“Poor usability is a huge barrier to wider open source adoption. Our backends have matured and we consistently achieve technical excellence. Usability is the one area we have not yet mastered. For some reason, we treat it as a mystery instead of looking at it as a problem we can solve the same way we solve all other technical problems.” Read the editorial at NewsForge.
Editorial on Open source Software Usability
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I admit I am annoyed by some things like not being able to copy-paste between some applications. Or middle click and ctrl-v pasting different clip board content.
Eg. ig if I ctrl-c this text and copy data from a terminal, ctrl-v is the text I copy, midle click is the terminal data, and right click paste is also the text, however right click copy will overwrite the terminal data but ctrl-c won’t?
My main problems are (in order sort of)
1) Depedency hell and installer inconsistancies
2) copy-paste is badly broken
3) I don’t agree with using LFS for regular applications.
4) Poor sound card support
5) Default texts are difficult to read at high resolutions
The headline suggests that open source software (i.e. all open source software) has usability problems. But it’s KDE which has such problems (of dramatic size imho). And why are people always assuming non-technical users when it comes to usability? KDE sucks for me and many other professional users just as much as for the well-known Joe Average, and that’s because it’s bloated. On the BeOS or Mac OS X i can get my work done in half the time (in gnome in two thirds of the time because GTK is a bit slow) because i don’t have to search for menu items and icons so long because there are not so many of them.
While in the BOFH stories, OS/2 is always used as a means of threat, for me that would be KDE (3.0 or 3.1).
This guy seems to equate KDE with open source. I’ve noticed huge strides in open source usability in the past 2 years, mostly driven by Gnome rather than KDE. It’s filtered out into the community though and it’s not just Gnome – look at what Mozilla have done with Firefox to see their approach to usability. See how far apps like Gaim have come (still got a long way to go but still
.
I think a lot of progress has been made, even in KDE, though I still rather think they are behind when it comes to that.
Usability *is* technical excellence.
Open source projects pursue and achieve high levels of usability in interface design every day. It is simply that the users for whom the usibility is designed are developers, not end-users.
When designing an extension of a c++ gui toolset api, the vast majority of what the designer needs for excellent usability design is available through introspection. This gets the design 85% towards excellence, and wide use grinds off the burrs and sharp edges, with new proposals coming back to the project leaders as patch submissions.
What we need are tools that allow introspective, agressive end-user-interface designers to work together in a similar “conversation”, in a way that preserves, accumulates, and shares their design expertise in the same way that open-source project source-code preserves, accumulates, and shares code-design expertise.
Basically, my theory is that we aren’t getting anywhere because we have nowhere to submit UI-design “patches”, no tools for experimentally “trying them out”, and no clear shared “language” in which to code them. All we have are either conversational language at one end, which is not preserved or shared well; or source-code patches at the other, which require a fairly large amount of unrelated expertise just to propose a reorganized version of a file-browser dialog.
I’m not proposing reducing the diversity of gui languages, platforms, or toolsets. That would be counter-productive and impossible. I don’t even think that would be a good idea in a dreamworld where it was possible.
But, what I *am* trying to point out is that “we” seem to be having a hard time having a sustained and effective conversation about user-interface-design, and one of the reasons, in my opinion, is that we don’t have any effective shared medium or vocablulary with which to have that conversation.
I imagine the big gui projects like gnome or kde must have some way of having these conversations within themselves. Perhaps there are some sort of tools there that can be generalised to make “gui-interface-design patch submission” more possible for interested, introspective end-user/designers. Maybe something like the xml files generated by Glade, but simpler and more universal.
Then the natural evolutionary pressures of the open-source environment will eventually create the kind of excellence we are used to seeing everywhere else in the open-source world.
Accept for the LFS issue, everything you list is a usability issue.
For reasons to lame to enumerate, I’ve spent a lot of time recently running XP on the machine that usually runs Slackware. I’m back on Slackware now and I’m considering ditching it for XP: No dependency issues; no soundcard issues (how many people’s Soundblaster Live! card starts squealing at them as soon as they configure Alsa? Imagine the fingerpointing if MS let that out the door.) Copy and paste works consistently.
Most important to me, as I stare at the screen all day, is display quality. KDE and Gnome do a very good job, but Linux lacks the ability to configure fonts across all applications. Something is always just not good enough. Fonts in GTK apps look one way in KDE and another way in Gnome. Ditto for KDE apps running in Gnome. Maybe someone will fix this someday, but the grief given to RedHat/Fedora doesn’t fill me with confidence.
For the most part, the apps I rely on in Linux are available for XP. I don’t have a stake in the KDE vs Gnome contest, or in the Free vs Closed software debate. So, maybe I’ll do the fdisk thing.
(Security: Maybe I’m lucky, but the only time I’ve had a Windows machine attacked was when I forget to pull it off the net during the install. I use commercial firewalls and antivirus software on windows. Yes, they’re an occasional pain, but they’ve worked for me. Never using Outlook helps, too.)
but Linux lacks the ability to configure fonts across all applications. Something is always just not good enough. Fonts in GTK apps look one way in KDE and another way in Gnome. Ditto for KDE apps running in Gnome.
This is my #1 beef with Linux as a desktop OS. The #2 beef is the software installation. People can say that having each individual distro making their own packages is not a problem, but I’m telling you that it is – it has been a problem for every distro I’ve tried.
I’ve had really good experience with fonts in Linux – all the apps I use draw on the same list of fonts, etc. Just drop a font in the ~/.fonts folder and I’m done messing with them. And they look great in all apps, too.
Packaging is a problem right now. I’m counting on autopackage.org to remedy that problem very soon.
…before Open Source is smooth enough for the average user. Linux is still young..
Fonts in Linux are good, but they aren’t consistently good. For example, I’m in Firefox in Gnome right now. If I switch to KDE and launch Firefox, the fonts display differently (IMO, worse). The same thing happens if I run Konqueror in Gnome.
This means that someone who wants to use KDE and Firefox must accept second-rate display quality. That’s a real problem. As far as I’m concerned, neither KDE or Gnome deliver anything that make me willing to put up with this annoyance.
On autopackage: I hope it does some good, but I’m skeptical. Dependency and packaging problems rest on the underlying library structure used in Linux and the development practices of most Linux developers.
I admit I am annoyed by some things like not being able to copy-paste between some applications. Or middle click and ctrl-v pasting different clip board content.
The Middle Click pastes what you just highlighted, CTRL-V pastes what you just copied. They aren’t supposed to be the same thing. They are two different things, and not difficult to understand.
KDE and Gnome do a very good job, but Linux lacks the ability to configure fonts across all applications. Something is always just not good enough. Fonts in GTK apps look one way in KDE and another way in Gnome. Ditto for KDE apps running in Gnome.
Then just use KDE apps or just use GNOME apps, and you’ll have your consistency. Under Windows you also use just one single desktop. If there were a second desktop available for windows, there would be these inconsistencies too, but thats the price of diversity. A way to bypass these would for example be to just not use this secnd desktop, or any apps developed for it. You really cant blame linux for its diversity, because this is the actal essence of free software and free evolution alltogether.
Just stick with something that was planned to be consistent all the way through. You cant blame Free Software for its diversity, or you cant blame two different developers on different sides of the world, who work on different projects for not cooperating to bring you the best and most integrated experience you can have. For that you just pay them to little – to be precise, you pay them nothing.
From their point of view, their software might be just perfectly integraded – with the environment its written for. Requiring for every project to integrate perfectly with every other Free Software project any other random developer worked on and developed on the other side of the world is a little bit too much to require.
The free software world alltogether evolves rather than being intelligently “created” by a single instance, and when you use it, you should think of that, and when you have a problem with diversity, ditch it and pay for your integration. You simply can not have freedom and free development and commercially motivated and tight integration at the same time, these two perspectives simply contradict too much.
View profiles are fine. But at least make it possible to have the home button to point to different web homes vs. file homes.
The file home should be ~/ while the web home should be an html site.
Right now the home icon on the toolbar can only point to one place.
A whole article about usability in the OSS desktop and not one mention of GNOME? I mean, common(!), KDE are the people who sacrifice usability/ease of use for features, while it is the GNOME people who sacrifice features for usability (also a11y, and ease of use).
And while MANY thing are still problomatic, they are usually known are being dealt with, just sometimes they take time.
Also GNOME has had alot of money put in by SUN to help do UI review and so forth, and there are a number of UI knowledgable people doing some serious fixes in this issue (The new HIG should be out soon, UI-review people making the desktop better one app at a time, a11y being an amazingly important issue in gnome, etc…)
My advice to the writer? Do some reaserch beyond KDE or even switch desktops for a while…
I do agree with him the the KDE people have what to learn about usability from the GNOME people (which is why I switched to Gnome with version 2.0)
The Middle Click pastes what you just highlighted, CTRL-V pastes what you just copied. They aren’t supposed to be the same thing. They are two different things, and not difficult to understand.
Its neither difficult to understand, nor has X changed this behavoiur in the last several years.
The problems come from the inconsistency with the behavoiur has with the way windows handles these tasks, and which many people simply require to have on any other “modern” system.
You cant do little when people will not take that a system other than windows is actually – different.
You wrote:
Packaging is a problem right now. I’m counting on autopackage.org to remedy that problem very soon.
And the autopachege homepage states:
“One of the most obvious ways in which they differ is file paths and library versions, but distros can differ in other ways too. There are two solutions to this: either ALL distros must conform to some standards such as the LSB, or a package manager must be built that is powerful enough to deal with the myriad differences.”
I thought that there is some better idea how dependancy handling could be achieved throughout different distributions, but as you can see, autopackage cannot solve that. It still requires the distributions to give up the way they handle things, and competely change their system to be compatible to the way another distribution handles things. So, for me, autopackage dioesnt suggest more than “for all of you to get compatible, just simply and all you have to do is to agree on autopackage as your new standard. Well, as I see it, this ist really much of a “sollution” because nobody will be doing this. As you see, even some distributions that use standard packaging system “rpm” Suse, Redhat, Mandrake, to name the biggies, arent really comaptible to eachother.
LOL, but.. autopackage’s sollution is just a wannabe standard nobody will ever care to standardize on.
The point of the autopackage homepage saying “LSB, or a package manager must be built that is powerful enough to deal with the myriad differences” is to say that autopackage is such a powerful package manager. The binaries in autopackages are relocatable (meaning they can deal with the different configurations on different distros). So, yes, they aren’t perfect, but they will solve a huge problem.
Lets try not to be so negative about the possible solutions to a big problem in FOSS – at least look at the first.
“Then just use KDE apps or just use GNOME apps, and you’ll have your consistency.”
Which massively limits the number of apps you can use. For example if you use KDE and it’s apps it means no OpenOffice, GIMP or Mozilla.
“If there were a second desktop available for windows, there would be these inconsistencies too, but thats the price of diversity.”
For the slight usability differences between KDE and GNOME that’s a price not worth paying IMO.
“You simply can not have freedom and free development and commercially motivated and tight integration at the same time, these two perspectives simply contradict too much.”
So you don’t think Linux will ever compete with Windows for consistent UI design? That’s a shame, but you’re probably correct.
You can use K-Meleon to have a Mozilla-based browser in KDE. Hope that helps a bit, though I understand the problem you’re having isn’t going to be solved with that one app.
consistency is badly needed between toolkits and desktop environments when it comes to behaviour and layout of menus, dialogs, behaviour of cut’n’paste, drag’n’drop, use of language, fonts, colors and icon themes.
Hopefully freedesktop can solve most of these problems but i’m afraid some of these inconsistensies will remain.
>>…just use KDE apps or just use GNOME apps, and you’ll have your consistency. Under Windows you also use just one single desktop. If there were a second desktop available for windows, there would be these inconsistencies too, but thats the price of diversity.
I have to disagree. As a user, why should I be willing to limit myself to software that works correctly with only one particular environment? That’s especaily true because neither delivers adequate tools to meet all my needs. From my perspective, the display inconsistencies apparent in Gnome and KDE detract from the usability of both. There’s no difference between Microsoft, or Apple, attempting to lock in user and developers and the approach taken by both KDE and Gnome. What’s prevailing is a sense of competition; what’s lacking is a sense of cooperation in the best interests of users.
I alos disagree that it is impossible for developers to interact to the degree needed to resolve this issue. It’s done in other areas, especially the kernel. The fact that it isn’t happening with this and other usability issues reflects the indifference many developers feel about the people they are coding for.
Your argument stands open source’s lauded diversity on its head: You’re asking me to limit my choices.
Why don’t we standardize the packages instead, like the way they’re laid out and such?
For example, if a package manager needed to know what kind of dependencies a package had, the package would include in it a file called dependencies.xml (or dep.xml for sure) it needed in a standard layout as defined in an RFC or something. That way, you could use any package manager you wanted and all of them would work any package that supported the standard. Of course, we’d need two standards – one for binary and one for source packages, but I think it’s doable.
An autopackage (a .package file) contains all the files needed for the package in a distribution neutral format with special control files inside, wrapped in a tarball with a stub script appended to the beginning. In order to install a .package file, you run it, and the scripts then check your system for the autopackage tools and offers to download them if they’re not present.
See, this is where I think autopackage is flawed. Why does (or should) a package need a script? For all intents and purposes, a package should be simply a collection of files, and not have to worry about where it’s going or how it should be installed.
For example, let’s say you’re trying to install a software DVD player on your Linux box. So, you download the .package file, and double click it to bring up the distro’s package manager (or run it from the command-line, whatever). Now, instead of the script in the package having to check and see what’s on the system, the package manager just looks at the dep.xml file and sees that it needs (for example) libdvdread in order to work. So, the package manager checks its database to see if the library is installed. This way, the developer can package up his app or library one way and have it work in all distros, and each distro could have its own package manager that works completely different than the others. Hell, you could even have package managers that are compatable with multiple distros if you want. The only thing people would have to agree on is how the package itself should be laid out, and then design each package manager around that standard. AFAIK, autopackage basically requires that you do the same thing (build the package in a certain way), but the way I’m talking about, you don’t have to worry about autopackage tools being on the system – just use whatever the hell you want.
If / had a layout that everyone could understand, then there wouldn’t be that many problems. When I run a make install, that file is just magically working. I have no clue where it’s gone to…
As a dozen people what / should look like, and you’ll get two-dozen answers. Here’s one of mine:
/
/cdrom <— or whatever
/floppy <— if at all
/gnome <— only gnome core
/home
—–/username
————–/desktop <— All user’s stuff here
————–/settings
————–/documents
/kde <— only kde core
/programs
/system <— common files, like CUPS
Autopackage can interact/interoperate with distro-specific packaging systems just fine. And you don’t have to worry about have the autopackage tools at all because if you don’t have them they are automatically downloaded and run for you. In fact, you don’t even need the root password, you can run the package, install the autopackage tools on the fly in your home directory, and install the app at issue in your home directory all without needing a terminal or worrying about what distribution you’re using. Try it out (the link on the autopackage homepage that says “Try It” will tell you how). Yeah, it’s more complicated than Mac OSX style installation, but the user won’t ever know that. They just double-click – or, according to an article by the autopackage creator, perhaps one day drag an icon representing the app to their desktop – and they’re done.
Enough said.
“That way, you could use any package manager you wanted and all of them would work any package that supported the standard.”
Yes, but you wouldn’t for certain be able to use a binary linked against libraries from distribution (or OS) X on distribution (or OS) Y. Given most compiled packages are compiled shared what does it solve? Why should i care wether RPM works on Debian? Most of the time i don’t care. If i do, i just convert the RPM to a DEB using Alien. What remains is source packages. Those aren’t a big deal either, because most of the time there’s already a source package available from the distribution. With that one can make a package on their own. Or use Alien to unpack/convert the source package.
You also forget that the package hierarchies (e.g. names) are nowhere but the same in every distribution. Versions aren’t always the same either.
“So, the package manager checks its database to see if the library is installed.”
What if:
1) the library has a different name in the ‘other’ package manager? Or the library doesn’t exist, a total different library provides the similar functionality (more likely on non-“Linux” than on “Linux”)?
2) the versions aren’t the same? You only want to check if the library exists. If versions aren’t the same, possibility is highly likely leads to segfaults, unresolved symbols and other inconsistencies.
Because of point 2 you are arguing in the end for every application version should be the same on every “Linux” distribution. Because of point 1 you are arguing in the end every library in a package must be named the same in every OS, and all OSes must provide exactly that library using that name — no less, no more.
“This way, the developer can package up his app or library one way and have it work in all distro”
No, see above.
So what exactly are the autopackage tools and why should they be needed? Does autopackage have to be developed in such a way that it ‘knows’ about each specific distro (Or in other words, does it have to know that if the distro is Slackware, the files go here, Fedora … the files go here)? If so, that is an inherent flaw. They should leave that all up to each individual package manager.
BTW: I’m not disagreeing that it works – just suggesting that there might be a better way
Linux was supposed to take over the desktop since 1998. I honestly think it never will. There will always be Windows, OS X, and for servers, UNIX clones like BSD and Linux. The fact is that usability is a concern only commercial vendors can really take seriously, because they have the resources to do interface research and devote entire departments to creating a standard interface. What is there on Linux? I have to actually install two entire desktop environments just to run each environment’s apps within each other. I have five or six competing APIs. An untold amount of window managers. It’s never going to gain mainstream acceptance, not with the way things are now.
“When I run a make install, that file is just magically working. I have no clue where it’s gone to..”
You specified that when you started building with e.g. configure. You could even change the paths from the Makefile, but most people who want their system to Just Work don’t (have) to do this. Heck, if i want to build a package from source in Debian i don’t even need to know this: i just use the Debian utilities which automate parts like this. All the flags were already specified in $PACKAGE_DIR/debian/rules. Note that i used “want”, not “need”. Need is almost never necessary, its very rare.
“As a dozen people what / should look like, and you’ll get two-dozen answers. Here’s one of mine”
(Haven’t you broke the POSIX and LSB compliance with this?)
You don’t really type cd /programs, do you? Ever thought about why normally everything in / only has 3/4 characters? Why the people who thought out this documented standard decided these names? (*) They’re well documented btw, so anyone would be able to learn why they’re named like they are and what they are meant for.
Assuming you’re using this set-up, i’m really wondering where you have X installed. /x? or /X? What if i have 5 WM/DE’s installed? /xfce4, /gnome, /kde, /afterstep and /waimea? How about 10?
On a sidenote i think i might even like to have my X programs not in /usr/bin together with my CLI problems there. But what if there’s one program which has both modes: blaat -g for GUI and without -g CLI (Links2 for example).
(*) I once had Cygwin and OpenSSH installed on Windows XP. So i SSHed into that computer and came in my “home directory”. I went to C: to do some stuff (something was very damaged) and after that i had to go to my “home directory”. So i had to type “cd Documents and Settings” and there was no completion available. Backspace didn’t work, so i had to type it flawless. It took me 2 times to get it straight, and the many times i’ve done this i had similar problems. Typing “cd home” is so much more convenient! Not to mention names like similar long names like “My documents”…
Yes, but you wouldn’t for certain be able to use a binary linked against libraries from distribution (or OS) X on distribution (or OS) Y. Given most compiled packages are compiled shared what does it solve?
Sorry, I’m not that technical … have no idea what you’re talking about here
However, I think I can solve your name/version conflict.
Regardless of whatever a package is called on a particular distro, it usually has a ‘generic’ name that the developer calls it, so we’ll call this the package’s real name. Inside the package itself, we could have a file called info.xml that tells the package manager what the name of the package is, what type of package it is, what the version number is, etc.
Now, let’s say your distro includes (for playing DVDs) the libdvdread library (version 0.9.3 for our example), but it actually calls the package read-dvd instead. So we have the real package name (libdvdread) and the distro’s modified package name (read-dvd). So when you first install the distro and this library is installed, the package manager adds the following info to its database:
REALNAME: libdvdread
MODIFIEDNAME: read-dvd
VERSION 0.9.3
So now it knows what the ‘real’ name of the package is, and can also keep track of its modified name, version number, and what else you want to add. Esentially, it’s taking all of the information from info.xml and implementing it in whatever way it wants.
Now, let’s say you download the Ogle DVD player because you don’t like the player that’s installed on your distro for whatever reason. So the package manager checks the dep.xml file for Ogle and sees the following (and please note I’m pulling all this out of my ass – I don’t even know xml, but am guessing it would look like this):
<dependency>
<name=libdvdread>
<type=library>
<version=0.9.4>
</dependency>
Here we see that this app depends on libdvdread and that it requires version 0.9.4 to work. When the package manager looks at the dependency info and sees that the package name is libdvdread, all it has to do is search its database under the REALNAME field as the search field until it finds libdvdread, and now it has all the necessary information. Since you’ve only got 0.9.3 installed, it’ll then have the option (if this particular package manager has auto-dependency resolving capabilities .. which would be completely optional), it can then download 0.9.4 for you, and it then can run both versions of libdvdread side-by-side on the same system and call it whatever it wants. If these library files need to be put in a certain directory for this to work right, the package manager should know in what directories to put things in each distro.
If it does download 0.9.4 for you, it’ll add another entry to its database so that it knows you have two different version installed:
REALNAME: libdvdread
MODIFIEDNAME: read-dvd
VERSION 0.9.4
But the package manager doesn’t HAVE to have its database set up like this .. that is irrevalent as far as the package itself is concerned, because the package has all the information that the package manager needs to do its thing in whatever way it wants.
It is up to the package developer to come up with the ‘real’ name of the package, and if the distro maker wants to modify it, it’s up to the package manager to keep track of all that.
The one weakness I see here is that how do you keep two different packages from having the same name or prevent someone naming a package the same as another one on purpose to mess up your system? Well, the way I see it, this is a problem with the current packaging system(s), so not like we’re introducing anything new here
Certainly, I would imagine that each package could have it’s own unique identifier or something ??
The Gremlin: I agree that the layout should change, but becasue of the open nature of Linux some changes will simply never happen. There are simply too many speratly managed involved parties for someone to stand up and say “hey. lets all start doing this that way now”. I takes only 4 words for my to say “your idea is stupid”, problem solved (j/k)
Actually, I don’t want 100 directories off of / any more than I want 2000 applications in /usr/bin with the rest of the application files in various other places on the dsik.
/mnt/cdrom and /mnt/floppy are fine becaue you also have to consider seperate partitions (windows) drives (cdropm2) and removable media etc. /mnt is cleaner
I don’t want /kde /gnome /blackbox /fluxbox /icewm and /xfce for all the same reasons.
I like the way Apple has an /Applications directory, there is no reason Linux could not also do this.
System files, services, X, kde-libs, command line tools etc. can continue to follow HFS, but GUI applications like Firefox, OOo, Gimp, K3B, Totem, Xmms, Games etc. should be in the applications directory.
/apps/xmms
This way configs do not clutter /etc, xmms-lib does not clutter the rest of the stuff in /lib etc.
Also, say you have a compatibility problem where xmms depends on a newer library version of random-lib, which if upgraded would break some other older applications.
In that case, if that library is not too large, you could just throw the new version in /apps/xmms. hardly efficient in disk space and RAM usage, but at least you have the option to have a working application.
BTW, the above suggestion does not break POSIX or LSB compliance.
Does anyone really have issues with copy/paste in linux? or is this one of those imaginary flaws certern people claim linux to have (not that linux is in any way perfect).
I’ve tried different distros in the little over a year I’ve been using linux, the last 9-10 month as my main desktop, and I have had only one problem easylly solved, I could paste with ctrl+v to a java-app, but ctrl+insert worked.
Just trying to understand what it is people are bashing.
GTK+ and QT use seperate clip boards, so you cannot copy and past form Gnome apps to KDE apps etc.
Also when you highlight it will auto-copy the info to one of the clip boards, so if you want to middle-click to paste a URL into Firefox, when you go to highlight the old URL to delete it, it is then copied over the URL you are trying to paste.
This feature used to be _really_ annoying when klipper defaulted to popping up an action menu when you had a htt://url highlighted.
Every time you would go to delete a URL you’d see, “you coppied a URL would you like to:”
Upen it in Konqueror
??
??
Do nothing.
Actions are not off by default Klipper, but removing it is usually still the fist thing I do on a new user account.
Actions are now* off by default in Klipper.
Actually _I_ can easylly copy between GTK and QT-apps (I use KDE), don’t know about others, that is the reason I asked since I havn’t seen it my self.
And I have never seen the “you coppied a URL would you like to:” when copying URLs or anything else, but I don’t use the actions in klipper and never did.
“Also when you highlight it will auto-copy the info to one of the clip boards, so if you want to middle-click to paste a URL into Firefox, when you go to highlight the old URL to delete it, it is then copied over the URL you are trying to paste. ” Huh, I don’t think I understand?
If I highlight the URL in the URL-field in either Konqueror or Firefox and middle-click then yes the old URL is pasted into it self. But thats because middle-click as I see it, is used to paste the last highlighted text into the spot thats middle-clicked, so I don’t see how this is an issue, since the middle-click does exactly what it is designed to. And I don’t see the problem between GTK and QT in this “issue” since I just tried middle-click in both konqueror and firefox and it worked as expected.
I’ve used multible distros, Redhat 8/9, Fedora core 1, Mandrake 9.2, SuSE 9.1 and Gentoo. All with KDE as DE, and of these issues, I know not (still why I asked).
“GTK+ and QT use seperate clip boards, so you cannot copy and past form Gnome apps to KDE apps etc.”
Huh???? So why am I successfully copying stuff from gedit to kwrite? It must be an illusion.
I suggest you to upgrade to the latest versions of GNOME and KDE. The clipboard problem has been fixed a long time ago.
Selecting a piece of text and pressing Ctrl+C are two different things. Forget the old broken behavior and read: http://www.freedesktop.org/Standards/clipboards-spec
It’s mostly people like you, who has gotten used to the old broken behavior, think that the clipboard is still broken, either because you haven’t upgraded your software, or because you’re so confused by the old broken behavior that you can’t understand the new proper behavior.
As has been allready mentioned copying between gtk and qt apps is no issue at all and you are free to use Ctrl+C. So either you and others who constantly bring up this nonissue don’t have the slightest clue what they are talking about, or you are simply trolling.
I would also like to thank all the gnome trolls that informed us that gnome has no usability issues whatsoever. Pleas inform the gnome devs, so that they don’t have to worry about improving gnome anymore.
All in all, a very interesting article that made some good points. It surely didn’t deserve a comments section full of trolls.
I am running mandrake 10 community.
kwrite 4.2
gedit 2.4.1
From KDE 3.2
If I run kwrite and gedit I cannot ctrl-c from gedit and ctrl-v to kwrite, in that order. If I copy something from kwrite I can paste it to gedit, and after that also paste back to kwrite from gedit, but not before then. If I close out both I can then reproduce the problem.
Also, highlight in gedit does copy to middle click but Firefox does.
There are also several other inconsistancies that are harder to reproduce without sitting down to mess with it. I am glad to see that you took one glance at this working on your distro and claimed I must be making stuff up. Typical really.
I like this part:
“It’s mostly people like you, who has gotten used to the old broken behavior, think that the clipboard is still broken”
Good stuff.
Works without a problem here.
Using the latest releases of XFCE4, firefox, thunderbird, nedit etc I find myself regularly having to use my clippy tool. Yes there is a problem.
The Gremlin: Funnily enough you’ve pretty much just described the filesystem layout in Syllable.
What’s most interesting about all of this is how defensive everyone has become, and the almost instant unhelpful finger pointing. We’ve learnt that it isn’t a problem with Gnome, it’s all KDE’s fault! X’s clipboard model isn’t broken, that’s the users fault! Maybe this goes some way to explain why Open Source usability in general is still so poor..
If people claim things that are simply false pointing out that these claims are false is hardly defensive. And sorry, I really tried it out now and I simply don’t have no problem with C&P whatsoever.
“As has been allready mentioned copying between gtk and qt apps is no issue at all and you are free to use Ctrl+C. So either you and others who constantly bring up this nonissue don’t have the slightest clue what they are talking about, or you are simply trolling.”
Copying plain text generally works fine, but copying images or other data between QT and GTK apps is still a big problem.
I find complaining about the number of window managers in linux to be somewhat dishonest. I’d also like to know how you expect this problem to be solved. If you want a single desktop distro, those exist and work quite well. But its not like theres anybody who can stop all these projects from existing.
Its also dishonest, because you have to ignore the fact that there are multiple window managers available for every desktop system. Blackbox for instance is almost universally available.
Complaining that Gnome apps, and KDE apps don’t quite work correctly together is rather like complaining that iTunes, and Windows Media player don’t look alike or work alike.
all of you are a bunch fo nitpicks. its like reading a board for a fps mod, people allways complain that they dont have the features from mod a,b,c…x,y,z and that the lateasts changes messes up the mod to the point of unplayability.
there are many applications on any os that dont care shit about usability studys and so on and just stuff the options where they feel that they makes most sense for the people makeing it.
copy and paste? its being worked on and the only software i know of that dont do it the windows way are xterm and similar software that are from the GNU only days of development. gnome and kde have agreed on a copy & paste framework that when fully supported by all apps in those groups should allow you to copy any item from one group to the other. what software that is made by other tools well i cant say about that.
personaly i dont care about one progam not looking quite like the rest, thats what i feel is a part of the charm of a linux desktop. oh and i do not run windows with its default color setup (with all the good and bad sides that come from that, like hardcoded black text in some application guis and so on).
application installation? if your distro dont come with a version then i wonder what distro your useing if its a common program. and if its a uncomon distro with a uncommon app that you want to install then i guess its the good old ./configure & make & make install that does the trick. if your new to linux then dont pick a distro like gentoo or slack unless your planing to lean and get your hands dirty.
to me it looks like windows and mac have made stupidity the norm, your not supposed to think as we the software supplyers have done the thinking for you. this is allso true for politics and other stuff to, we go with the flow, the adds, the campagian slogans. its like some roman emperor ones said: “give the people bread and circus and they dont care what you do” (or something like that)…
linux is the thinking mans os, its unforgiving if you dont think about what your doing, just like a car. forget for one moment what your up to and your in the field if your lucky or going head to head with a truck if your unlucky. im tired of awnsering the same question from the same person 5 times in a day. sure, useability can help this but so can a willingnes to learn from the side of the user…
“The Middle Click pastes what you just highlighted, CTRL-V pastes what you just copied. They aren’t supposed to be the same thing. They are two different things, and not difficult to understand.”
Why is it using Control-C and Control-V at all? There is a clash here because Control-C means “end task”.
It should use the “platform” key (the Windows key on most keyboards, the Command key on Mac keyboards, the Amiga key on Amiga keyboards) with ZXCVQS etc. Using the CTRL key seems to be a typical case of Linux blindly copying a mistake made by Microsoft.
If there is one thing wrong with the Linux GUIs I have seen, it is that they are far too much like Windows.
You can use K-Meleon to have a Mozilla-based browser in KDE.
K-Meleon is software for windows.
The Linux desktop does have barriers to usability. This is known. However, there is so much work going on right now to overcome these issues. Also, those of us who have been using Linux as a desktop (mostly in a dual-boot setup until I made the full leap earlier this year) since the late 90’s are astounded with how far this work has come.
You really do need to go back and try to use Redhat 5.0 again to see how much improvement there has been. Think Netscape 4.x with the grosses fonts ever…. think fvwm95 with the ugliest windows rip-off interface ever.
The future is in standards that KDE and GNOME both agree to implement. Both are going down this road too, from what I hear. We have freedesktop.org to thank for laying some of this groundwork.
freedesktop.org is hosting a ton of work to bring KDE and GNOME closer together:
http://freedesktop.org/Standards/Home
They work on ways to beautify and enhance the graphical appeal of X:
http://freedesktop.org/Cairo/Home
Project Utopia will bring ease of use with hardware peripherals to the linux desktop:
http://freedesktop.org/Software/hal
http://freedesktop.org/Software/dbus
Both KDE and GNOME do need to enhance their usability, though I do agree that the simpler look of GNOME and the increasing role of the GNOME HIG in its development both show that GNOME is farther along in the usability area at this time.
KDE needs to reduce the clutter, pare down and simplify their desktop in some ways. Choose one text editor for example.
Both desktops need to continue to interoperate better, as neither is going away anytime soon and each offers unique advantages that make both really fun to play around with and use.
Here is an idea. I know there already too many distros out there to even to begin to justify another distro however this may be exactly what is needed to solve usability issues.
The idea is that this distro could be built around a community which votes on usability issues and creates things such as the Gnome HIG to guide the distro. By creating some kind of complanince standard for applications on this distro it could ensure a consistent user experience at least for these apps. With such compliance standard it may be able to influence application designers especially if the usability constraints are sensable and clearly defined.
Another step is that small glue applications could be built to enforce usability and interoperablility between different applications or parts of the OS in general. An example is the building of a package manager which relys on the suggested XML info, and dependency formats. Utilities could also be made to give a virtual view of the file system allowing it to appear as though /applications or /settings really exist without breaking the internal workings of applications and how things are installed on Linux.
I would be overjoyed to join an effort in this direction if some more people with more Distro development experience would be willing to help out.
The best comment I read so far.
I can agree with nearly everything you say, but for one point.
The neverending story of the kde text editors. 🙁
Kwrite and kate are just different frontends for one editor, one simple interface for basic text editing, one complex interface for advanced tasks.
The only reason why kedit is around is that kate does not yet support bidirectional writing. AFAIK the devs are working on this feature and the second it is implemented kedit will be gone and there will be only one editor with two frontends.
Having used kde and standalone windowmanagers in the past, i was impressed when i tried out gnome today for the first time in years. it feels very polished and easy to use.
I think i will stick with gnome for a while to see if the feeling stays with me. it definately seems that gnome is headed in the right direction! makes me happy to see.
If developers aren’t writing software for other people to use, then they’re pointless.
There is a well-trumpeted myth that free software is only about developers writing for themselves, and that users should be happy to take whatever they can get.
That’s an incredibly arrogant position. What other kind of engineering discipline only builds what an individual engineer wants to build? How many physicians only treat patients with problems that interest them? How many teachers only teach students who intrigue them?
If developers really believe that the code they write is to be use only by themselves, then they should not release it.
There is a well-trumpeted myth that free software is only about developers writing for themselves, and that users should be happy to take whatever they can get.
That’s an incredibly arrogant position. What other kind of engineering discipline only builds what an individual engineer wants to build? How many physicians only treat patients with problems that interest them? How many teachers only teach students who intrigue them?
Your contending that you know what is best for you. Let’s use your examples as proof of why this is not always the best case:
Engineers may not be able to go around building exactly what they want, but if you asked them for something that was poorly designed, they would tell you. You tell them what you would like, and they build it, but that doesn’t mean everything you want can be created.
Physicians are a perfect example of what is wrong with your argument. Just because a patient wants drugs doesn’t mean they should be given drugs. A patient doesn’t know the best treatment. Why should they dicate what the physician does.
Finally, teachers know better than the students what the students need to learn. By your assertion, if the students wanted to learn one subject, the students know best, and the teacher shouldn’t teach anything else but only what the students want to learn.
Obviously, this is not what you wanted to say, but it is what you are implying; it is what you are saying. Developer’s do write software for themselves and other people. However, just because you think what they are doing is wrong doesn’t mean it’s wrong. You being a user of their software doesn’t give you “phenominal cosmic powers” in being able to tell them what is right and what is wrong.
Simply put, most of the comments from users in the open source software space simply don’t matter. People feel as if their voice should be correct, and should be heard, and she be obeyed. Fine, the developers hear you, but that doesn’t mean your opinion really matters. And it shouldn’t. Everyone has an opinion, but not everyone’s opinion matters.
Their is another case that the majority is not always right. The majority can, and has been, wrong in many, many cases. If software developers only listened to the majority, then the Mozilla team would have shut down and Linux wouldn’t be around at all.
You don’t get it. My analogies weren’t about physicians or teachers rejecting requests when they know that is in the best interests of the patient or student.
My analogy was about physicians or teachers avoiding something that is in the best interests of the patient or student simply because they aren’t interested. That’s the appropriate parallel with developers who contend that open source is not about users and, hence, take a “users be damned” stance.
The software universe is full of developers who will argue, for example, that emacs or vi, and fvwm, are the ultimate in interface design. If people don’t like them, they will contend. it is only because those people lack the will or the ability to learn how to use them.
Blaming problems with bad software on users makes no more sense than blaming problems with bad autos or bad refrigerators on their users. Arrogance is behind it.