What is a distribution and how does it differ from the distribution next door? Do they provide a different-enough experience to the user who is in search of a capable desktop?Note: The author is Norwegian, please excuse any grammar/syntax errors.
What is a distribution? A distribution is the thing that makes Linux an operating system – in addition to the kernel it includes basic tools to put together an operating system. As my technical reference puts it, “software designed to control the hardware of a specific data-processing system in order to allow users and application programs to make use of it.”
Each distribution has its own goals. While a commercial product might want to hit a more mainstream group of users, a distribution can be targeted against anything from a huge data server to a client system on a USB hard drive. Commercial companies like Microsoft and Apple tries to encapsulate the main hop, and has succeeded, dare I say.
For a desktop user, there are many dependencies, that is, the user expects that the system is equipped with all the tools required to use, maintain and upgrade the system – programs for doing the main tasks and the managing of software and hardware. The programs should be strong and flexible, and the settings connected all-together in an easy-to-use panel. No distribution reassembles these factors today.
Why do they not? The open source applications are reaching a very high-quality level – I would say that the most recent version of the K Desktop Environment (KDE) is in itself easier to manage than the Windows environment. The problem many new users experience is the huge leap in software architecture – libraries, source code, the quite complex way of installing programs and the managing of the kernel. The Linux kernel is high-quality work – the problem of it lies in the complicated way to upgrade and maintain it.
But this architectural leap as mentioned, first of all the organization of files, folders, settings and the understanding of environment variables, is the base of Linux. It cannot be changed, but it can be made more understandable to the user. Mac OS X managed it, although it is fairly different (based on another branch of UNIX), the system, software and hardware management in this system is ideal for the desktop.
Many of the commercial distributions, of Xandros and SuSE tries to reassemble this – and has come significantly closer in simplifying the architectural leap, but not as good as the complete novice would be comfortable. For the Linux to rival Microsoft and Apple on the desktop, standards must be developed. The programs that are included in most modern distributions are superb, but when the user wants to add more possibilities to the system – for example by increasing his hardware, or by applying for more recent versions of software, he gets stuck – either in missing libraries, or a dependency problem. And the usual question after a somewhat succeeded installation, “Where’s my program, I can’t see it?”
Experiences from modern distributions
The management of software was made very easy for me as a desktop user when Xandros released its second version around Christmas last year. The process of installing had now been simplified significantly. Although I of course am aware of, that this way of management is not unique for this distribution, and that the methods have existed for many years, this was the first distribution that made it easy for me, a quite experienced user to seamlessly manage my system.
My experiences from Red Hat and Fedora was of professional and mature ones, but not ideal for desktop use. At first, nothing comes out-of-the-box. The immediate support of multimedia – is very important. I shall also mention that I have tried Lindows, and that my experiences from this one were rather bad. I felt that the track was laid for me, that I was not in charge of the system, and that the possibilities were rarely beyond those defined by the distribution manufacturer.
Click-And-Run is indeed a very nice system, but too commercialized for me to use– then I would rather stick to Mac OS X or Windows. Although Xandros also is a commercial distribution, I felt that Xandros Desktop was more of an all-rounder than Lindows, and much easier to use and with much more potential and possibilities. In an operating system I want to do whatever I want – I felt claustrophobia in Lindows, and liberated in Xandros.
However, the biggest and most serious problem in all distributions is, that when one is released, the kernel has either moved to the next stage, included important hardware support, or the software that’s bundled is out of date. The very first version of Fedora, dated November last year, is now old, providing the kernel three from 2001 (2.4) and the older version of KDE which through the last months has gone through a complete refreshment. A novice would be completely helpless in upgrading this system to a more recent state, without installing a more recent binary version from a CD, or through a network.
The Challenges
The challenge is to emerge all this fantastic software, into a superb distribution, where one can combine strong elements from each product. Last week Novell, the owner of Ximian and SuSE, announced a merge between GNOME and KDE. While I think both sides have strong points, I think a merge between those two would be great to less experienced users. The freedom of choice is the main aim in Linux, but making the choice easier for a novice is a requirement if Linux shall succeed on the desktop.
The challenge of a desktop operating system is to make the core of it very easy to manage and seamlessly to use – and then use the potential of this environment to extend the possibilities by adding more software. I think that is where Linux is lacking from the desktop users today, and where Microsoft and Apple have succeeded. Although Linux is a very good alternative for experienced users, the desktop distribution maker’s challenge is not only to do deploy easy-to-use applications, they have to clarify the architecture of Linux, and put together the different pieces of it. This is not the software developer’s job. While the developers should be coordinating their products in accordance with for example FreeDesktop.org – it is the distribution manufacturer’s job to shape the work offered by the developers.
As you can see, I do not think that there are too many Linux projects, but that there is too many distributions trying to assemble these projects.
About the author:
I have been using Linux regularly since 1998. Missing a fast internet connection, I was during the first years of my career stuck with a retail version of Red Hat. Over the past two years, I have been trying most of distributions.
The article reads like it was translated from Norwegian into Swahili and then into English with one or two other languages in between. Sorry for the cranky comment, it’s just that the article was so bad that one couldn’t even tell what it was about.
I’ll mod this comment down myself, thereby joining every other comment on this story so far…
Heh. I’d love to see you write something in the author’s language and to do a better job.
“Sorry for the cranky comment, it’s just that the article was so bad that one couldn’t even tell what it was about.”
I did understand the article. Maybe all those native english speakers should just try a little harder.
Nice article Magnus!
since I’m not an english syntax and spelling Nazi like some others on this site, I actually tried to read the entire thing…I could be way off but is the whole article about standardizing the best parts of linux so even a n00b could use cool features i.e. like updating a kernel?
foo
Oh, yeah, we really need a 6 DVDs Special Edition Linux having everything on board, from the future 900 megs KDE merged with the Gnome Desktop, including all the developing software ever created for linux + all the games/fortunes, etc and a quad cpu version of OpenOffice in order to please a windoze user on his journey to the outter space… common, do you remember those days when linux was about small/fast/secure? what happened?
I see no reason for linux to look/feel like windoze in order to be used. It’s an alternative after all…
It’s just me after a few days of playing with ucLibc and getting my memory and hard drive back 😉
Its the distribution I’m using. Fully up to date with Kernel 2.6 and KDE 3.2. A large amount of software as well. I have been using Linux for almost three years now and out of all the distros I find Mandrake the easiest. 10 isn’t as buggy as 9.2 was and the official version will be out soon. I recommend it to both beginners and experts.
Depedancies are automatically resolved, program icons are automatacally installed on the “Yellow Star” menu once installed. Software is easy installed by Clicking Star,System,Configuaration,Packaging,Install Software.
Installing Hardware is a simple case of plugging it in, and *poof* an icon appears on the desktop. My devices such as a scanner, digital camera, printer and TV card all just work.
Most of your problems are solved in Mandrake, and I personally feel terrible using Windows after using Mandrake 10, because Mandrake is so much easier its painful to use windows!
What is the guy talking about! Here he is talking about Lindows
“Click-And-Run is indeed a very nice system, but too commercialized for me to use I would rather stick to Mac OS X or Windows”
I find it odd that he says Lindows is too commercial and would rather use windows and MAC, which is the most commercial OS on the market today.
How did he come to his conclusions, what standard questions did he ask, how did he do his comparison. What was his objectives when looking at these very different distros.
The mind boggles over some of these people who seem fit to write an article on Operating Systems. Please tell him to go back and do it again, but this time with facts and objectives.
I actually liked this article. Usually any time I seem someone not from the US review Linux they give it a 10 of 10 for not being Microsoft. I thought the author made some good points in the article, such as: “As you can see, I do not think that there are too many Linux projects, but that there is too many distributions trying to assemble these projects.
I believe that part of the reason there are so many Linux distros is becasue having your own operating system us bragging rights. It used to be 5 years ago you could brag about having a web site, now everyone has a web site.
Regardless to the issue of translation, some of his information is just plain wrong. Gnome and KDE aren’t being merged, go read the releases again.
As for to many distros, for the love of all that is holy stop harping on this. People make distros because they want to .. thats their RIGHT when it comes to FOSS. They can do what they want within those rights granted to them by the licenses they use. So if someone wants to start Bobs Linux .. GO Bob .. have fun. People like this guy and Jimbo seem to think that others shouldnt have the right to use THEIR time the way they wish .. instead they should contribute to some project that they think needs assistance. Guys its up to the individual what they contribute and how, if someone wants to make a new distro .. its usually because they don’t like all those other distros. Do you get that? They tried them didn’t like them and went out to make what they think is the best.
This is good. See things that come from this, such as Synaptic. This was created by a Linux distro group, and now others get to use it too. Thats how the system works. If you dont like that system and want ONE VISION and ONE WAY .. go pay for that and get that from MS or Apple. But dont think that your continuing whining is going to make people stop trying to make something their way. Wake up .. its not so bad when you actually understand the system. You just have to give up the tunnel vision for a little while.
Peace
Er, an essay perhaps? If it is, it looks like he would need to do some more reading. I believe it would have been better if it was posted in the forums first where the discussions to follow could be constructive/informative. That it was ‘published’ as an article… well.
I have nothing against the author but this ‘article’ sorely lacks substance, IMO.
1. Less bloatware included with Distro. A 6-CD pack or a DVD when Windows2003 Server fits on less than CD? Excuse me!
Yeah, I know it is better to bundle all that compiles and ensure it works in your distro than make end user suffer dependency hell, but it leaves bad taste.
2. Less distros. When http://www.distrowatch.org is reduced from 100+ distros to 2-5, then Linux will be ready for desktop: because then third-party developers can manage it. Having one package tested on 100 distros is not an option. And no, giving user sources to compile instead of binary install is not a good idea.
3. Less hot air, more substance. Linux has bugs, get over it. Linux developers are humans, it takes them days and weeks to fix bugs- not milliseconds. Millions of users can view the sources but at most few hundreds will be interested to contribute code.
4. Money does matter. Documentation writers want to be paid. Testers want to be paid. Managers want to be paid. Tell end users they are supposed to pay for the product if they want the product to survive and improve.
5. M$ is not an enemy. M$ is not an Evil. M$ is not the worst thing which could happen to the world. FOSS is not about destruction of M$, does not matter how much it pleases Oracle and what is the hidden IBM agenda. FOSS is about software by the people to the people.
6. Stop bending for Big Corporations. No, better kernel 2.6 scalability on 8 processors and better managing of hundreds of Gigabytes of RAM does not impress me. Yes, it does impress IBM and Oracle, but for my sore Celeron 700 with 512 Mb RAM and 5200 RPM disk the fact that kernel 2.6 can write to SCSI drives faster means that developers are not interested in me, the end user.
It does not contradict with “Money matters”- but you put yourself into position where you tell end users “it’s free, free to share, free as in free” and then have to ask ruthless corporations what do they want- because your money comes from them. So, you sold your soul to IBM and Oracle today and the government of China tomorrow.
People can wait, right? Wrong!
7. Remember, it is just OS, part of a moderm office equipment, not a religion. Not supposed to be.
If someone does not like yours: tell them politely they have a choice to use what they like and a freedom to express their opinion, no matter how wrong or ridiculous or uneducated it could be. Peace, man.
eidtorials are often the editors of diffrent newspapers own personal view on something, its often vague and up in the air when it comes to solutions and so on. this isnt a article and therefor dont need to be accurate more then on the basic level.
hmm, i wonder, have mandrake tryed to make urmpi/rpmdrake able to say take a rpm package that is clicked to launched with urpmi “packagename” and handeld it like it was part of urpmi/rpmdrakes existing list of rpms? ie, solving dependancys and so on for that rpm based on what rpms that it knows about and then requesting cds or getting them from servers. and if not available then request them from the user in some way. it could in many ways help with the task of installing new stuff in linux as the rpm would for all the user cares act like a install.exe file in windows (just click it to install).
hmm, i allso wonder if one could run config tools in sudo so that the user didnt have to input the root password every time he want to use them. these two parts should help in the useability department i think.
then if we could have one unified place to put a symlink to add a item to every desktop/windowmanager app menu (should be nice to have in the lsb i think) then the problem of makeing install files for diffrent distros become that of the kernel and the need for a recompile of every driver that is not part of the kernel when you modify the kernel somehow, and that problem is only for companys like nvidia.
Very well said Russian-Guy. Props.
> Less bloatware included with Distro.
Very good point. IMO consumer distros should just provide one system to the user. One DE, just server or just desktop, … not all-in-one distros like SuSE or Mandrake.
> Less distros. … to 2-5
A very bad idea IMHO unless they’re like gentoo: Less distros -> less different systems -> less testing.
> No, better kernel 2.6 scalability on 8 processors … my sore Celeron 700
Scalability would interest you if you had a P4 with hyper-threading.
> Less hot air, more substance. Linux has bugs, get over it.
Who said the contrary?
> Peace, man.
Stay calm.
You forgot a ‘c’ in English. That should actually read:
Sprechen Sie Englisch?
1. Less bloatware included with Distro. A 6-CD pack or a DVD when Windows2003 Server fits on less than CD? Excuse me!
Yeah, I know it is better to bundle all that compiles and ensure it works in your distro than make end user suffer dependency hell, but it leaves bad taste.
Umm, that’s not bloatware. If you had to install every single piece of software on that 6-pack CD, then yes it would be. But the default install doesn’t require anything close to that.
2. Less distros. When http://www.distrowatch.org is reduced from 100+ distros to 2-5, then Linux will be ready for desktop: because then third-party developers can manage it. Having one package tested on 100 distros is not an option.
There are not very many main Linux distros: SuSE, Mandrake, and RedHat I would say are the top three.
And no, giving user sources to compile instead of binary install is not a good idea.
Amen! I hate when people suggest the ./configure && make && make install dance. Not acceptable. However, that’s why the AutoPackage (see autopackage.org) is being actively developed. It aims to solve this very exact problem.
5. M$ is not an enemy. M$ is not an Evil. M$ is not the worst thing which could happen to the world. FOSS is not about destruction of M$, does not matter how much it pleases Oracle and what is the hidden IBM agenda. FOSS is about software by the people to the people.
Your opinion. I believe they try to stifle the competition and break standards.
6. Stop bending for Big Corporations. No, better kernel 2.6 scalability on 8 processors and better managing of hundreds of Gigabytes of RAM does not impress me. Yes, it does impress IBM and Oracle, but for my sore Celeron 700 with 512 Mb RAM and 5200 RPM disk the fact that kernel 2.6 can write to SCSI drives faster means that developers are not interested in me, the end user.
Stop whining. Does improving performance for servers inhibit your performance? Umm, no. Everybody I’ve heard from says 2.6 performs even better on their desktops. So what the hell do you expect? If you don’t like it, use something else. Please.
You’re forgetting something here… the linux distributions are not just redundant in offering choice (the ones targetted at Joe Schmoe often *do* standardize on one desktop i.e. lindows) they offer *more* than the MS Windows CD you get… more games, applications (at least one office program, almost all of which are more powerfull than wordpad) among other things….
6. Stop bending for Big Corporations. No, better kernel 2.6 scalability on 8 processors and better managing of hundreds of Gigabytes of RAM does not impress me. Yes, it does impress IBM and Oracle, but for my sore Celeron 700 with 512 Mb RAM and 5200 RPM disk the fact that kernel 2.6 can write to SCSI drives faster means that developers are not interested in me, the end user.
and
It does not contradict with “Money matters”- but you put yourself into position where you tell end users “it’s free, free to share, free as in free” and then have to ask ruthless corporations what do they want- because your money comes from them. So, you sold your soul to IBM and Oracle today and the government of China tomorrow.
People can wait, right? Wrong
So the kernel developers now all work for big corporations ?
Your statement is exactly the reason that Linus left trans meta and joined osdl. Companies like IBM can add anything they want, employ as many developers as they like to get something into the linux kernel which will improve performance for their servers. Thats the whole point of opensource, Wake up. Are you trying to honestly say no improvements were made for the desktop ?
so what was all of the premption / low latency patches about then ? why has hardware support for desktop’s improved why has there been so much work into improving performance for desktops.
Im sorry your statement is absolute crap. When did they sell their sole ? Linus is a millionaire and became a millionaire before IBM started funding so many developers into improving scalability, SGI has been adding loads of inhouse scalability patches. That has nothing to do with the core linux developers. Linus himself has pointed out they are working on improving the kernel specifically for desktops, probably because big companies are paying to improve the server side of the kernel.
You dont like the opensource model, where any one with half a skill set can produce a distro, well then dont use it. Dont forget properiatery vendors only have to deal with the big four imho
SUSE, Mandrakesoft, Redhat and Debian.
Linux is all about choice i truly hope choice is not removed from the user. If you like the windows or apple model of a fixed choice, then so be it. I want to be able to pick and choose and linux gives me that right. Even gives me the right to produce my own special distribution. Freedom is a great thing. If your Russsian, how did communism treat you ? did you like the fact you were given no choice in having a religion in believing in GOD, in being dictated to about what you can do ? you didnt ?
If i develop a program and i release it under the gnu/gpl i dont expect to get paid, if some one decides to pay me to add features they need all good. But money is not the be all and end all of software development, if i can do it just for my own self gratification, and have users that enjoy using the program then that is my payment. If i want to make software from a program i would stick it under a eula and charge. Or even dual licence my gpl program, so that people that can pay for windows, can pay for my software. Do you get the drift. Its not as cut and dry as you try and make it sound. Yes we all need money to survive, but money is not the be all and end all of our existance. You may not agree with it thats your choice, but for you to dictate to me that i must want money to write documentation or that i want money for my software, or whatever is just blatent crap. Maybe i want to write a how to after working out how to get something set up. Maybe i want to support the community with my efforts, for what they offer. Whether it be with writing a domcumentation, or even paying money for a distribution. If i dont want to do anything i dont have to. I dont even have to pay for anything, do what you think is right. But dont force your view point as the view point of every one.
Russian Guy: forgot one biggie – you didn’t even mention the ridiculous GUI latencies that, IME, every single window manager suffers from. I think it’s Xfree’s fault. Top-end hardware and lean, mean programs should never suffer from GUI lag when there’s almost nothing running in the background.
I’ll not ditch Win2K until I see some compelling evidence that foss developers actually care about making gui interaction a thing of joy; I really like the direction of GNOME (ie, simplifying it – opposite of KDE) but it still sucks. Users like responsiveness.
That’s my big gripe. That’s about it.
forgot one more: how about detection/support of 5-button mice? I’ve yet to find a distro that does this. Sure, you can always go to your XF86.conf and possibly add some lines of dubious text from a newsgroup post that “WFM” but left others with a broken mess, but I think it’d be nicer to just have it work.
I use my mouse all the time, and going back to 2/3 buttons is like using a Mac, for all the good that those extra (and incredibly handy) buttons will do me.
Wish somebody, anybody, could make it work.
Detection of mouse should go in Project Utopia if you ask me. I don’t know wether such can be autodetected immediately, or only when one of the additional buttons is pushed.
1. “Less bloatware included with Distro. A 6-CD pack or a DVD when Windows2003 Server fits on less than CD? Excuse me!” — Linux Desktop, Windows Server? Excuse me! First of all this could be because the CD’s support multiple architectures as is the case with Debian GNU/Linux and OpenBSD. Second, why does this hurt when you buy it? Third, when you download it, you don’t have to download all the CD’s And fourth, you can always do a netinstall with: 6 floppies with Debian GNU/Linux (x86), 1 with Slackware/OpenBSD (but in that case you’ll have to get the right one, with the right drivers). YMMV.
2. “Less distros. When http://www.distrowatch.org is reduced from 100+ distros to 2-5, then Linux will be ready for desktop: because then third-party developers can manage it. Having one package tested on 100 distros is not an option.” Ever wondered when the world would be a better place: when there were 2-5 countries or when there were 100 countries? With 100 distributions, of which not 100% is for the desktop, there’s choice enough, and unfortunately for you there’s nothing you can do to take this choice away. I don’t give one rat wether you like this available choice or not, i only wish people find a pleasant Linux distribution in their effort. “And no, giving user sources to compile instead of binary install is not a good idea.” Gentoo seems to do fine on this. They appear to have good documentation and a good support forum. I don’t like to compile from source either, but it tends to be popular at the slightly more experienced user. You don’t want to take their choice away, do you?
3. “Less hot air, more substance. Linux has bugs, get over it. Linux developers are humans, it takes them days and weeks to fix bugs- not milliseconds. Millions of users can view the sources but at most few hundreds will be interested to contribute code.” The point of this writing is: …?
4. “Money does matter. Documentation writers want to be paid. Testers want to be paid. Managers want to be paid. Tell end users they are supposed to pay for the product if they want the product to survive and improve.” Oh and Eugenia agrees with this yet she chose not to pay people who write for her site from which she profits. Something ain’t right here. (‘im doing what you propose, though).
Anyway, i prefer not to — i prefer to say something like “You are using software which was given away to every person on earth in the hope someone will improve the software. You can help improve the software by writing documentation for it, developing the software, donating money, give feedback to the author, review the software”. Notice how i shift from the usage of the software from various contributions, which is the global point. Ofcourse there are more options, depending on the situation. More in-depth examples are also Good. This, instead of hard wording with only 1 choice like you put it “supposed to pay”. There are other ways to contribute too, and nobody has to.
6. “Stop bending for Big Corporations. No, better kernel 2.6 scalability on 8 processors and better managing of hundreds of Gigabytes of RAM does not impress me. Yes, it does impress IBM and Oracle, but for my sore Celeron 700 with 512 Mb RAM and 5200 RPM disk the fact that kernel 2.6 can write to SCSI drives faster means that developers are not interested in me, the end user.” I don’t see how improving SCSI performance is bending for big corporations, unless you mean the people who benefit because of that which would mostly be servers using SCSI which are primarily used at corporations. What’s the problem with that? Linux isn’t only about desktop computing. You better start living with that fact or you’ll experience more headache every time some improvement on the kernel doesn’t directly benefit you or the desktop. Here, you were either trolling or plain ignorant.
“It does not contradict with “Money matters”- but you put yourself into position where you tell end users “it’s free, free to share, free as in free”” I think it does contradict because regulary money involves other priorities than freedom has. The freedom you state is already guaranteed by the GPL, and like i explained earlier the freedom doesn’t involve that one has to pay for the software.
7. “Remember, it is just OS, part of a moderm office equipment, not a religion. Not supposed to be.” It isn’t an OS, we are talking about multiple OSes (“Linux distributions for the desktop”, according to you hundreds). It also applies on the BSD’s. These don’t come necessarily with Office equipment, and Office equipment ain’t necessarily the most important part for the user.
—-
What the “Linux desktop” (the FLOSS desktops, rather) needs imo is:
1) Project Utopia: easy, plug & play hardware usage.
2) Freedesktop.org X server with alpha blending, composite, or more looking into alternatives like XDirectFB (FAST!); perhaps something like project Looking Glass, in C.
3) More popular, native games perhaps and other niche projects maturing.
4) x86 emulation on non-x86 architectures; WINE maturing; making it easier for users to contribute (i have some specific suggestions for that); non-proprietary GUI WINE administration program; native NTFS write support, without hacks.
5) Generally more contributions from non-coder people, especially in the form of feedback, bug hunting and documentation. The bridge should be short. However, it exists, it has always been, and it appears to continue. I hope it does.
6) I can’t think of any sound-related issues, except supporting soundcards which are supported by OSS but not (correctly) with ALSA. If they want to make ALSA the default (and OSS perhaps vanish), ALSA must support at least the same hardware equally enough. Can’t speak for *BSD.
7) Plugins for kcontrol/gconf so that browsers can configure the system (“Webmin for your DE/WM”)
And my personal, for the majority most likely less important ones include:
1) Less flamewar by users of various DE’s and WM’s; more looking to alternatives of the big players; thus more diversity.
2) Support for a number of unsupported SGI/MIPS computers which support was dropped as of IRIX 6.5.23. Luckily, SGI helped the Linux/MIPS project with the Indy/* Indigo2/R4400. NetBSD is also catching up.
3) A full, FLOSS Java implementation (thus no more emulation on *BSD, and for non-x86) so i can contribute developing Mir.
4) OGG Theora; native, legal Real/WMV9+DRM/Quicktime support (thus also support for non-x86 and *BSD).
6) Native The Palace; Native Rez.
7) Libvisual!
Open Source means anybody including Microsoft can contribute to it. It is never going to be Unified. Forget about it. There is never going to be one desktop. Just give up on that. There is never going to be one Distro. Nope. It is never going to be Apple’s OS X or Microsoft Windows. It just ain’t happening.
Open Source is not just Linux. It is BSD, OS X, Linux, Windows operating together seamlessly. We shouldn’t be fighting for another OS X or Windows on Linux. Instead, we should be fighting for KDE, GNOME, XFCE4, ICEWM, *BOX working seamlessly together on Linux, BSD, OS X, Windows, etc, as well as within each other.
Linux and open source is organized chaos. If you want sanity and no choice, by all means, stick to Windows or OS X. But trying to turn Linux into Windows, or OS X or BeOS is just selfish and sickening. And thankfully, it ain’t going to happen. We can learn from their mistakes, but we will be doomed if we repeat them.
Okay, I’m just going to respond to some myths/problems in a few comments and the article itself.
Package management is better in good Linux distros than it is in Windows.
Package management is better in good Linux distros than it is in Windows.
I said it twice because no one seems to understand.
Windows has no framework for installation of packages. Yes, they have MSI files, but all they do is hackishly throw shit all over your system, and keep a short log of where the shit went so that. If I want to get the new version of a piece of software I own, I have to hope the programmer coded a “web updater”, as is popular these days on Windows. Guess what? Linux package managers have the web updater built in, and it works for every program. apt-get upgrade.
Also, good Linux package management systems have an automated replacement for the routine task of “Google piece of software, visit website, download, unzip, run installer.” It’s called apt-get install.
As for upgradeability of the kernel, you can’t upgrade the Windows kernel unless you’re talking about booting off an XP CD and installing it over Windows 2000. So the fact that to upgrade a Linux kernel you need to do make bzImage && make modules && make modules_install shouldn’t be upsetting.
Bloatware: Linux distribution CDs are NOT comparable to Windows CDs. Windows provides ONLY the kernel plus a DE. It does not provide meaningful software beyond that. Every Linux installer has an option for a “minimal” install, minimal in Linux terminology being what “full install” is in Windows terminology. Apples and oranges.
Finally, this idea that “there should be less distros” or even there should be only one distro is ridiculous, anti-capitalistic thinking (of all things). Competition is good. The best distro wins. This monopolistic thinking is exactly what allows MS’ code to be poor without much reprucussion (like in the case of IE).
“forgot one biggie – you didn’t even mention the ridiculous GUI latencies that, IME, every single window manager suffers from.”
I recently noticed this too. I’ve been using abiword on linux for ages, and a few weeks ago i installed the windows version, i was shocked to see that not only did it start faster on windows (instantly) it also felt much faster and responsive than the linux version. I find that very disturbing since it’s a gtk app, and i’ve allways had the feeling that gtk on windows was not very good (i remember playing freeciv a few years ago on windows and it was really slow and crappy).
This is true for OOo too. But I can have all OOo apps opened (calc, math, writer, impress) with multiple docs opened, multiple instances of konqueror running, KMail, the GIMP and an opened KPDF without swapping and lags when switching between these apps. In w2k I only dare to have six windows opened at once, because of swapping and loss of responsiveness. Both OS’s suck in a different way, it dependents on your preference and beleive in open standards.
Yes! Here it is again in case anyone missed it.
Package management is better in good Linux distros than it is in Windows
@Russian Guy:
Having one package tested on 100 distros is not an option. And no, giving user sources to compile instead of binary install is not a good idea.
I agree with this statement, but I think that the answer is not to cut down the number of distributions. It’s too late for the big players like Red Hat, Suse and Debian to change, but the minor distros should work together and support a more universal installation system.
If they have a really great package manager, they should of course use it, but they should also support a common system that allows closed source binaries to be installed without having to include every library statically.
@pixelmonkey:
Guess what? Linux package managers have the web updater built in, and it works for every program. apt-get upgrade.
It doesn’t work for every program. It only works for those programs that the people behind the distro has included in the reposistory!
Try searching for a package for Opera for Debian…
http://packages.debian.org/cgi-bin/search_packages.pl?keywords=oper…
>>Guess what? Linux package managers have the web updater built in, and it works for every program. apt-get upgrade.
>>> It doesn’t work for every program. It only works for those programs that the people behind the distro has included in the reposistory!
Try SuSE Yast – works on every rpm I’ve found.
>> Try searching for a package for Opera for Debian…
Commercial software does work to different rules – yup. I reckon the people who don’t charge for stuff think the commercial boys can sort out their own installers – most do.
>>”forgot one biggie – you didn’t even mention the ridiculous GUI latencies that, IME, every single window manager suffers from.”
Try IceWM – it seems quicker to me on my slow machine.
http://www.apt-get.org:
deb http://deb.opera.com/opera/ stable non-free
deb http://deb.opera.com/opera/ testing non-free
deb http://deb.opera.com/opera/ unstable non-free
Add one of these to your /etc/apt/sources.list and presto.
hmmm, I’m using Suse Yast right now…it’s not that good.
Nicely done psilo!
foo
>>hmmm, I’m using Suse Yast right now…it’s not that good.
Like I said, its done it for me, the newest version anyway seems to autoinstall every rpm I download – but I admit I’ve got most of my stuff in off the DVD anyway except for 3 or 4 off the wall packages. I admit it does have a few other curious habits though……….
I don’t want to hijack the thread, but re:Yast…as far as rpms go, the latest version is ok-ish, today it’s failed me more than helped
But such is the nature of rpms…
For the last year and a half, I keep returning to debian derivatives. I’m glad it’s working for you logdog…having the DVD probably makes things easier with Suse. Do you use any online Suse repositories with Yast? Not for updates, but for actual software installation? I want to try one and give it another shot.
foo
Thanks for the excellent article. The author should ignore the petty mean-spirited comments about his “syntax”, etc. I had no toruble understanding any of it and I’m a professional writer.
The reason Linux seems to copy Windows or Mac is not because those mature OSs have come up with such outstanding insights into the OS of a computer. Rather, those companies have had a chance to presage what a user needs in order to do certain tasks. Someone once invented a music keyboard built like a typewriter. It didn’t succeed. The standard piano keyboard is what we use for music, and the typewriter is what we use to control computers, along with the mouse. These precendents can be dressed up in different ways, but Linux, as the saying goes, needs not to reinvent the wheel.
I don’t remember exactly… but I seem to recall a few years back that using hdparm to change the drives to non-blocking i/o seemed to make all the difference in the world in terms of gui latencies. I am not sure if this is still the case but it was a night and day difference one my OLD celeron 400… there was also a poorly documented option for the kernel having something to do with timer granularity…
Just my 2 cents worth…
1) I have my own distro because everyone else is doing it wrong. I may however have rights to brag a bit on it’s release but that is only a side effect.
2)Windows may come on one cd but if you want to do anything real with it you have to back to the store and buy the, “RESOURCE KIT, (2 cds), OFFICE, (4 cds, or is it 8 now)”. They last complete operating system sold by them was dos 2.0. Apple used to refuse to give up their fdisk and format utilities, at least they are trying to atone for their sins now.
Russiam Man, I hope you could be in charge of a linux project for the End user because you know what linux needs to meet the end user requirements. This is sad some people here see linux as a religion. Some of them seems to be using linux to feel they are good because linux is not easy as windows… Boy i’m glag that most linux projects are not directed by such pple because linux would’nt advance.
“forgot one biggie – you didn’t even mention the ridiculous GUI latencies that, IME, every single window manager suffers from.”
I used to agree with that but not any more since I’ve installeed ML10.0. The GUI is _way_ more responsive on my old PIII 733 with ML10.0 than with ML9.2. It’s probably thanks to the 2.6 kernel.
Well put!
Can’t help much – SuSE’s own repository at suse.de is very comprehensive, but mostly I use the DVD ‘cos we live as far beyond the end of the known universe as you can get in a crowded country like England – so no broadband. Other stuff such as ‘dillo’ (I needed a _really_ lightweight browser for an old machine) I just tend to bang into google with the program name + suse + version, hit download or drag and drop from an ftp site whereupon yast pops up in the browser and I hit the ‘install with yast’ button, and it does.
I suppose I’m a SuSE man thru’ inertia – I started with SuSE 6, when I needed some software for work that I could no way afford, and there were no Windows ports of good OSS stuff like gimp, or blender then. I hadn’t touched a computer since 1984 and was told SuSE with the old text yast was the easiest install – and it was (oh, and Caldera, glad I didn’t take _that_ road). Mostly then if it wasn’t on the 6 CDs I had to compile it to make it work, so I’m getting even lazier.
A friend swears by the SuSE apt/get repository at ftp://mirrors.mathematik.uni-bielefeld.de/pub/linux/suse/apt/SuSE/… – but if you’re on suse why apt-get? (or if you’re on apt-get why suse? – whichever) -end of off-topic burble-
From the article:
“The programs that are included in most modern distributions are superb, but when the user wants to add more possibilities to the system – for example by increasing his hardware, or by applying for more recent versions of software, he gets stuck – either in missing libraries, or a dependency problem. And the usual question after a somewhat succeeded installation, “Where’s my program, I can’t see it?””
This is something I can relate to. Sure apt-get works fine most of the time but.. but.. when it doesn’t its next to impossible to get back to where you started and there’s no way in hell for most of us to know what’s been messed up.
Another thing the author mentioned is the organization of files, folders, settings, etc. in Linux. Yeah, that bugs me too. And on that note, I sure would like to see some rapid development on GoboLinux towards more simplicity of use. That would be great!