is completely useless. It plain sucks, terribly. First of all, it's crashy, very crashy. See the bugzilla database on gnome.org, serious crash bugs in sawfish 2.0, definately NOT release material. Second, sawfish was designed with extreme configurability in mind, every aspect of sawfish is meant to be configurable, but now they have completely removed 90% of the configuration options. They supposedly tried to choose sane defaults, but with something as configurable as sawfish, that's simply not going to happen. There are some serious problems with the default settings. The new sawfish control panel... what can I say, it plain sucks. The tabs are across the top now, and you have to use the dumb little arrow buttons to scroll across the stupid things. This makes it an extreme pain to search for settings.

No favorites menu. I always found this very useful, I always put all the little utilities I often use in there. It's gone, and there is no equivalent replacement. Now your stuck browsing through the damn apps menu. A very poor decision in my opinion.

Now those problems are all extremely annoying, time draining, and basically make gnome 2.0, simply put, not ready for prime time. It's simply not release quality at this point, not even close.

There are some positive aspects though, quite positive actually. Fonts, gnome2.0's font rendering is really, really great. Fonts are rendered very cleanly, not blurry looking, and not jaggy, they look very good. Speed, despite what the reviewer was saying, gnome 2.0 is pretty speedy, speedier than 1.4. It loads up really quick, probably 4 or 5 seconds on a reasonably fast machine. The menus are much less cluttered by default, a plus in my book, they were simply full of junk before. GTK+ 2.x is much better. The default theme actually looks pretty good, file selectors work better, save dialogs don't wack the filename when you change directories(!).

All in all, I have to say that I'm pretty disapointed. It's not a lost cause, but it seems to me that gnome may be heading in the wrong direction.

And that's all I have to say about that.

And this is a surprise how?

[marm]

GNOME 2.0 has been rushed out of the door, just like GNOME 1.0 was.

The 1.4->2.0 development cycle has been a lot longer than originally anticipated, due to a big influx of developers (Thanks Sun!) and lots of core systems changing quite radically, coupled with some pretty piss-poor project management (now where have I heard THAT before?). In the mean time, KDE has been gaining a very large amount of traction as the most popular Linux desktop, and Sun has been wanting to push Solaris 9 out the door ASAP.

So GNOME had to release now, really, if they had any hope of keeping the users they have and for Sun to get Solaris 9 out approximately on schedule.

KDE underwent a similar change about 2 years ago, in the 1.1.2->2.0 transition, and not everyone was convinced then that KDE would survive, but it did, and look where they are now. Of course, KDE had the advantage of doing it first - although KDE 2.0 was far from perfect UI-wise, it had a considerable lead on GNOME in changing to a component-based architecture, so there was a very big Unique Selling Point for it at the time which GNOME 2.0 does not now have.

It took KDE 2 further major releases to turn the framework they built into a really nice desktop, and I suspect it will be similar for GNOME. The big question is whether the framework that was built for GNOME 2.0 will be good enough for their future plans... time will tell.

Personally I'm sticking with KDE3 for now. There are certainly issues with KDE, mostly in terms of speed and size, which themselves mostly stem from the choice of C++ rather than C, but these are being fixed one by one. KDE3 is now quite snappy, actually quite a bit faster than GNOME 2.0 on my Debian machine once you've got past logging in (all those double-buffered GTK+ 2.0 widgets are smooth and dandy, but they sure as hell ain't fast). Also, right now, KDE absolutely has the edge on both functionality and usability. Konqueror in particular is way out in front - indeed, for me at least, it's the best file manager on ANY platform. Nautilus is good, don't get me wrong, but Konq is breathtaking.

I'll reassess the situation when GNOME 2.2 is out. 2.2 should be the first mature release of the new framework, then we'll really get to see whether it's good enough to compete. I'm hopeful, a lot of the new framework looks good but either needs loose ends tidying up or needs someone to use it properly. Let's keep our fingers crossed. KDE is a class act to go up against though - they crank out the releases on time every 6 months, they seem to have a consistent vision of where they're going, they know where their flaws are, and they have yet to make a serious error. GNOME can't afford any more releases like this one if it wants to stay in the game.