Slashdot Mirror
A Look at the Upcoming GNOME 2.6
unmadindu writes "GNOME 2.6 is just around the corner, and I figured out that many GNOME users would like to know what's in store. So I installed GNOME 2.5 (development version for 2.6) in my box, and came up with a list of the new stuff that are coming up. (and just in case, copies of the article are also available here and here)."
I'm still trying to get a 2.4 stable version working. Toooo many new features too soon.
Is there is stripped down version gnome?
As much as I like the idea of it being smaller and faster, it's still kind of strange.
The windows take up 1/4 of a 1024x768 screen. I don't want to have a bunch of gigantic nautilus windows filling up my small screen.
Part of the problem is that developers, being geeks, tend to have all the latest kit. So the GNOME hackers working on their 512M / 1G 3 GHz box won't be concerned about performance, but the millions of desktop users running lower-spec machines will.
Let's be clear about this: the vast, VAST majority of machines on the planet, in homes and in businesses, have 32M, 64M (and occasionally 128M) RAM. That's nowhere enough to run GNOME/KDE, OpenOffice.org and Mozill at a realistic and usable speed. When did we become just as bloated as Microsoft?
If the GNME developers don't step back, look at the problem and concentrate on efficiency and clean design (rather than flashy features and bloat), it'll lead to long term damage for Linux on the desktop. They're doing a great job bringing Linux to the masses, but the masses are going to be less enthusiastic about Linux when it keeps requiring hardware upgrades...
But the performance gains from highly-optimised compilation are negligible at best -- around 3-5% for the majority of apps. In some cases, it makes them slower (-O3, function-inlining, increased code size and more CPU cache misses etc.).
Going to Gentoo won't suddenly make gconfd eat up half the RAM, or OpenOffice.org and Mozilla to run speedily together on a 64M box. It's not so much about CPU usage (as most of these apps spend their time waiting for user input), but outrageous Microsoftian memory usage.
That's fine for us geeks, but it's terrible for businesses, curious home users and third-world countries. We shouldn't force an upgrade path; yeah, you can use AbiWord and Dillo and IceWM etc., but no mainstream distro has such a fast desktop as standard, and they're lacking in features.
Bottom line: if a company has to splash out money to buy more memory or new boxes for the "Linux migration", they won't be so chuffed (and may end up sticking with MS for a bit longer).
And don't underestimate Microsoft. They improved stability in 2k, they improved boot time in XP, and they could improve elsewhere. We don't want to have the most memory-hogging, slowest-booting and bloated OS around...
I've always thought that the reason having two (main) desktops (KDE and Gnome) is good is not necessarily because of the competition, but because there is a need to interoperate between the two, so sensible 'generic' programming interfaces need to be created. This should create more modular code, and modular code makes successful open source projects.
However, to what extent is this true? Can I, for instance, use just the Gnome file manager in KDE, and vice-versa? Is it an aim of these projects to make this level of interoperability a goal?
It'll be interesting to read a decent "neutral" KDE 3.2 vs Gnome 2.6 article though! And it also has to be said that the competition between KDE and Gnome really had driven both communities to excellence. Als competition has not deterred them from cooperating in freedesktop.org - something to be encouraged until hopefully one day somehow the libraries can be unified.........
Of those to whom much is given, much is required.
It is certainly not just geeks who will want or need to type in file names. Skilled typists will not want to move their hands from the home row to open a file. Making them use the mouse to open a file is a bad idea.
It's important to remember that some users are much more skilled at using some aspects of the computer than are developers, and that "easy to learn" is not the same as "ease of use."
No, the majority DON'T though (Google for surveys etc.). Yes, the majority of OUR boxes have, because we're computer geeks on Slashdot, but you're totally ignoring companies and poorer countries.
If RH, IBM, Sun and co. could go into a company and say: "Don't spend money on hardware upgrades for XP/2k, just install our Linux and save!" then we'd be sorted. But the average box running NT4 or Win98 is nowhere near capable enough to run a modern desktop Linux, so companies have to buy new hardware anyway. And if they're splashing out on new boxes, they may as well stick with Windows for the time being...
It's like the recent Slashot story on MS EOLing Win98, in which loads of people suggested the company was doing that to stem the tide of Linux. No way! The average Win98 box has 32 or 64M -- try running Fedora, Mandrake or SUSE on that with modern, fully-featured apps. It's horrendously slow. And yet, shouldn't the Linux community be HELPING these people, making old hardware fully usable on the desktop? Sure, they can become servers, but we're bloating the desktop right into Moore's Law.
Have to say it, this was one of the best written personal review article submitted to slashdot in recent past.
It covers the functionality well, does not break the continuity and was fun to read.
If only we had more articles like this, slashdot might gain few more subscribers.
- mritunjai
That's the same argument Microsoft used to say that Windows 95 is faster than Windows 3.1. And on a system with plenty of memory, it is. But most people's experience with the hardware available at the time was that Win95 was much much slower, thrashing horribly with less than eight megabytes and still rather uncomfortable with less than sixteen.
Making a program twice as fast in CPU time but at the expense of using twice as much memory may not be a good trade-off. If you start running low on memory then you get a very steep performance drop from paging to disk (or not having enough RAM for disk cache, which is effectively the same thing). The most important benchmark is how it performs on a machine with, say, 64 megabytes of RAM, or whatever minimum level you want to require. Not shaving a few fractions of a second off times on recent hardware.
-- Ed Avis ed@membled.com
According to Google's Zeitgeist", Windows XP represesnts 45% of the market out there (well, of their customers/users). Windows 2000 represents 18%, and although it will run in 64MB, I don't view anything less than 128MB realistic. Therefore, I would guess that the majority of people are already using machines with 128MB or more.
I would assume it's because nautilus is a lot bigger than a gtk file selector. Anyway, a file selector is still required because people will choose to run your apps while the whole DE is not running. For instance, I run a number of GNOME and KDE apps on XFce4; I may have konqueror installed, but it never runs and I certainly don't have nautilus installed. Even if they were installed, if they were required to do file operations from Cervisia or Gnumeric I'd have to wait for those browsers to come up from a standing start when all I wanted was to open or save a file.
"Nothing was broken, and it's been fixed." -- Jon Carroll
GNOME vs. KDE will perhaps be one of the holy wars of this millennium,
Yes, and like most holy wars, it's about obsolete ideas. Gnome and KDE are both serviceable desktop environments, but let's not kid ourselves: imitating Windows and MacOS should not be the future of computing.
Personally I'll keep Mac OS X on this for the moment, if only to avoid kernel recompiles and incompatibilities arising from that,
Whatever makes you happy, dear. Personally, I dumped Mac OS X because I got tired of the manual upgrades and install hassles; Debian has been much less effort to maintain and has a lot more software available for it. And kernel upgrades just work, with no recompiles, with Debian.
Did you forget everything else? Antialiased fonts, a configuration system which supports multiple backends (XML or LDAP or anything you want; important for businesses), heavy use of graphics that have millions of colors, bigger screens, more advanced underlying architecture (important for third-party developers!), MIME type sniffing, etc.
Yes OS/2 Warp did some of that but it also looks bad by today's standards and isn't nearly as advanced or polished.
Did they REALLY make a file requester widtout pattern matching? WHY? Even windows got that feature, and it is so usefull that there are NO reason to leave it out.
Since you seem to be part of the team, I'll bite.
/usr. I'm not even argueing that it should be a click and drool install, just that it should be possible. On top of that I find that any good installation documents are non-existant, I'd be more than happy with a simple "install this first, then this, then that, which satisfies X's dependencies..." If I could install it I'd be using GNOME right now, but since I can't I'm using KDE 3.2. Konstruct alone makes KDE kick GNOME's butt.
I've been trying to install gnome off and on for the past month and a half. Problem is that it's a major bitch to install, even with neat little shell scripts like CVSGNOME I run into at least a dozen quirky dependencies. Each dependency is hell and a half to get working, and requires at least an hour googling and tweaking to get running.
The first time I tried to install I got up to the point where I needed some XML library, which needed a font library, which needed font config, which needed three other libraries. Got the three libraries working, tried font config. Came up with a build error. Edited the source files myself. Recompile. Came up with another error. Asked about it on the font config mailing list. Was told "get the CVS version". Done. Tried running the shell script that generates the configure and makefiles. It crapped out. Spent a good hour modifying that script trying to get it working. Nada. Finally gave up.
This last time I stopped when it got to the XML DOC DTD so I could install GTK DOC or something of the sort. Spent four hours trying to get it to work, after posting on several boards asking for help when I was done banging my head against the keybard.
My bug report - It's design is based on a virtual clusterfuck of dependencies and is impossible to install, especially in my user space since many of the programs and scripts assume that they'll be installed in
And please don't mention "just update your distro/use Debian/etc", I'm using Linux from scratch. GNOME is the only program I have found to be totally impossible to install.
Isn't Evolution a mail reader and not a file browser ? Or did you mean Nautilus ?
Or you could leave both options open and let people use whichever they want. Like it's done now.
Besides, if I have both mplayer and xine installed, how does the One File Browser know which one to launch ? Or Emacs and Vi ? Or whatever ?
And yes, I realize you can set this in preferences; but suppose you want to use different tools for different tasks, despite the file format being the same ? Or if I just want to try out a new program ?
Because that would mean resizing application windows to fit them besides the directory windows, and be a lot more hassle than simply using a selector window ?
No, it's a useability feature. Lacking a separate file selector would give users unneccessary grief.
The more features you bundle into a single program, the less likely it performs any of them well, simply because different features (such as useability and low learning curve) conflict with one another.
Forget magic. Any technology distinguishable from divine power is insufficiently advanced.
None of the things you mentioned jutify the outrageous memory requirements. Heck, RISC OS had full anti-aliasing on 30 MHz and less machines.
I'd rather see new features being implemented efficiently and carefully, helping to make more of current hardware. If you're happy buying new kit every couple of years for the latest stuff, that's fine, but it's so Microsoft-esque. Some of iuus were hoping that Linux would get us out of this constant treadmill.
This new spatial apperance of the new Nautilus reminds me of old MacOS finder. I liked it back then and I will probably like it in Nautilus.
But I am a bit worried, some folder hierachies in Unix is quite deep.
Perhaps they should introduce something like the Mac spring loaded folders.I.e. if you want to move a file down in the hierachy you just drag and hold it over a folder, after a short while the window opens, and you hold the file over a folder in that window, until that opens and so on. When you finally reach the right folder you drop the file, and all windows you encountered on the way is closed automatically.
God is REAL! Unless explicitly declared INTEGER
Not trolling or anything, but here goes...
As a developer, I have always been interested in writing software for Gnome since 1.x. The one thing that has really set me back from doing so is the fact that with each and every iteration, something in the very core of Gnome changes and more often than not, those changes mean that you would have to recode large chunks of your software to cope with the changes.
Yuh, sure all your Gnome 1.x apps will still run but it won't be able to use any of the new features in 2.x. This comes naturally, since this is after all a "major release" upgrade. They've really done it with 2.x this time, something major changes with each "minor" version is released. I know this is all about bringing Gnome closer into the "integrated desktop where you have everything you need to do everything you need" that it is trying to achieve.
Case in point, this whole new-fangled "Object-Oriented" metaphor. Now not only do I probably have to learn a whole new set of interfaces to get desktop integration going for programs that I write for Gnome, I also have to learn how to operate this contraption. I mean come on! Do we really need all this HIG crap?!? My UI was "usable", at least for me, before all of this HIG things were implemented. If the developers want to implement this HIG thing, then go ahead and do it but it would also be nice to let users with "bad habits" choose to revert to the old UI behavior when they want. And for heaven's sake, leave the API's unchanged until the next major release! Being a developer for Gnome is a lot like being Sisyphus.
Now I realise why there are more apps written for KDE than for Gnome.
</rant>
Yuh, I know this rant probably doesn't make any sense to you. But maybe that's because you haven't been around when Gnome 1.x was new and Miguel was still sane.
(puts on asbestos underwear and ducks under the sink)
Maybe they never used Windows 95 either!
... and of course, the option to disable it completely if you want an explorer like UI.
True Spatial Navigation is quite good really. As long as you have options for opening the parent folder, and autoclosing the parent folder when opening a new folder to keep clutter down.
I use ctwm on a pentium 120 with 16MB RAM, its perfect for such a machine.
dillo for browsing and nedit for editing works fine on it. For real text processing even abiword is workable.
You can use such a machine perfectly well, you just have to be picky about what you use on it.
Jeroen
Secure messaging: http://quickmsg.vreeken.net/
I think this is a really the wrong approach. Designing primarilly for new hardware hurts lots of end users, especially in poor countries. You would also be surprised to see what is still being used on the desks of most corporate users.
Core software designed for a broad audience needs to be designed for the middle ground. Not the lowest common denominator, but not brand spanking new. You then need to make it tweakable (either run-time or compile-time) to make it acceptable on low end hardware, and run well on high end hardware. Personally, I think the middle ground is a low lower than most people think.
I expect make buildworld and portupgrade gnome2 to take a long time, but the performance hit with Gnome 2.4 really shocked me.
It's open source. You have the power. Think outside of the proprietary software box!
I find this a little condescending. I started using linux in late 1991 dual-boot and by mid 1992 my machine was linux only. I use FreeBSD now, but I am well aware of how all of this works.
However, I am also a professional developer with a 9-5 (ha!) job, but I also have my own open source project, other extra-curricular professional activities, and a family.
Where do I dedicate my time?
General OS
User-interface
General applications
My project
Developemt enviroment that my project relies on
When I was a college student and single, I was able to dedicate lots of time to the early linux development, but not I have to cut back and focus on my project.
(S(SKK)(SKK))(S(SKK)(SKK))
Why not? Just because it isn't what you would usually use. When I am forced to use a Windows machine the first thing I do is change it to single click activation.
I actually had several people making fun of me because I used Windows in this way. I just can't understand it... Why click twice when you could click once? Your mouse button lasts twice as long. I guess I am just lazy.
Umm... the activation doesn't occur until you release the mouse button, so if you click it and hold it, you can move it as you please. Once you start moving it, when you release the button it drops the icon. How does this make it not easy?
Here's why. We need to get away from this obsession with 20+ year old graphics servers running 20 different inconsistent toolkits, all working extra hard to "work well with each other" when we should really be focusing our efforts on one major desktop competitor running on Linux.
"Sufferin' succotash."
It seems that users these days are going more the other way. Most Windows users (no idea about Mac-heads) seem to not realize that they have a file browser. It really comes down to whether the modern GUI is application-centric or file-centric, and I can see that for most basic home users it is heading towards application-centric. Unless people know they need to put something on a floppy disk or usb drive to take with them, it seems that in almost all cases they just save to the default location or a subdirectory thereof. IMHO, this is a good thing: It means all their files are in one place to be backed up, and that if their profile/home directory/whatever is stored on a server that they don't have to worry about saving their files someplace where they can access them from any machine. Unless they need to transfer a file somewhere, in which case they resort to using a floppy or emailing it to themselves. Most of Thomas' points are already dealt with in some manner, or a matter of personal opinion. If he had used Word in the last 7 years he would realize that it already auto-saves after the first time. If he had used a non-Mac computer he would realize that increasingly programs are heading towards an SDI instead of MDI. No more quit to close programs, just close any open windows! As for the issue of copying/moving files around: This is a symptom of another issue. The core issue that needs to be resolved is that you have to move your files around to have access to them everywhere in the first place. Different people have different ideas, one being a portable personal server that has all your files on it. The other idea is Internet connected file storage. Both have their own ups and downs but I believe that soon one or the other will become mainstream.
I think a more progressive model for a user interface is a more verb based one if you will, rather than noun based. With cheap, fast, pervasive Internet connections it's no longer important where your files are (or hopefully it won't be soon). Its more important what you want to do. For example, say you have a JPEG image: What you want to do with it is more important than what the file is named/where it is. When you find the file that you want among an assortment of different files, you still have to choose whether you want to edit it, view it, add it to another document, etc. I would argue that having to know where all your files are is the outdated cruft, not having open/save dialogs.
Personally, though I think that more choices for ways to do things is usually better, as long as one doesn't get in the way of another. There are definitely situations in which I find it easier to use a file browser than to use an open dialog.
Ok, this is kinda fragmented but I have to go to class now.
-Mr Lizard
^I'm with stupid.^
2. If you are advocating a single desktop that runs on Linux, I guess you want to get rid of FreeBSD etc?
3. What happens when the desktop project starts being unreasonable? The only reason we aren't all stuck with XFree's tantrums is that competing projects like fd.o give users leverage. Do you really think that the XFree team would have backed down on the license change if they didn't think everybody would switch away?
4. There's no clear "winner". KDE and GNOME have different design goals, and lots of people agree with KDE's set of design goals, and lots of people agree with GNOME's design goals. You'll never get consensus, and you don't provide a reason for getting rid of one of them.
5. You state that X/KDE/GNOME are dead in the water, when in fact they are moving along at a rapid pace, and then point to Y, something that is barely out of the vapourware phase.
6. You state "we should be focusing all of our efforts...". Exactly what effort are you focusing? Or are you just an armchair know-it-all?
6 points, and you get called a troll. Bye now.
Why is it that drag & drop fanatics are always trying to force their preference to everyone else?
EVERY TIME there's talk about file selectors, someone pops up and seriously suggests an option that not only encourages the need to use mouse, but actually requires it.
Especially for saving... instead of hitting ctrl-s (and quickly typing a name if it's the first time), I'm supposed to a) resize application window, b) locate file manager from open windows, or open one if it isn't running, c) drag icon somewhere? Excuse me, I think I need to puke.
Certainly none that matches the convenience of the Filer windows in RISC OS where you would double-click (or drag) a file to open it, and drag from the application to the Filer to save
:wresult3.txt, and in many contexts I'd be right.
That's subjective... I could claim that needing to drag an icon from a text editor to a whole other window (which I'd have to find and make visible first) is painfully slow compared to
That points to one big advantage of the dialog box approach: keyboard compatibility. Desktop environments which offer DnD should provide some (optional) way to perform equivalent actions from the keyboard, but I'm not aware of any having done so.
Digressing down that topic:
There have been some small steps made towards keyboard-controlled DnD, but I haven't seen any adequate yet. Of course, some systems let you push a button to steer the mouse from the keypad, but that's too awkward to consider. Some file managers include an abuse of the clipboard metaphor (like a Copy button which makes a "shallow copy", instead of a "deep" one like every other Copy command besides Excel) to provide features that could be better solved by enabling DnD via keyboard. There are assorted taskbar-applets which provide a "shelf" to set down an icon in mid-drag; enhancing one of those to be controllable by keyboard would be the most direct implementation of a solution.
I should note that I am not at all surprised someone smoking crack modded me down as "Troll." What makes Linux so great that it's above criticism? XFree86 for that matter? Here on Slashdot, if you dare criticize something and hold a differing opinion than the majority, you get modded down.
I'll take your post point-by-point.
1. Just because something is 20+ years old, it doesn't mean it should be thrown away.
If you had bothered to RTFP (read the fucking paper) I linked to, you'd see *exactly* why using that 20+ year old X technology is a bad idea.
Plenty of people like the idea of heirarchical filesystems. Plenty of people like mice, monitors and keyboards. Plenty of people like icons and windows.
I have no idea why you're bringing all that up. Those are abstract concepts and ideas, like the idea of books, pages, and chapters. I never mentioned anything about those.
The X protocol in current use isn't 20 years old. It's matured and gone through a number of revisions, just like any other technology.
The X protocol *is* 20 years old. It supports extensions really well. As the paper outlines, many of these extensions are breaking other extensions. A typical desktop runs about 26 extensions. Not to mention that adding certain graphical effects requires changing a lot of architecture in the source code because of the way XFree86 is set up.
2. If you are advocating a single desktop that runs on Linux, I guess you want to get rid of FreeBSD etc?
Port it to BSD if you want. That's not even a relevant issue to what I was talking about--removing this silly desktop "competition" going on between competing products, completely scaring away any serious company wanting to write an app for X. Who do you write for? GTK? QT? Motif?
3. What happens when the desktop project starts being unreasonable?
What if both GNOME and KDE become unreasonable? You can't argue with a "what if" because it could apply to absolutely anything. You haven't even defined what you mean by "unreasonable."
The only reason we aren't all stuck with XFree's tantrums is that competing projects like fd.o give users leverage.
It's a fork of XFree86. It still has its tantrums. X doesn't even have its own widget set. The issues with X are much more fundamental than you are discussing here.
Do you really think that the XFree team would have backed down on the license change if they didn't think everybody would switch away?
What does the license have to do with anything? I'm talking technologically and logically--XFree86 needs to be gone, and the two competing desktops need to get their acts together. All these incompatible toolkits have driven away serious desktop development all in the name of "choice."
4. There's no clear "winner". KDE and GNOME have different design goals, and lots of people agree with KDE's set of design goals, and lots of people agree with GNOME's design goals. You'll never get consensus, and you don't provide a reason for getting rid of one of them.
Application support. There. Which one should Adobe port Photoshop to? If we had one seamless desktop with proper binary installation/uninstallation routines and a sane programming library that retained backwards-compatibility with each major release, you don't think companies would take a look?
5. You state that X/KDE/GNOME are dead in the water, when in fact they are moving along at a rapid pace, and then point to Y, something that is barely out of the vapourware phase.
Vaporware? You can already download it and compile. Developers are writing the base widget set as I type this. I don't get this bizarre aversion to change that Linux users have.
6. You state "we should be focusing all of our efforts...". Exactly what effort are you focusing? Or are you just an armchair know-it-all?
I'm currently writing a design paper. Sto
"Sufferin' succotash."
Which is of course *exactly* what is wrong with KDE. You only need one completion method that works well.
"Well *my* kettle boils water in twelve different ways!"
-mike
-- "So, what's the deal with Auntie Gerschwitz et all?"
go fuck yourself