Ubuntu founder Mark Shuttleworth announced Wednesday that his company, Canonical, will hire professional designers and interaction experts to improve the usability of the Linux desktop software ecosystem. They will work closely with upstream developers to bring a better experience to users of the open source operating system.
.. what I would like to see is way more colaboration. Freedesktop should define stuff like OK dialog button order and password stores (Gnome has its keyring, KDE Kwallet and Mozilla has its own ;( ) etc.
And there is soo much tech duplication and NIH in Gnome and KDE. (Beagle, Tracker, Strigi etc .. the list goes on for ages)
The Linux Foundation should really work on these things. They seem really kernel fixated.
Don’t wait and hope the Guadec/Akademy double event next year will fix all those issues!
But the recent advancements in Xorg development and this make the future a little bit brighter I guess.
Thanks Mark.
> stuff like OK dialog button order
there actually was a de facto standard for this before GNOME pulled a switch-a-roo for no measurable benefit other than to create a new area of inconsistency.
that said, KDE4 and Qt4 appications adapt their button order to the platform: KDE, Windows, Mac or GNOME. a technical solution to the problem, for sure, but it proves that its not only possible it works pretty well.
> The Linux Foundation should really work on these things.
as long as they work with the rest of the community. i’m a little concerne with this announcement that Shuttleworth is tap dancing to work but all by himself and to a tune he’s not letting others listen to.
organizations like LF an Canonical can provide a lot of value, but if it’s done without working with the rest of the community it’s value wasted.
people working on the Linux kernel and other server side bits tend to understand that … why it gets lost as soon as we hit the desktop? well, i have thoughts on that, but i’ll keep them to myself for now =)
> And there is soo much tech duplication and NIH in Gnome and
> KDE.
there is a lot *less* “NIH” in kde4 than there was in kde3. in part this is because the rest of the ecosystem is maturing and gives us more options, in part its because the KDE project recognizes the value in using what is available.
as for strigi, we actually tend to go through Nepomuk these days and when strigi popped around there really weren’t any good alternatives. “building what needs to be built” is not NIH.
moreover, Strigi is not tied to KDE in any way. (which actually makes things a bit harder for us at times..)
> Don’t wait and hope the Guadec/Akademy double event next
> year will fix all those issues!
it won’t. it’s a step further down the right path, but it won’t be a silver bullet event.
there actually was a de facto standard for this before GNOME pulled a switch-a-roo for no measurable benefit other than to create a new area of inconsistency.
Im sorry but you are wrong, there weren0t even a standar when GNOME implemented it, they just followed the Mac OS style, KDE followed thw Windows style.
BTW? What the hell is the KDE code of conduct for?
You are still trolling with or w/o code.
I was so righ when I predicted you will be the first one to break it.
Yes there was. There were Unix and even Linux desktops before KDE and Gnome came along, what with CDE, Motif applications and guidelines et al, and the accepted standard was left-to-right as it was on Windows. Many of those applications still exist and many people run them still under Linux systems. You see, some of us actually remember this stuff?
Now, what then happened was that some Gnome developers came along, looked at the Mac and said “Goodness, that’s what we want to be. I’m moist!” and so the right-to-left button ordering was implemented with no regard for previous Unix desktop history or even with any evidence whatsoever that the change improves usability. There is still no such evidence. It was simply pulled out of the Mac’s UI guidelines, shoved into Gnome’s UI guidelines and implicitly accepted as fact. The Mac approach isn’t better, nor is it wrong. It’s just different, and that’s what fails to sink in still. That’s before you even bring locales into it……….
It’s nice to see that Mark Shuttleworth is continuing in that fine tradition of Mac worship, and neatly painting over the really critical things such as attracting developers to Ubuntu, creating a software development target that developers will want to go for and allowing Ubuntu users to install software in a sane manner. But, whatever.
You think that was trolling? Bugger. Mind you, there are some people around here who either can’t accept the truth, or worse, they think the Linux world and its Unix heritage started when Mark Shuttleworth started throwing CDs at people.
Edited 2008-09-12 20:29 UTC
I couldn’t agree with your statements more. And, speaking of Apple worship… There seems to be an awful lot of Ubuntu worship in the FOSS community these days too.
There was evidence, but it is just hearsay. Apple spent a buttload of manpower and money in usability testing back when they started growing in a massive way. Some of the tests involved the ok cancel thing.
The apple way is that input dialogs shouldn’t be using generic words like OK and Cancel. They should be using verbs describing the action about to be performed (Save / Don’t Save). This works against people developing patterns of clicking OK as soon as they see it, and fits in well with the public domain usability knowledge we have so far. The reason that Save is always in the right hand corner is for consistancies sake, if you use a system designed the mac way you know that the default choice will always be in the same place, no matter what the words say or how many options you have.
By contrast, there has been zero research done anywhere on OK / Cancel. The reason that it was chosen was because that is how one would say it in english. The reason that most dialogs are Ok or Cancel is because those are the default choices.
Even though the apple research was proprietary, the reasoning behind it makes alot of sense.
Actually, dialogues should not be constructed in a way that button ordering is a problem.
Dialogues like:
Do you want to continue doing X?
(OK) (quit) (cancel)
Should instead look like:
Process X is still running, Do you wan’t to:
(Quit X) (Continue X)
The most probable choice should be the default action, unless that action somehow could cause irepairable damage.
I think this was the idea behind the change in Gnome.
If you just have OK and cancel buttons, people tend to press OK without giving a thought to what they actually say is OK.
Thanks for the insightfull answer, Aaron.
Yeah, I remember. And I wasn’t blaming KDE
Yeah, that is probably the way to go. KDE adapts, because Gnome cannot/or is not willing to ;(
But in the end it will be a testament to KDE that they can swallow a little pride and do what is best for the free desktop, although they had most of the technologies first.
I think it is because Gnome has a lot of corporate people with an agenda working on it .. way more than KDE.
OK. I probably did not pick the best examples, but it just feels like there are lots of things that might work for both desktops.
At Guadec2008 someone promoted C++ for Gnome development. Maybe if they would adopt that it would keep them from reimplementing a lot of things KDE already did in C.
That would be sad ;(
Isn’t it strange that Gnome was built by two guys who had close ties with Microsoft long time ago !?
I trust Mark, that’s all.
It’s about time somebody had the guts to do something like this.
Go Ubuntu, go !!!!
But how? First, people have pointed out that Freedesktop.org is “just a website”, not a standards body. To define a button order, both camps will have to come to an agreement. But both thinks that they’re right and the other is wrong. What do you do then in such a situation?
By the way, GNOME’s just following the Mac button order. And they’ve actually moved away from “Yes-No-Cancel” buttons a long time ago. GNOME apps usually use action verb buttons, e.g. “Cancel – Don’t Save – Save”, where the order doesn’t matter as much. Usability studies have also shown that people scan the right bottom area of a dialog first, which is why they’ve chosen to put the most appropriate button there.
Just as an additional note on this topic:
the GNOME Keyring maintainer and the KWallet maintainer have already started to work on a unified solution, see https://bugs.freedesktop.org/show_bug.cgi?id=16581
The Linux Foundation could probably employ one or more of the Freedesktop.org sysadmins so they have the time to actually respond Free Software desktop developer needs within reasonable time.
Right now the latency between responses is measured in months, e.g. have a look at the link above
Not their fault, this is what their members are most interested in. Once more desktop oriented members join up, this will very likely shift.
With all the advances in Linux over the years, I still can not believe that they still have such glaring problems such as the menu, and the lack of a control panel.
The menu system in Gnome / KDE is still what I would consider pre windows 3.1 level. Why can’t you just drag an application into the menu. Windows has had that since what, windows 95!!!. Why do I have to manually edit .desktop text files to change a menu, this is ridiculous.
The only other thing Linux / Ubuntu desperately need is a control panel type application where ALL system settings can be changes. Windows has had that since what, windows 1.0!. They also need a device manager type thing, Mac, Windows both have them, why do I have to dig through /proc to find hardware info.
Kernel binary compatibly would be real nice also. It is pretty lame if I install a new security patch, I have to re-compile my video card drivers.
“The only other thing Linux / Ubuntu desperately need is a control panel type application where ALL system settings can be changes”
You mean something like YAST?
YAST is SUSE only.
Garbage likes garbage.
“YAST is SUSE only.”
He said Linux/ubuntu . Plus there is nothing Stopping other distros from integrating it.
“Garbage likes garbage”
lol. I dislike yast but I would hardly call it garbage, it does what it sets out to do.Suse is really quite good these days and there is no better kde distro than suse.
Edited 2008-09-12 19:36 UTC
Well, I wouldn’t count Mandriva out – it is still very much a fine KDE distro. Also, Suse and Mandriva are also good Gnome distros. ;}
SUSE and Yast are good, they are not “garbage”. You’re carried away.
Control panel (Control Center) is available in Ubuntu Linux. If it doesn’t appear on your System> Preferences menu you have to activate it. Right click on System toolbar. Click Edit Menu. One in Menu editor, go to System Preferences and make sure there is a check mark on Control Center. Done.
Someone mentionned YAST, but The Mandriva control center is worth to talk about too…
There’s many many tools in there…including parental control, setting up web,ftp servers etc.
regards
YAST sucks, and as already pointed out, SuSE only. I have many issues with Yast / SuSE although I will say 11.0 is pretty slick.
I don’t understand the control panel thing. I have always thought the idea of a control panel stunk. We already have the majority of tools under the Administration or System menus.
What would be nice is integration with more hardware into the gnome-system-tools (or whatever KDE uses). Something like a Udev / dbus / hal configuration frontend. That would solve the hardware issues. Also something like UPS support under the Gnome Power Manager would also rock. In fact if I get time today, I’m going to put that in a request.
My bad, apparently gnome-power-manager HAS UPS support already, I’ve just never seen it, ’cause my UPS is plugged into my server, which of course doesn’t have gnome on it…..
The menu system within KDE and Gnome are far superior to the current system in Windows.
I will call this a troll, since both KDE and Gnome have menu editors and I have only once edited a .desktop file and that was between 2.16 and 2.18 of Gnome.
For the record, I also think the whole idea of a ‘control panel’ sucks. It’s much easier to read an alphabetically organized administration or preferences menu, than it is to read a big window full of icons. Even if those icons are alphabetical. It also helps if they are named appropriately. Everyone was so used to Add / Remove Programs under win9x to winxp, that when they changed it to Programs and Features under Vista, it thoroughly annoyed me. Changing, just for the sake of change is retarded.
I still haven’t found a standard OK / Cancel order within KDE dialogs. It always seems that it’s opposite of what I’m used to, but then again, maybe that’s because it IS opposite, because I use Gnome.
I had high hopes for KDE4, but it annoys me so far even more than KDE3. I try every version, usually even the x.x.+1 versions just to see if they’ve fixed enough things to make me want to try it…. I get as far as changing the clock to 12h format, and then realize that not only has KDE4 made it even more obscure where the setting is, but that they still make it over complicated.
And I’m still confused at what the purpose is of being able to rotate your desktop icons in one big tab….
Not to degrade what great work has gone into KDE over the years, but I’ve always had this feeling that they were trying to be too Windows-like, and that’s the reason I switched over to Linux in the first place, because I hated the UI of Windows. Yes, I made the switch long before viruses were such a huge issue, and spyware hadn’t even been heard of.
If you use applications ONLY from synaptic, then you do not have to edit .desktop files, but try to use an app that is NOT on synaptic, such as eclipse 3.4 (yes I know eclipse 3.2 is on synaptic, but I need 3.4).
And yes, I had to edit .desktop files just to get a menu item. On the Mac, I just drag it to the dock, thats it.
I think Linux in general really needs a some standard way of dealing with applications, I think it is nuts for someone to write an app, then have to wait for the package maintainers to create a package for distro ‘x’.
And no, this is NOT a troll, I like Linux, I just point out some flaws that annoy me. This I think is the other fundamental problem with Linux in general, if everyone does not agree with some comment 100%, then they are trolls. This attitude has to go. Face it people, nothing is perfect, not MacOS, and not even Linux believe it or not.
As long as third parties insist upon using installation methods outside of the normal mechanism for the OS, in this case the Debian family, it won’t integrate well with the OS. If Eclipse is not using debs and rpms it should.
Unzip an archive into a directory in Windows and then look for the program in the start menu. Doesn’t work. Obviously Microsoft needs a better way to install apps before it can go mainstream, right?
It’s 2008 and dumping a tar ball onto an ftp server should not be considered a proper “release” for OSes which are supposedly supported. It would be preferable if they put the release into a standard or 3rd party repo, so that security updates could be automated. But if they choose not to, a deb would do just fine. But they cannot operate completely outside of the distro’s application management framework and expect seemless operation. Duh?
Application providers need to use what is freaking out there for them to use and has been for years and years.
Edited 2008-09-15 13:52 UTC
It still doesn’t explain why he had to edit a .desktop file.
The only reason I can imagine is that the .desktop file as shipped with this new version of Eclipse was broken, but a broken installer will fail on any platform.
It really doesn’t matter in which way the software has been packaged or how it was installed as long as the content of the package are correct and don’t have to be edited after the installation.
Well, menu integration does work even when using a third party installation method.
In the case at hand it didn’t work because the software bundle itself was faulty (had to be manually changed to become correct)
It most certainly does. Here is the Eclipse installation procedure for Linux. Installation is as easy as 1-2-3:
1. mkdir eclipse
2. cd eclipse
3. tar zxvf eclipse-SDK-3.4-linux-gtk-x86_64.tar.gz
And you’re done!
That’s why I said that dumping a tar ball on an ftp server shouldn’t count as a real release today in 2008.
The real problem is that we’ve been treated this way for so long that we simply expect this kind of treatment from 3rd parties. IBM should know better. And users should complain to the software provider about their shoddy installers, or total lack of an installer.
First, I wouldn’t call that an installation independent of the platform, that’s exptracing an archive, and second the other poster wrote he had to edit a .desktop file, not to copy one from the “installation” directory to one of the appropriate target directories.
I see, that part previously didn’t make any sense to me since I was assuming any form of actual installation.
I am not sure, because I definitely wouldn’t expect this from any 3rd party.
I remember using small tools on Windows which were just a single .exe file and primarily launched through file association, but I wouldn’t expect something as complex as an IDE to be just runable from an extraction directory. Ever.
Absolutely!
Must be one of the pitiable cases where software vendors coming from the proprietary Unix times haven’t updated their distribution practises yet.
I agree that a better centralized (GNOME) “control panel” might make sense. GNOME and Ubuntu try to collect configuration apps under certain menus, and one can configure things in GNOME using many handy GUI apps even now – but a centralized configuration panel could unify things and ease usability. If you need to browse different configuration settings, why would you have to open the main menu and make several clicks several times instead of just opening one centralized configuration panel and finding what you need from there?
It could be the same if you had all your tools lying around home or even in only one room versus having them all neatly arranged in a well-designed tools cabinet. Collecting tools and things together that are related to similar purposes makes sense (just look how shops arrange their shelfs, for example).
SUSE YAST is just one example of such a centralized control panel app. Maybe most advanced GUI operating systems and desktop environments have something like that because it just makes sense and makes users’ life much easier. For example, the Xfce desktop environment (maybe the closest competitor to GNOME) has its own centralized settings manager too even though being more minimalist and lighter alternative to GNOME otherwise.
I will believe Mark Shuttleworth once we see the code and it is being integrated upstream.
Before then, this is just marketing.
I don’t think this requires professionals to think up ideas to improve Linux. Look at the Ubuntu Brainstorm, you will find numerous realistic, great ideas that are being ignored. The “regular” users are generally much more in touch than the “experts”. Cheaper too.
Edited 2008-09-13 04:14 UTC
That may be what he means.
Afterall, how many releases has it been since Ubuntu users have been promised a new default theme? I would assume that doing UX for a full OS, working upstream etc etc will be much more difficult, even moreso when Shuttle refuses to work upstream. His words:
(reply to first comment on http://www.markshuttleworth.com/archives/162)
ie if you want us to work with you, come to us.
/yes, I think Ubuntu is a bad thing when it comes to developers, and probabaly a bad thing longer term when it comes to users too.
There is no shortage of ideas. The challenge lies in implementing the ideas, not coming up with ideas.
I can’t believe that people are arguing here about the order of buttons on a dialog. The real usability problem of Linux is the CONSTANT regressions. Things that used to work fine suddenly break spectacularly. E.g. Evince on Hardy now apparently refuses to print half the pdf files thrown at it: https://bugs.launchpad.net/ubuntu/+source/cairo/+bug/227186 .
The explanation is of course simple. One needs (a lot of) real humans to do regression testing for GUI applications and applications that interact with the hardware. In the Linux development model the users do double duty as testers. One should simply be honest about this.
The real usability problem of Linux is the CONSTANT regressions.
Oh man, yeah. It’s not even just at the app or DE level. Stuff in the kernel breaks. For a few years now I’ve been using a true hardware RAID card for RAID 1 across OSs with no drivers needed ( http://www.cooldrives.com/inbaymosaiir.html ). This has worked in Linux, Windows, OSX: anything. And why shouldn’t it? To the host computer it looks like a single SATA drive.
Well starting with linux ~2.6.24 it doesn’t. The kernel guy responsible (seems to be one Tejun Heo) is aware but ignores it as the chip is “insane”. The way it provides a second partition from which one can read the status info is weird, and has only ever worked for me on Windows, but the blinkenlights are sufficient really. I can do without the info partition, what I’d like is for the RAID functionality to continue working as it has been.
It works fine with just one drive plugged into it, but that kinda defeats the purpose. Two drives: fail. The bootloader comes up, but the kernel barfs. If I can’t count on something that has worked cross OS for years (including in Linux) what can I count on? It’s slightly aggravating.
*EDIT*: btw don’t actually buy from that site I linked. I actually ordered http://www.cooldrives.com/insaiira1mic.html as I don’t have free external 5″ bays. They sent me the wrong product, which I now have mounted backwards internally, behind an intake fan Similar stuff is available elsewhere.
Edited 2008-09-14 13:35 UTC
The only solution to that is for people who have regressions to complain loudly on LKML. Not just to their distro’s bugzilla. Not just to the dev responsible. And to make sure that Andrew Morton notices it. That way, when they discuss regressions at the kernel summit, and devs pipe up to say they don’t think there’s really a problem (and oh, btw, will you merge my shiny new code?) Andrew will have plenty of ammunition and a big stick to beat them over the heads with. Andrew is listening. Some of the rest of the crew have a rather “speak no evil, hear no evil, see no evil”[1] attitude about the problem. Regressions should not be tolerated at this point.
[1] http://tinyurl.com/6l7aab
Edited 2008-09-14 15:28 UTC
Funny how, after all these years, GNOME fanboys and KDE fanboys still promote that silly rivalry.
Both the GNOME project and the KDE have long since stated that the “Linux Desktop Wars” are over. KDE is using GNOME code. GNOME is using KDE code. There are many examples of those two project working hand in hand, and yet the fanboys are still trying to make a flamewar out of every KDE/GNOME article?
Why dont’ you just stop? Each of the two desktops has its merits and its drawbacks. KDE is pushing technology forwards, breaking new ground and shaming both Apple and Microsoft, whereas GNOME has been setting new standards on consistency and usability.
The two projects are working together through freedesktop.org, the developers of both projects are having their conventions together. And still the fanboy users need to talk war? What a shame.
Just my 2 cents, but I see this and similar announcements from Ubuntu (such as an Ubuntu-organized multi-distro release schedule) as Shuttleworth’s belated recognition that Linux on the desktop is looking a little shaky. He’s well aware that if a major component of your OS is struggling – the x-server, for example – then everything else can get dragged down.
OTOH, it’s hard to see his new initiative making much difference. Partly because it’s a drop in the ocean and partly for the reasons Eugenia gives in her blog entry on the subject. Shuttleworth will always come up against the problem of people preferring to be big fish in a small pond rather than take the risk of becoming just another fish in a much larger pond.
Linux has been taking increasingly accurate shots at the desktop for about a decade now. That’s a long time in the IT business. If Linux was going to either unseat the incumbents or carve out a significant presence – say, 10-15 per cent – then it would have done so by now. But it hasn’t. Instead it seems to have settled at 0.5 to 2 per cent depending on whose figures you believe.
Nothing personal in this. We could be talking about washing powders in the detergent market instead of computing. But I suspect that this is how consumer demand, the supply chain (think OEM), competition and technological development pan out in this particular market. If you want to move out and take on the corporations at the desktop OS level, then the perennially underfunded open source development model is the wrong way to go.
Tough perhaps, but if after a decade of trying it hasn’t really worked in terms of capturing market share, what makes people think it will start to work in, say, five years’ time? Open source software at the application level – Open Office or Firefox, typically – has been vastly more successful.
So yes, as Eugenia points out, if Ubuntu continues to be a stand-alone OS, it will offer only more of the same. A very frustrating prospect for someone who’s fairly young, ambitious and mega-wealthy.