During the majority of my time working with computers, Windows was the operating system of choice. Reason being, it’s all I’ve known. In 2002, I took a college course titled “Linux Administration” which entitled me to a few cd-roms of Redhat 7.x. While this course was nothing more than a few extra credits for me, I fell in love with Linux and went through the entire textbook a week into the class. It was a nice feeling to use something “different” than what I was used to.Disclaimer: Some of this article may seem as if I am dissing Linux, quite the opposite, I love it, but I feel that it could be much better.
Since then, Linux has always resided on my 20GB hard drive. (My 80GB drive used for Windows). The smaller of the two hard drives housed whatever Linux Distribution I was using at that particular week, depending on what mood I was in. As time went by, I grew tired of Microsoft breathing down my neck. Until recently, I didn’t have internet access (checked email from a college computer) so upon reinstalling Windows XP a number of times, my software indicated I needed to call Microsoft to get a new product key. I had to walk to a pay phone and dial the 1-800 number and explain to the representative on the phone that I only had it installed on one machine. At that point, I had it, I needed “freedom” to use my computer as I see fit, and began trying to find a Linux distro I could migrate to.
Most recently, Linux is now housed on my 80GB drive, leaving 20GB for Windows 2000 (I had it with the activation of Windows XP) and the partition for Windows 2000 is now used primarily for gaming. At this time, I am switching back and forth between Fedora 2.0 and Suse 9.1, having a hard time deciding which one to stick with, although I am leaning more toward Suse.
Using Linux primarily now, I see more and more that I think could use improvement. First and foremost, usability. Mind you, I am a geek, so I am able to use Linux regardless, no matter if I have to compile from source or search for an RPM on Google. My point is, I shouldn’t have to.
Installing packages is my biggest gripe about Linux. Sure, I could use Apt-get to install everything, but what happens if I forget to pay my internet bill? I can’t install anything, then. One of the things that makes Windows so popular is that if you download a setup program and activate it, it installs your new software without a hitch, providing the developer made it compatible with your system.
In my opinion, the way RPM’s are handled is excuseless. Why should I search for a dependency, then search for a dependency for that dependency? It’s annoying. I think there should be a “new” kind of RPM (Maybe call it “Enhanced RPM”, or ERPM or along those lines) that contains all the dependencies you may need for that program. If the dependencies in the package are necessary, the system installs them. I may be behind the times, but I don’t think this exists yet, and it’s unfortunate.
The more I use newer revisions of the Linux operating system, the more impressed I get. For example, KDE 3.2, abosolutely amazing. KDE 3.2 contains all of which made the previous versions such a hit (as far as I know), and improved upon everything. To me, that’s what all operating systems should be, instead of “Get a new Windows version, search for all of your software updates to install on the new version”.
While KDE improved on the GUI, it doesn’t seem like anything was done on the “ease of use” feature, to install a program, it still takes 20-30 minutes of research on Google, as of the time of this writing APT isn’t working too well right now.
In the same manner, when you install Windows, you install the core operating system with nothing extra. For the most part, you insert the cd-roms of your applications and install them, if necessary. The core Windows OS is on one cd-rom. I installed Suse 9.1 last week, and not only did it have 5 cd-roms, I also had to choose from a list of hundreds of programs, to make sure I would have all dependencies resolved. I would like much better a one cd Linux distribution, that contained nothing but the core OS and a Window Manager, with support for “ERPMS” that would work on the fly.
Upgrading software in Linux is another thing that could use a facelift, back when I was using Fedora Core 1, installing KDE 3.2 was almost a rite of passage, but I did get it installed, but not without totally destroying my Gnome configuration. God help me if I wanted to upgrade that as well! In a similar subject, I still cannot get my joystick to work manually, however, the newest Fedora and Suse versions do this for me right out of the box, which is a huge step toward usability, and is definitely a step in the right direction.
Now, this article, as I said before, is not to diss Linux. Sure, it’s hard to use, but as I said, I’m a geek, I like a challenge. Even though it’s annoying to resolve dependencies all day, the feeling I get out of getting something to work more than makes up for it. However, if I was not a geek, I would have thrown my computer out the window by now. Aside from that, I will continue to get to the point where I can migrate to Linux completely.
With Windows Longhorn fast approaching, the Linux community has 1-2 years to get everything in order. If no improvements are made, casual PC users will buy/download Longhorn as soon as it’s released and ask questions about stability and how proprietary it is later. (That’s what Desktop Linux needs to focus on, the “casual User”. In order to thwart Microsoft, we need a Linux version that does everything it does and better. Microsoft’s ease of use plus Linux, equals the best Operating System the world has ever seen, but unfortunately, I don’t think it’s in development yet.
There’s a misconception out there about exactly what Joe, Mum and Grandma are capable of. The short of it is that a lot of them are capable of very, very little! Joe Grandmas have trouble using cut and paste, and are hardly able to install any Windows software let alone install the OS. Essentially there’s a whole world of people out there who struggle with computers every day, and they rely heavily on friends and family to help them out whenever something goes wrong or needs installing. It’s unreasonable to expect Linux distributions to be so usuable that Joe Grandma can fully administer it, either today or in ten years’ time.
To reduce stress levels in the Linux computing environment please use one of the following:
Debian – General Users (I just want it to work)
Gentoo – Power Users (I want it my way)
Slackware – Control Freaks (My way is the correct way)
There’s a misconception out there about exactly what Joe, Mum and Grandma are capable of. The short of it is that a lot of them are capable of very, very little! Joe Grandmas have trouble using cut and paste, and are hardly able to install any Windows software let alone install the OS. Essentially there’s a whole world of people out there who struggle with computers every day,
The misconception is that they should be able to do these things without a shred of knowledge which I find absolutetly horrible and impossible to make reality.
and they rely heavily on friends and family to help them out whenever something goes wrong or needs installing. It’s unreasonable to expect Linux distributions to be so usuable that Joe Grandma can fully administer it, either today or in ten years’ time.
Damn right. If Windows and OS X cannot do it because these people are incapable of learning the concepts and commands, then leave it at that and make it easier for those who administer the systems to dumb them down for them. These unrealistic expectations of “But my grandma wants to print out her taxes and she has to push more than two buttons, confoozing! Linux/Windows/OSX/BeOS/Whatever is just not there yet for the desktop!!!!” are so nuts.
Linux will be ready for desktop use by average users when a full-featured, powerful distribution is offered with a set of tools allowing for complete command-line-free use and administration.
Linux will be ready for desktop use by average users when a full-featured, powerful distribution is offered with a set of tools allowing for complete command-line-free use and administration.
You are spot on, dave. I could not have said it better myself.
I never liked RPM very much though. It’s one of
the reasons why i prefer Debian which comes with APT.
That’s just ignorant.
The .rpm and .deb formats do the SAME thing: they combine program files and dependency information into a single file.
The Debian equivalent of the ‘rpm’ tool is ‘dpkg’. Both install and uninstall packages, but neither can automatically resolve dependencies.
‘apt’ uses the lower-level services of dpkg to provide access to local and remote package repositories and automatically resolve dependencies. As do SuSE’s yast, nd Mandrake’s pmi/rpmdrake, and Fedora’s yum.
After reading over this article ant the appended comments I at least feel that I have a better understanding of some of the same problems I’ve been having with Linux. A lot of people seem to be bashing Mr. LaCroix for posting his oppinion about the state of Linux usablity but I must say that his points as some have already mentioned are at the heart of GNU/Linux being widely adopted as a Desktop OS.
The fact is that Linux, at least the distributions that I’ve tried, are much to difficult to use at the level that I and other experenced windows users can use windows and still be satisfied. The point is that if I want to install some new piece of software to do a particular task, I learn about it and then I download it or retrieve from somewhere and then follow the quickest possible proceedure to get it up and running. Where it seems Linux fails is in the last part, because knowing about dependencies and using the command line shouldn’t be nessecary however it seems that some distros it still is with the default configuration. I should note too that I’m talking about any software made for linux, not just the binaries that are avialable for your particular distribution. Gaim for instance disn’t have an updated package I needed, so I tried to build from source only to find I needed to satisfy a dependency first… ok so now I have to go look for that too? what if that one also has a dependency? The reason why windows users complain about this is because they are not used to the task of installing software to be a chore but instead quick, painless, and rewarding. This doesn’t mean Linux couldn’t have something better than Windows in the future or that some distribution that is hidden under a rock doesn’t already have it.
Another gripe is that I think goes along with the comment made that Linux doesn’t have a place for the Windows Power User, is that once something goes wrong it’s all man pages and command lines. In all my days of using Windows, there hasn’t been one time I could fix problems with my computer and it almost never required use of the command line no matter how screwed up things were. Now to compare that to Linux, if one thing goes wrong it’s time to read a book long man page, sift through a flat organization commands, and memorize a buch of cryptic options and switches just praying that something goes right.
Now don’t get me wrong, I love the OSS movement and Linux however anybody who can’t understand and respect the value simplicity and clarity in OS (or should I say distro) usability may be doing more damage to the community that good.
Thanks everybody for all the great info in this discussion and thanks Jeremy for bringing up this topic before I had to.
I agree with your aricle that RPM sucks, which it does suck. I cannot believe redhat or fedora cannot improve their package management and system management tools.
One suggestion is to switch too Gentoo Linux. One everything is built by source and you can customize your system to your needs. Another clear advantage is Emerge. Emerge tool is awesome for package management and system management. If there are dependencies for a certain package emerge will download them and then build them before building the initial package. Also, you can keep your system updated by using emerge as well. I will not go into detail you can check out http://www.gentoo.org. One downfall is the installation of gentoo. check out http://www.najatech.com for an installation script written in python and using xml it is open so you can get the code and play around with it
Thank you guys for reading my article and for your comments, you guys are great!
i have had to use the commandline in win2k a lot lately. for some reason i got some screwed up files while trying to download them and windows refused to delete em as explorer was blocking. the only way to delete them was for me to boot into safe mode with commandline and delete them that way…
allso, i love the man files. atleast they cover every last bit of a command or file if its in there, and you can access it all from the commandline (and its a godsend if xfree figures out that its lockup time). atleast one have something to fall back on as the entire os can be configured from the commandline. from what i recall, if the gui in windows gets messed up its reinstall time. in fact windows users are so used to doing reinstalls that you will find them reinstalling linux over and over if they mess up something small at times…
like i stated before in this post, the only thing rpms are missing are url info about their dependencys when and if the cds or ftp servers dont contain the needed files. other then that i would love to see rpmdrake/urpmi in mandrake have the ability to be deliverd a rpm that is outside of the known packages, like say “urpmi -if (install file) somefile-1.2.3.rpm” and a browse to file function in rpmdrake. maybe you can even link rpm files to rpmdrake in kde/gnome that way so that when a user doubleclick on the rpm they get a friendly installer that requests the root password (alltho it would be interesting to use these kinds of programs via sudo)…
I just wanted to answer jeff’s comment. It doesn’t seem like many people are bashing the OP. They are giving helpfull comments. You should know the difference and learn to deal with some criticism.
Second of all, why is it such a pain to use a command line? The command line is the reason linux and other unices are much more powerfull and in the end usable. If you want to learn how to use linux, learn how to use the command line. That’s just how it works.
It seems that everytime you hear somebody bashing linux he is a “windows power user” who apparently expects linux to behave exactly the way his windows system does. If that is the way you feel, go back to your windows system. You will be more happy. If you want to use something that is completely (well, not entirely, but a bit) different you should take the time to learn a bit about it. I haven’t met one windows power user that hasn’t formatted his drive at least 12 times in the past year, or that has never messed around in the windows registry, or doesn’t know how a batch file works. But apparently windows is so much easier to use.
I especially agree with the person’s (intentional?) misspelling of this sentence: “n all my days of using Windows, there hasn’t been one time I could fix problems with my computer “. That’s what I know about windows. Mostly you just end up reinstalling the damn thing. DLL hell is ten times worse than dependency hell. Mostly because dependency hell is something only the noobies have to go through.
I think there should be a “new” kind of RPM (Maybe call it “Enhanced RPM”, or ERPM or along those lines) that contains all the dependencies you may need for that program.
The problem with this is that the dependencies go all the way down to the kernel. You’re going to end up with HUGE packages. For example, kbiff, a cool little system tray applet, needs kdelibs. kdelibs needs qt, xml2, gettext, and lots of other stuff. qt needs xlibs, png, and other stuff. xlibs needs fontconfig. Etc. So a tiny 100k program suddenly gets stuffed into 100Mb package! That’s overkill!
man i hate that term. power users tend to people who are self-taught and computer literate. what that means is someone with huge holes in their knowledge base, and alot of rote memorization. linux is a hell of alot easier to a newbie then a windows power user, because a windows power user expects things to look and work a certain way, and if they dont then everything else they know about computers falls apart.
here is where the difference is. a linux “poweruser” learns how and why things work the way they do. a windows “poweruser” learns how. very big difference. if you dont understand why doing something in windows does what it does, then when faced with a similar thing in linux, you will be totally lost and without a clue. if you learn why something does what it does, along with how to operate it, you immediately grasp the underling concept behind any implementation you will ever run accross.
#1 golden rule for all microsoft power users who wish to learn linux. linux is different then windows. windows is built from the gui down, linux is built from the commandline up. that means you can do everything on the commandline, and you can do most things with the gui, rather then the other way around like in windows.
this means invest the time and learn bash. thats the first thing you need to do. ignore stuff like nautilus and konquorer, do all your file management in bash. when you mess up a config file in x, go to runlevel 3 and fix it, dont reinstall the os cause the gui is broken.
as for dependancy hell, please, for the love of god, dont use an rpm based system. compiling from source is a billion times easier and you have will alwas have access to the latest versions of everything. just how hard is ./configure && make && make install? i mean honestly? it works 95% of the time, which is more then i can say for rpm -i. if tarballs are too much, use a debian based system, the debian repositories contain everything you will ever need, and it is far easier to install stuff then windows. fire up synaptic and you have a list of hundreds of applications that are freely available to install with just a double click.
linux is a mess from a usability point of view. windows is a mess from a usability point of view. the only people i have ever seen come close are apple and be, and look where it got them?
as for the whole “erpm” thing, its impossible. you would have to include the dependancy chain all the way down to glibc and then the kernel, which needless to say is wildly impractical. what you want is what windows does, static compilation and mass duplication of code that is extremely similar, both in ram, and on the hd. theres a reason you need the kind of ram you do to run windows… personally, i am happy to sacrifice ~10 minutes to install a new app. i make tgzs of what i want to keep, and afterwords its a matter of an installpkg *.tgz to install the app. how is this hard? how is it harder then windows? it sure is different, it requires knowledge of a very different os, but that doesnt mean hard.
Ok, guys. Look at the following statements from the opinion article:
Opening line:
Linux In 2004 – Usability Needs A Facelift
Closing statement:
That’s what Desktop Linux needs to focus on, the “casual User”. In order to thwart Microsoft, we need a Linux version that does everything it does and better. Microsoft’s ease of use plus Linux, equals the best Operating System the world has ever seen, but unfortunately, I don’t think it’s in development yet.
Now, read this:
kromagg said: Second of all, why is it such a pain to use a command line?
Obviously this guy is not being truly heard. I bet if you stop some people on the street and ask them what a command line is you will probably get a response from most of them like, “uh, isn’t that like a term they use in English class or something?” Only IT geeks like us and and the most adventurous home user would know what the command line is. Much less use it. And “DLL hell” and “dependency hell?” Once again, only those in the same category I just mentioned will know or care what those phrases mean.
To be a direct competitor with microsoft, linux needs to be more usable for the home user, period. Since that is the market we have to corner in order to take microsoft. That is the market share that microsoft owns like 90-something percent of, and the vast majority of that market are users who know crap about dynamic link library, or dependencies, much less care to know. For the most part windows has been able to leave the use of the console to a very small minimum.
In my opinion, we need a distro that rivals and exceeds the usability of microsoft, (i.e. – NO command line, no manual editing of files, and maximum gui use for maximum ease) so we can appeal to and corner that part of the market and win it. And still have our bread and butter distros for those that truly know and love linux.
I agree wholeheartedly with what dave said about 10 posts back:
Linux will be ready for desktop use by average users when a full-featured, powerful distribution is offered with a set of tools allowing for complete command-line-free use and administration.
Exactomundo.
Excellent posts. Kudos!
Okay, everyone repeat after me: you don’t have to use the command line to install software in Linux! There are GUI tools to install software and automatically solve dependencies on all major distributions.
Someone has remarked that you don’t always find the software you need in a distro’s repository. Well, tough luck. There are over 7,000 packages in the Mandrake repositories, and over 10,000 in Debian’s. If you really, really need to use the latest, unstable version of a particular software and can’t wait the couple of days it usually takes for it to get to the software repositories, then you are not considered a “normal user” but rather a power user…
The problem is that Windows Power User want all the experience they’ve acquired on Windows to carry over in Linux. I know, I was one myself a couple of years ago. Now, I’m all for making Linux easier to use for newbies (in fact, my very non-geek gf uses Linux at home now, and finds it as easy as Windows). But we certainly don’t have a responsibility towards power-users. They’re bright enough to have figured out Windows, they can figure out Linux.
Newbies will never have to use the Command Line in Linux. Power Users, however, should consider learning the basics as they will find that it is an extremely powerful tool. I’m a GUI person myself (I remember my unabated excitment the first time I saw a Lisa, oh so many years ago), but I regularly use the command line at home – just because it’s convenient for particular tasks. In fact, I’ve installed Cygwin on my work computer so I can use it on Windows as well.
That’s not to say that the command line is not newbie-friendly either. Remember, there was a time when newbies didn’t have a choice – all PCs used a cli of some sort. And you know what? They managed to learn it pretty quickly. However, you don’t have to use it in Linux – there are GUI tools for all newbie tasks.
Someone said:
In my opinion, we need a distro that rivals and exceeds the usability of microsoft, (i.e. – NO command line, no manual editing of files, and maximum gui use for maximum ease) so we can appeal to and corner that part of the market and win it.
We have those distros already: Lindows, Xandros, Lycoris. Heck, even regular distros you can use without ever going to the CLI (my GF has no problem using Mandrake 10, and she’s as newbie as they get). Usability is not an issue anymore – and in fact, the Mac shows that superior usability is not that effective in gaining over market share.
(BTW, removing the CLI and text files would be pretty stupid – the important thing is to have GUI tools that can modify the text files and perform command-line action, and most current Linux distros have those.)
Anyway, at the risk of repeating myself, it’s not newbies that complain about Linux ease-of-use, but power users. Just like someone who’s come to know everything about a particular app won’t like to change to another app that does the same thing, but differently – whereas a newcomer will be a lot more receptive.
I’ve seen linux work of the computer illiterate and I’ve seen it work for the techno geek. But you say it doesn’t work for the windows power user?
Let me tell you, I was a windows power user! I knew all the tricks and trades to keep windows safe, secure and stable. But guess what even with all the knowledge in the world its nothing compared to the stability, security and ease of use that linux is on the desktop just because there is no ‘tweeking’ involved. The biggest windows power user is the people that play games. Well I never play games on a pc and to the I always say go with a concole. Its thousands of dollars cheeper and you don’t have to worry about security and stability issues!
Does anyone know if there’s in Windows something similar with “dmesg” in Linux (& *BSDs). In other words, is there a way to show the boot messages afterwards in Windows?
Okay, nun.
I use Xandros OS Version 2 BE, and to say that one never has to go to the command line in that version, despite how easy it claims to be, is a blatantly wrong statement.
Now, I have said this on a topic prior to this one, mainly to debunk what Solaris up there was stating on that topic: if one installs a linux distro as easy as xandros with the many default apps and NEVER installs any other software on it, then of course, no one will have to go to the command line. But what if someone plunks money down on a distro, decides he/she wants to install some new hardware, i.e.- ati video card, and needs to install the drivers for it. Then configure it, which I guarantee, requires some command line intervention, because I have had to do it, with the very distro that you claim doesn’t require it. And logitech quickcam usability, which I also have, and am trying to get working with xandros. Oh, a driver (phillips, if I remember correctly) is available, but it requires a friggin kernel recompile, which requires guess what? Right. It is little things like these that can potentially have a huge impact on the use of linux on the desktop.
The problem is that Windows Power User want all the experience they’ve acquired on Windows to carry over in Linux. I know, I was one myself a couple of years ago. Now, I’m all for making Linux easier to use for newbies (in fact, my very non-geek gf uses Linux at home now, and finds it as easy as Windows). But we certainly don’t have a responsibility towards power-users. They’re bright enough to have figured out Windows, they can figure out Linux.
What the hell are you getting at there? We certainly don’t have a responsibility towards power-users? No one’s asking you. And this is about linux’ success on the home desktop. If linux wants to succeed in that realm, changes in ease of use must occur. It doesn’t matter if it’s the windows newbie, casual user, or power user.
Actually the control file for .deb format has more (useful) information than .rpm format, in particular Conflicts, Recommends, Suggests, Replaces and Priority fields. dpkg also has special config file handling. All these allow for the *possibility* of seamless installs/upgrades, though if one is tracking testing/unstable it may be a problem.
Another thing is that although rpm does have a “provides” thing, it does not make extensive use of it. Debian uses it for virtual packages and then in dependencies; e.g. package xyz requires mail-transport-agent and exim/sendmail/postfix… all provide mail-transport-agent. Similiarly, debian makes extensive use of installation scripts to provide alternatives/menu/mime/debconf/defoma … systems.
Someone mentioned about many freebsd’s would have the same problems; but there are already commercial debians (xandros, linspire) and these problems are not there. It is because the distribution model debian (and BSD’s) followed from the start: that of online repositories which allow dependency tracking. So the commercial debians at least try to be compatible in package names and dependency etc. (their problems are different, that current versions track testing to provide latest/greatest) I agree that the problem of debian for *desktop* users is the long release cycle and tracking testing/unstable may not be pleasant always.
I think this problem will also disappear soon as the desktop has matured enough, and one would not feel the need for latest/greatest; after all there are better things to do than downloading/installing packages or latest versions of distributions.
Debian has an open Debian Policy for packages, … and the commercial debians try to follow it. Linux world would have been much better if Redhat had followed the distribution model of Debian/BSD, an open policy regarding packaging and community based repository since the very beginning.
While it’s a good idea in general to solve the dependency issues, the solution offered by the author isn’t. If you would have to bundle all dependencies with each download, they would grow 2;3 or up to 100 x their normal sizes. It would only make the whole network slower (and more expensive).
The ERPM should be based on the ‘bittorrent’-principle, but instead of looking for several parts of the file it should look for each dependency. I don’t know how familiar you are with bittorrent but here ‘s the main principle: the torrent cuts up a file/folder in several pieces. It then gives each piece a checksum. The bittorrent itself is about 2 or 3 kb. If you open it with a bittorrent client it will read the info: how many pieces there are, how large, their checksum and the checksum of the complete file (hash). Applied to ERPM: instead of cutting it up into pieces cut it up into dependencies. Each ‘dependency’ has an unique ID-number. The ERPM managing client should ‘construct’ the file combining all dependencies and packages, so it would ‘build’ the installation for you. But you would still have larger downloadfiles I hear you say. Therefor it would be better to keep the ‘5-cd distribution’. In most distro’s you already have a package manager. If the ERPM management software would also look in that database first for the needed dependencies it would differ alot in downloadsize/speed. For packages not included on the cd’s it then could connect to a main database, looking up the Id of the needed package and rederecting the software to the apropriate downloadpage where the ERPM software would get the package from.
The result should be an executable file that would install like a normal installation file in windows.
Kind regards,
Sjaaksken
I use Xandros OS Version 2 BE, and to say that one never has to go to the command line in that version, despite how easy it claims to be, is a blatantly wrong statement.
Okay, truth be told I haven’t used Xandros, so I shouldn’t have talked about that particular distro.
However, I still stand by the fact that a newbie will never have to use the command line if he doesn’t want to. Of course, if he adds hardware, he should make sure that it’s supported by Linux first. That’s only common sense. If the hardware is supported, then it won’t be necessary to install drivers.
The exception, of course, is video cards. I can’t talk for ATI, because I have a NVIDIA card. If you want utmost performance from a NVIDIA card, you will have to use the command line to run their installer (you don’t have to, the nv driver is set up by default, on Mandrake at least, so you still boot up to a GUI).
However, installing new hardware such as video cards (I’m not talking about plugging in a USB device, here) is not something which newbies will tend to do – rather, that’s what I’d classify as “power user” activity, even on Windows.
Let’s compare apples with apples here: when people buy a computer, Windows is already installed and configured. Newbies don’t install and configure OSes (I can’t tell you how many times I’ve installed Windows for people…). Let’s compare that with a PC with Linux pre-installed and configured. In that scenario, why would a newbie ever need to use the command line?
if one installs a linux distro as easy as xandros with the many default apps and NEVER installs any other software on it, then of course
Uh…why do you talk about software, then immediately after give an example about hardware? In fact, a newbie can install tons of software on a Linux system without resorting to the command line. As I’ve said before, there are plenty of GUI tools to install software that resolve dependencies. Additionally, autopackage.org is starting to make its way among developers (though still slowly…)
Please, if you’re going to make an argument, don’t confuse the issues.
What the hell are you getting at there? We certainly don’t have a responsibility towards power-users? No one’s asking you.
Power users don’t want to change their habits – i.e. many seem to want Linux to behave in a way identical to Windows. If that was the case, why bother switching anyway?
And there’s no need to be all uppity about it. I gave you my opinion, that’s all. My point is that Linux is a different OS than Windows. Newbies don’t have much problems using it, as my experiment with my girlfriend has proven to me. It’s power users that complain.
And this is about linux’ success on the home desktop. If linux wants to succeed in that realm, changes in ease of use must occur.
If that was true, then Mac OS X would be taking away market share from Windows, as it is much superior in ease of use. However, this isn’t happening. Another good example is BeOS. It was better and easier to use than Windows, and yet it withered away.
What Linux needs to succeed on the home desktop is a) preinstalled machines; b) more people using Linux at work, so they bring it home (which is how the PC won over the home market in the first place); c) more marketing presence.
It doesn’t matter if it’s the windows newbie, casual user, or power user.
Listen, I’m not saying that ease of use isn’t important – it is. And in fact, great progress has been made in this area, and more will be made. But ease of use is mostly important for newbies and casual users. Power users will always have to unlearn a few things before they can learn new ones.
Okay, first things first. I am not getting uppity, moo. I was responding to the fact that you more or less stated that windows power users are to be ignored in the effort to simplify linux for the masses.
Power users optimize an OS to it’s maximum performance and usability. I consider myself one. And I use windows still (for gaming and stuff that I can’t yet readily do in linux, i.e.-webcam use with MSN messenger), but I have steadily used linux for a little over a year. So, power users can change and many want to.
However, I still stand by the fact that a newbie will never have to use the command line if he doesn’t want to. Of course, if he adds hardware, he should make sure that it’s supported by Linux first. That’s only common sense. If the hardware is supported, then it won’t be necessary to install drivers.
Cool, stand by it. And if a newbie downloads or purchases a distro based on the fact that his hardware is supported and/or newer drivers are available for it, but he has to download the drivers and install them, then what? Well, I thought my hardware was supported? Yeah, it is. Now, open up a terminal dude, and get to it. Oh, shoot.
Uh…why do you talk about software, then immediately after give an example about hardware?
Obviously you didn’t get that I was trying to give credibility to the huge linux software base included with the majority of distros. One may not need to install anything else on a system with all the apps that linux provides. And that installing apps without command line is possible in Xandros and Lindows, etc. But, search the Xandros forums and see what it is taking for many to get UT2004 running in Xandros.
I am mainly talking about installing drivers and configuring hardware which presents many a headache. Easily getting full functionality out of the hardware they already have (and enjoyed with windows) when switching to linux is what I am talking about. And even “newbies” who’ve been using windows for a while upgrade hardware on the fly using windows, not worrying about why the X server doesn’t fire up after installing a new GFX card for example. This is something that many prospective linux “newbies” have been able to do with ease in windows. And to get much of that same functionality requires what some have said is not necessary, the console.
If that was true, then Mac OS X would be taking away market share from Windows, as it is much superior in ease of use. However, this isn’t happening. Another good example is BeOS. It was better and easier to use than Windows, and yet it withered away.
OS X was pretty proprietary last time I checked and Apple has focused alot on education and in multimedia. To me, that doesn’t leave much option into any lack of ease. It works and works well with the hardware and software made for it. And one doesn’t have to worry about hitting the terminal because there is really no choice unless someone wants to mess around and config themselves. Hence, part of the reason it’s not taking away market share from windows, IMHO.
What Linux needs to succeed on the home desktop is a) preinstalled machines; b) more people using Linux at work, so they bring it home (which is how the PC won over the home market in the first place); c) more marketing presence.
I agree with you there. It can help push linux to succeed. But not in it’s current state.
Power users optimize an OS to it’s maximum performance and usability. I consider myself one. And I use windows still (for gaming and stuff that I can’t yet readily do in linux, i.e.-webcam use with MSN messenger), but I have steadily used linux for a little over a year. So, power users can change and many want to.
Of course. I was a Windows power user myself before coming to Linux. I’m not saying power users won’t be able to learn Linux, and that we shouldn’t support documentation efforts. But power users, however, have to accept that Linux and Windows are similar on some aspects, but fundamentally different in others. The Linux world is certainly more chaotic, which can be unsettling for some.
Take text config files, which normally you shouldn’t have to edit (but may have to, especially if – like any good power users – you want to tweak your system). Some see this as archaic, or not user-friendly. On the other hand, a lot of config file (especially for common apps) are not that complicated, and some contain basic instructions inserted as comments. Most importantly, as text files, it’s pretty easy to program GUIs for, which is why there are GUI tools to set most things up…
Cool, stand by it. And if a newbie downloads or purchases a distro based on the fact that his hardware is supported and/or newer drivers are available for it, but he has to download the drivers and install them, then what? Well, I thought my hardware was supported? Yeah, it is. Now, open up a terminal dude, and get to it. Oh, shoot.
Unless they are proprietary drivers, there’s no need to download them – simply upgrading the kernel will do for kernel drivers. Upgrading a kernel can be done from withing such advanced package managers as rpmdrake, a GUI frontend for urpmi.
For proprietary drivers, well then it’s really up to the hardware vendor to provide packages or GUI installers, isn’t it? It would be illegal to distribute them along with the kernel for free distros (although I think that commercial kernel rpms with nvidia drivers are available to Mandrake Club members).
In other words, while this is valid criticism, it’s not up to the community or distro makers, but rather the hardware ventdors, to remedy the situation.
(That said, the command-line nvidia installer is a charm to use – I don’t see why they don’t make a GUI version…)
I am mainly talking about installing drivers and configuring hardware which presents many a headache.
Well, it depends on the hardware. As I’ve said, the best is to make sure that it’s supported. If it is, installing the vast majority of hardware is a breeze, especially USB peripherals. Note that hardware detection and support has increased dramatically for Linux over the past two years. Graphic cards are a special case, for the reason I’ve given above.
And even “newbies” who’ve been using windows for a while upgrade hardware on the fly using windows,
Uh, no. I think it’s been a while since you’ve dealt with newbies. Except for USB peripherals (which by and large are as easy to use with Linux as they are with Windows), other peripherals can be a headache to install, especially since so many people are still using Win98 or (the horror!) WinMe. Hardware detection has improved with Win2K and XP, but it’s still not a walk in the park for a lot of newbies…
not worrying about why the X server doesn’t fire up after installing a new GFX card for example.
Graphics card are definitely a sore point. However, I haven’t had much problems on that front. I don’t know how fickle ATI’s modules are, but I know that NVIDIA’s haven’t failed me yet.
This is something that many prospective linux “newbies” have been able to do with ease in windows.
Then again, many others have had hard times installing stuff for Windows as well. You make it sound as if installing stuff on Windows always works on the first try – which is often not the case.
Heck, even as a Power User, I couldn’t succeed in installing the supplied Windows drivers for my MSI GeForce4 Ti4400 cards on Win98. Every time I’ve tried to install them I’ve corrupted the system (it would crash during boot, the only recourse being the dreaded Safe Mode.) I’ve also struggled to install such simple things as Network cards…
And to get much of that same functionality requires what some have said is not necessary, the console.
Okay, we agree that there should be graphical installers for proprietary drivers, since they can’t automatically be installed by the distro. Do you agree that it’s the
vendor’s responsibility?
OS X was pretty proprietary last time I checked
I don’t see what this has to do with anything.
and Apple has focused alot on education and in multimedia. To me, that doesn’t leave much option into any lack of ease. It works and works well with the hardware and software made for it. And one doesn’t have to worry about hitting the terminal because there is really no choice unless someone wants to mess around and config themselves. Hence, part of the reason it’s not taking away market share from windows, IMHO.
I’m not sure how the fact that you don’t have to worry about not hitting the terminal is part of the reason it’s not taking away Windows market share. Unless you’ve made a typo, that paragraph apparently contradicts the argument you’ve been making so far…
I agree with you there. It can help push linux to succeed. But not in it’s current state.
So we kind of agree. I just think it’s ready enough for quite a few desktop. Working on ease of use is always good, but to me that’s not the most important aspect of the equation right now.
And, really, ATI and NVIDIA should distribute graphical installers for their drivers…maybe someone could send them a link to autopackage.org… 🙂
Editor- how about requiring your submitters to run a simple spell check on submissions? I mean seriously- ‘abosolutely’! It’s bad enough the authors that you accept can barely write cohesive thoughts; you could at least police the stuff you approve by seeing to it that it contains proper spelling. Please! What a stinky article.
Ok, I just read the editor’s bit about being Greek and not having English as the first language, sorry about spelling, blah blah. Ok, understood, but there really is no excuse for not using a spell check, even if you do a site “for fun” (while taking advertising revenue from some significant advertisers).
Dunno, from what I recall from me windows days, this usability is in some ways a myth. Don’t remember for which programme exactly, but I had to install a newer version of msi, which was a pain in the *** cause I first had to find it etc. Kinda like dependency surfing now… 😉
Also, I once installed Win2000 and SuSE 9.0 on the same box. Win2000 took about 7 CDs and a marvelous total of 14 (roughly) reboots. Whole process took some 2 hours. SuSE needed 30 mins, one reboot, and then it worked. Office, Email, Browsers, everything was there.
To me, that was far easier to install. 🙂
Anyway, I sometimes think Windows usability is a myth. Tried to fix a friend’s CD burner problems once (XP). Turned out that her user account didn’t have the rights to burn, but couldn’t find where to do that. Now, with KDE I know there’s KUser, which is fairly easy to use. XP? Dunno…
Malte
!@@#%$@#^$@#
“Linux could be better”
<snip>
“…RPMS suck…” (paraphrase>
<snip>
RPMs suck. Yes. I agree. RPM = RedHat Package Manager. Therefore, if you used a different (and not RH based) distro, you wouldn’t be using RPMs. And most anything is better than RPMs.
I don’t understand this article… If you want ease of installation, there’s so many package managers to choose from. It’s been done to death in the comments, so I won’t go into that.
But the other point is configuration. I can spend hours upon hours looking for a menu item linking to a dialog, and that having infinite tabs, just to configure one single setting in Windows. Often, in most distros, the config. tools are so centralized, that you can make these changes from a single place. And if you’re like me, and prefer text files in /etc (or wherever), you can go edit them by hand. Configuration of every distro I’ve used is far easier than in Windows.
That 3 or 5 CD SuSE you tried using? You do realise that you’d have more than that number to get a complete Windows system up and running, don’t you? I’ll enumerate: Windows – 1 CD. Office – 2 CDs (yikes! Oo.O is 79 MB, last I checked). Photoshop is on another CD. Winzip, the same. And then, you’ve got to go on the net to get an IM program (if you don’t use MSN), update your system (don’t forget to install the blaster and sasser patches in XP) and so on.
I use Gentoo, so there’s no comparison for CDs. But Debian, which I’m going to build for a friend, is 2 CDs to get it completely up and running, and one more for some other programs he might want. That’s it. One shot.
And you’re claiming that Windows is “easier” to setup? Hahahahaha!
You ranted about dependancies. Well, every installer in recent times resolves dependancies beautifully. Also, like others have mentioned, dependencies in the package are going to be bloat worse than XP!
So, there you go. All the reasons why you’re not talking sense! 🙂
So here is my approach: Desktop Linux needs a set of standard applications and libraries that are used in many apps (call this the “core system”). It would for example include the Kernel, system libraries, and kdebase. This “core system” should be agreed upon, released yearly, and be available in all distros. The “core system” would be installed in the usual Unix filesystem hierarchy (e. g. /etc, /bin etc.)
Now, all additional components should come in “self-containing app directories, or AppFolders”, that is folders that contain everything thae application needs to run on the base system, including libraries, sounds, etc.
An AppFolder would not have to be “installed”. It would just be copied anywhere to _any_ distro that contains the “core system” and would just run. No apt or rpm needed, no “dependencies” other than the “core system”.
And stop, before you say “impossible”: Apple has this with .app folders, Rox does something similar on Linux and finally, on klik.berlios.de you can download AppDirs for Knoppix (which in this case serves as the “core system”).
In fact, it would be quite easy to change to this system for Linux globally. Small “core systems” (like Windows), and self-containing AppDirs.
The problem with Linux is that the filesystem layout is designed for the programmer, not for the user. “Package managers” are just a crude hack around this.
Mac OS 1985 had it this way: A “System folder” where eyerything was what the system needed to run. And “Application folders”, e. g. “MacPaint”, where the MacPaint application resided, along with all what it needed to run.
That’s the way to go. Not make it more complicated, make it easier! “Installers”, “autopackage” and the like are the exact wrong way because they do not remove, just hide the real mess.
Linux 2004 should learn simplicity from Mac OS 1985.
Why should I care where my apps are? I’d rather have an integrated package and update manager. Also, what about multiuser systems (increasingly common, even for home desktops)? If I want to install some apps for everyone, and other apps just for some users?
“Hiding the mess” is what operating systems do, even the classic Mac OSes. What Linux does is that, instead of “bolting the hood shut”, it lets you look inside if you want. Most people won’t, but some do. Choice, in other words.
Also, having apps as folders is not efficient when programs share libraries – let’s say there’s a vulnerability in a certain library. If each program uses its own version of the library in their own folder, how do you go about updating all of them easily?
There’s nothing wrong with the current package systems, as long as we have intelligent package managers to resolve dependencies and such.
If it ain’t broke, don’t fix it…
Get redcarpet. It deals with all the dependency
issues u have – just select a package and press
install.
Btw. windows servicepacks and not the same as
linux packages. Servicepacks only contain software
from one company. Software based on multiple rpms
usually depend on work by many different independent
programmers og companies. Windows has nothing to
offer the kind of service that redcarpet has.
Remember what ur mathteacher told u –
u cant compare apples and peares.
It must be possible for each app to have its own version of the libs it needs. If appA needs libC-0.96 and appB needs libC-0.97-fzg and both libs are incompatible – there _must_ be a way to have both installed!
And 3rd party software packages must be binary compatible between _all_ distros (this can only be achieved by not relying on package managers…).
It must be possible for each app to have its own version of the libs it needs. If appA needs libC-0.96 and appB needs libC-0.97-fzg and both libs are incompatible – there _must_ be a way to have both installed!
It is possible. In fact, on my system, I have both KDE 3.2.2 (for everyone) and KDE 3.1 installed (for a single user). I didn’t use a package manager in the second case – it’s not because you have a package manager that you’re forced to use it, you know…
And 3rd party software packages must be binary compatible between _all_ distros (this can only be achieved by not relying on package managers…).
You mean, like OpenOffice.org? Or various software that use the Loki installer? These are compatible with package managers, in the sense that one can have both type of software installed on their Linux box.
However, I disagree with you when you say “it must”. It “must” do diddly squat. If distro makers want to package a third party software for their distro, they’re free to do it. Personally, I like the package system a lot – it takes a lot less time, and lets me easily track what’s installed on my PC. I know it’s different from how they do things in Windows or Mac, but that doesn’t mean that it’s not good.