I’ve been an Arch user for roughly 3 years. I’m pretty much familiar with it all – The way it boots, its configuration and its package management. I’ve also heard a lot of good things about Ubuntu, and wanted to try it for a long time. So, two weeks ago, I took the plunge. These are my findings.Installation
I’ll have to be frank – I did not install the latest Ubuntu, but used an older version which I had downloaded before. I used the Ubuntu 6.06 LTS CD in order to install the system. I booted the machine with the CD inside, and after a little while I was presented with a graphical interface. It seem to have detected my monitor pretty good, as it gave me a 1600×1200 resolution for my 17″ monitor. I started installing the system using the installation wizard. At first I didn’t quite understand the process, so I had to press the “Back” button a few times during the partitioning part of the installation. This resulted in the installation wizard crashing right at me. I was a bit puzzled. I started the installation again and this time I did not over challenge it – and It worked just fine.
At the end I had the system installed, along with GRUB, which detected my Windows XP installation and added it to the menu by itself.
Upgrading
So, I end up with a Ubuntu 6.06 installed. That’s a pretty old horse by now, and indeed, I am reported that there are 520MB worth of upgrades. As it was already late, I decided to go for the upgrade, and left the machine to do it overnight.
After a restful sleep, I wake up and go check on my machine. The upgrades were downloaded, but hardly any of them were installed – A debconf screen was waiting for my input on some package (which I do not recall), which blocked all further upgrades.
This is the first thing which I believe needs improvements. A package installation should not be interactive, according to me (and to my previously installed Arch). A valid solution would be using default values, which exist for the most part anyway (All I did with that debconf screen was pressing the Next button a few times). Also, it should display all debconf screens together at the end of the entire upgrade, so you just finalize some settings at the end.
After waiting for a while, all upgrades are complete and I’m using Ubuntu 6.06.1. Since I’m a desktop user, I’d actually much prefer using Ubuntu 6.10. So, I fire up Synaptic and add a new mirror which contains Ubuntu 6.10. Luckily, Ubuntu has plenty of mirrors, and I was able to find one in my residence (Israel) which resulted in very good download speeds. In Arch, mirrors were a bit more limited. So I perform a dist-upgrade, and apart from some more debconf screens, the upgrade works like a charm. I reboot, and I can see I am using Ubuntu 6.10. Excellent.
Configuration
I now have a fully recent installation. Time to get dirty and configure things to my liking. First of all, networking. I fire up Firefox – Oh goodie, nothing to configure here. That was nice to see, even though my network configuration is one of the simplest around (I use another Windows computer as an Internet gateway, via DHCP).
Next one – Samba. I start searching the menus, can’t find anything. I fire up the Add/Remove Programs GUI, a very nice one for installing packages. I search for Samba, and find gsambad, a configuration GUI for Samba. It installs fine, including shortcuts being added to the menus. I fire the GUI up. Wow, that’s a mess. It basically spits out every option Samba has to offer, so you can’t actually find anything useful there. I start defining some basic settings there, such as the workgroup. I also try to add a few shares. I then look at the generated smb.conf file – It is loaded with garbage. I give it up, and edit the file manually. So, graphical Samba configuration is a no go for me.
I Then try printers – In here, the GUI is so basic it is also rendered useless. Not good.
All in all, perhaps I’ve missed something there, but configuration is by no means something which can be done by a novice, even in Ubuntu.
Kubuntu
Being the KDE lover that I am, I prefer the KDE desktop rather than GNOME. So I go about and install the kubuntu-desktop metapackage. Everything is installed fine, and I decide to reboot. Unfortunately, when the system boots up again, I am not presented with a GUI at all. It’s all CLI. I need to perform a ‘debconf kdm’ in order to fix that. Again, not something which is trivial to a new user. Once I’ve done that and got KDE up and running, I wanted to change its resolution. I used KDE’s tools to do so, but the settings were not applied – I had to restart X for those. I recall it worked fine in Arch, without restarting.
Package Quality
It definitely feels that the package quality of Ubuntu is sufficient. Konqueror for example, is a lot more stable than it was on my Arch installation. The packages are obviously not the latest and bleeding edge, but they are sufficiently up-to-date, and the seem to be going through a reasonable QA process.
Package Management
Installing Kubuntu, I’ve also received Amarok 1.4.3. I tried to add my collection to it, but it failed indexing my collection, due to a bug which was fixed in Amarok 1.4.5. So, I wanted to install Amarok 1.4.5. One of Arch’s strengths was the ease of building packages. I wondered how to do the same thing in Ubuntu. The entire thing seems a lot more complex than building an Arch package – There are 3 forms of building a package for Ubuntu, each way consists of writing several files, where the absolute minimum is 2 files (please correct me if I’m wrong here). There’s a control file which specifies the package’s metadata, and a rules file which is in essence a make file. To be honest, after reading Ubuntu’s wiki, I still don’t quite get how it works. Still, I think Arch’s way is more straightforward, using only one file for most packages, a file which includes both the metadata and the commands for actually building the package.
Another issue I’ve find out in Ubuntu includes metapackages. A metapackage is a way to group a lot of packages together to allow easy installation. For example, installing the kubuntu-desktop package will result in installation of many KDE packages, plus some extra packages which are defined. The metapackage is a nice concept. However, I find its implementation somewhat lacking. From what I’ve gathered, a metapackage is simply a blank package which *depends* on all the packages it actually installs. So when you install a metapackage, the package manager wishes to satisfy its dependencies, and so it installs all of the packages we really want. The problem starts when something goes wrong – Let’s say I accidentally deleted a very important file which is used by KDE. A possible solution would be reinstalling KDE altogether to restore the lost files. This, however, seems impossible to do in a straightforward way – Since metapackages are in essence empty packages, apt is unable to differentiate between a normal package and a metapackage. So reinstalling a metapackage will actually reinstall an empty package – which will result in absolutely nothing. We have to track down its dependencies ourselves, and install them separately. It seems to me that metapackages are nothing but a hack – and no hack should become fixed. A real grouping of packages must be supplied. If any Ubuntu developer is reading this, please take it into consideration.
Summary
So, I’ve been using Ubuntu for 2 weeks now. All in all, Ubuntu works well. The packages go through a reasonable QA process and the system really is stable. Still, Ubuntu is lacking in several respects. The first one would be that I still do not believe it is user-friendly enough for the average Windows power user. Installing new software is simple and easy, configuring software is still lacking in many respects. For the advanced Linux user, I’ve seen areas which need improvements, such as the unnecessary interactivity when installing packages, the metapackage issue, and perhaps simplifying the package building process even more.
I still find Arch to be simpler in many aspects, and I hope Ubuntu can perhaps learn some valuable lessons from it.
If you would like to see your thoughts or experiences with technology published, please consider writing an article for OSNews.
I’m really kinda of surprised to hear you say Arch is easier to use than Ubuntu. Ubuntu is aimed at new Linux users and tries, when it can, to do the most sensible thing it can.
Now, nothing against Arch, but Arch’s own wiki says “Arch is targeted at more advanced Linux users.” In general, most people’s experience is that Ubuntu is the best place to start. I wonder, if you didn’t know anything about Linux, if you would still think certain things in Arch are “easier” than they are in Ubuntu.
Remember, what’s easier to a long-time user isn’t always apparent to a new user.
Actually he didn’t say that Arch was easier to use than Ubuntu. Just that it’s much simpler (in the sense of not being complex), which is absolutely correct, IMHO.
Indeed. Arch is simpler down below. What this means is that if you ever need to change something in a script or something, you will have a better luck with Arch’s clean design rather than Ubuntu’s 15 year old Debian craft.
But when you see it from the higher level, as an end user who would never need to fix anything, then Ubuntu is a better solution for that user.
I am using Arch for 2+ years now and it’s my main distro. However, many things don’t progress as fast as I would like in the higher level part, and so I am also thinking of using Ubuntu in the future. But you see, in order for me to use Ubuntu, I mist never have to delve in its under the hood scripts like I do for Arch to fix something — because I know that if I do, it won’t be a clean design as Arch’s and it will piss me off.
So basically, personally I am looking forward to leave Arch for Ubuntu, but this won’t happen before Ubuntu becomes near perfect in its user experience. Until that day comes, I prefer to fight with Arch’s command line to fix things rather than Ubuntu’s.
Well, Eugenia. I feel much like you do. Arch is simply the greatest distribution I ever tried (believe me, I’ve tested a whole lot of distributions in the past 9 years) but, unfortunately it has some user friendliness “problems”.
Well, they’re not exactly problems in the sense that they’re just not what the distro is focused at.
What I’d really like to see (and even contribute if there were enough people involved — I used to customize distros to meet some enterprise’s needs a few years ago) is a sibling project with Arch, to use the Arch code base to develop a truly user friendly distribution on top of it.
ABS and Pacman (in its upcoming version) would make the job a charm to complete… well… one can dream, right ?
Fortunately I can live without this userfriendliness and will happily live with my Arch system.
I think this falls into the “userfriendliness depends on the user using it” category.
“Ubuntu is aimed at new Linux users and tries, when it can, to do the most sensible thing it can.”
Um…NO. It’s not aimed at new Linux users. It’s aimed at DESKTOP Linux users…Ubuntu isn’t easy as pie for new users..
Out of the gate:
1. They can’t play music, they can’t play video
2. No flash, no Java
3. Gnome isn’t an easy transfer from Windows…which is what most are accustomed to.
SimplyMEPIS is a better solution to this…or PCLinuxOS. But Ubuntu is NEVER about NEW users. You need to get your thinking cap on.
Forcing a user to install something like Easy Ubuntu or Automatix is lame and means that the distro isn’t “aimed at new users”
I agree that Ubuntu isn’t easy enough for new users, but I don’t buy your explanation that Ubuntu is for “desktop Linux users.” After all, most desktop users need items 1 and 2 on your list. I also don’t buy item 3. GNOME isn’t really any more different than Windows than KDE is. I scoff at the suggestion that just because KDE has the K Menu at the bottom left and GNOME has its Application Menu on the top left (by default on some but not all distros), that KDE is more “Windows-like.” They’re both significantly different from Windows and will require some amount of adjustment. I’m not going to dwell on this in the interest of avoiding the usual flamewar.
My explanation is that Ubuntu’s direct markets are the enthusiast and power-user crowds. The ones that love Linux and don’t mind a little tinkering here and there, but not the hobbyists that are willing to really get their hands dirty to achieve the most optimal system. Ubuntu is also positioned as a superior project infrastructure for supporting derivatives with different target markets in mind. In essence, Ubuntu is taking everything that made Debian such a great mother distribution and repackaging these qualities with an emphasis on timely releases and a focus on Intel platforms (plus SPARC servers).
My theory is that all of the Debian-based distros that want the “easy as pie” market will be deriving from Ubuntu within 18 months. MEPIS/SimplyMEPIS and Linspire/Freespire are already onboard, PCLinuxOS and Xandros can’t be far behind, and I have a sneaking suspicion that Mandriva might even kick its RPM habit if things keep progressing as they are.
We have a three-horse race shaping up in mainstream Linux land, and this consolidation is what Linux needs to become even more compelling for proprietary ISVs.
Edited 2007-03-01 21:58
We have a three-horse race shaping up in mainstream Linux land, and this consolidation is what Linux needs to become even more compelling for proprietary ISVs.
Interesting. That’s what a lot of folks seem to think at the moment. In some ways, it could be a two-horse race, with .rpm for the enterprise and .deb for soho (or mainly so). Or a two-horse race anyway: the one that’s vulnerable is SuSE, since it comes a long way second to Red Hat in the enterprise and a long way second to Debian-base distros (if you add them all together) everywhere else. On the other hand, racing is well-known for doping and already certain “trainers” from Redmond have been spotted skulking around the paddock …
I did try to install Arch a few weeks ago as a tester but gave up when the installer said I had to consult a text file then add in the details of what time zone I’m in by hand. Just my 2 cents, but a distro that wants more than hardcore niche appeal has to do a bit better than that. It’s like asking someone to crank a handle before using the telephone. I’m having more fun with Zenwalk.
Edited 2007-03-01 23:02
You might try Ubuntu again when Feisty is released. By default, when Feisty attempts to play an audio file it doesn’t have codecs for, it searches for and displays the package needed for the proper codec. Not only that, after the package is installed, it restarts the media.
The magic part is that is works and works well. Here is a few pictures of the process:
http://www.flickr.com/photos/28313023@N00/375559390/in/set-72157594…
When you install windows, it does not come with flash or java. As a matter of fact (unless your OEM pre-installs them), you have to install them after the fact.
In Ubuntu, firefox will prompt if you want to install java or flash and it seems to work fine. Also, you can install both from the repositories using synaptic or apt (the command line) very easily.
My computer ignorant parents seemed to have 0 problems switching from Windows XP to gnome. The only real issue is that my father needed to be taught how to resize images in the gimp.
So there, everything you said has been debunked by someone who actually uses the system.
The only real issue is that my father needed to be taught how to resize images in the gimp.
I suggest gthumb (image viewer) which has some basic/simpler image manipulation options built-in.
Edited 2007-03-02 13:53
Well, it depends.
If you can tell a user, if you want to change your IP address, change this line in that file and do “/etc/rc.d/network restart”. This is something rather easy and the user can write it down.
Arch is so “KISS” that most operations needed by a desktop user basically aren’t going anywhere beyond.
I don’t think an Ubuntu user without advice would survive a really critical long time. If he would, he is not the “new user” we are talking about. So if he gets some advice, arch _can_ be easier to learn, as it is more straight forward (“here is your editor, that are the 3 files you need to know”)…
Seems like he went through some hurdles getting things updated to the 6.10 release, but overall is seemed to go well. I would agree that GUI config tools for many things in Ubuntu are lacking.
It seemed a little convoluted … the guy actually wanted Kubuntu 6.10, but he installed Ubuntu 6.06, upgraded to 6.10 and then to Kubuntu 6.10. How is that for weird?
Anyhow, it’s commendable that the resulting system worked fine afterwards. Coming from a Windows world, I’m not used to upgrades that work fine (or that work at all for that matter).
The upgrades were downloaded, but hardly any of them were installed – A debconf screen was waiting for my input on some package (which I do not recall), which blocked all further upgrades.
I agree this sometimes happen, but did you just apt-get dist-upgrade or did you use Synaptic? I think Synaptic automatically selects a default choice for you for most packages.
A package installation should not be interactive, according to me (and to my previously installed Arch). A valid solution would be using default valuese
The problem is when you have modified some config files and the new package provides different options. What do you do with that? Is hard to make a sensible choice without affecting the user’s system. Machines are not good to make decisions.
I Then try printers – In here, the GUI is so basic it is also rendered useless. Not good.
I’ve stoped using Samba for sharing printers. CUPS IPP is better and standard. You can setup a printer from Windows XP easily with something like 192.168.0.1/printers/YourPrinterName without messing with Samba. You can also access SMB shares without installing and configuring Samba, if you use KDE.
Installing Kubuntu, I’ve also received Amarok 1.4.3. I tried to add my collection to it, but it failed indexing my collection, due to a bug which was fixed in Amarok 1.4.5. So, I wanted to install Amarok 1.4.5. One of Arch’s strengths was the ease of building packages. I wondered how to do the same thing in Ubuntu. The entire thing seems a lot more complex than building an Arch package – There are 3 forms of building a package for Ubuntu, each way consists of writing several files, where the absolute minimum is 2 files (please correct me if I’m wrong here)…
You should have checked http://www.kubuntu.org fo updates. There you would have found a new Amarok release and how to setup apt sources to download and upgrade. No need to build anything. But if you prefer to build from sources, it’s as simple as apt-build the package.
Let’s say I accidentally deleted a very important file which is used by KDE. A possible solution would be reinstalling KDE altogether to restore the lost files.
You shouldn’t be messing around with system files. If you do, then you probably are above the average user and know how to track down the file you deleted with dpkg and then reinstall the package manually.
EDIT: OSNews guys, I think we are past HTML . How about supporting semantic tags?
Edited 2007-03-01 19:53
Actually, even in regular Ubuntu you can connect to samba shares without having to fully install samba; just type in smb://hostname in Nautilus or use “Places->Connect to Server…”
I also agree, IPP is the only way to handle printer sharing these days, though standard Ubuntu will connect to SMB printers quite easily too.
Yes, you have to use Connect To Server because browsing shares is broken Gnome. That is really lame.
I agree this sometimes happen, but did you just apt-get dist-upgrade or did you use Synaptic? I think Synaptic automatically selects a default choice for you for most packages.
Whenever I do a dist upgrade on Ubuntu I use “sudo update-manager -d” which still causes all of the prompts (i.e. doesn’t take defaults)
The best way to do an upgrade without prompts is to do it from a terminal (tty1-6) using sudo apt-get dist-upgrade. Better yet use aptitude which handles the dependencies better. I really suggest you DON’T use the graphical tools when you do these things. Using the graphical tools will give you prompts besides, even though apt-get supports it doing an upgrade that may entail upgrading x while you are in it is not a good idea. I know its not really necessary but its an old habit from the mess that was fedora upgrades.
I’m sorry but this is not true. I just updated a machine yesterday from edgy to feisty with your method and I got half a dozen prompts.
Edited 2007-03-02 19:28
The problem starts when something goes wrong – Let’s say I accidentally deleted a very important file which is used by KDE. A possible solution would be reinstalling KDE altogether to restore the lost files.
Maybe reinstalling KDE altogether is the recommended Arch way to solve such issues but wouldn’t it be simpler to only reinstall the specific package that contains the file you deleted? If you can remember the name of the file you deleted, “dpkg -S file_name” should tell you which package contains that file.
If you dislike the command line, or you want to search for filenames in a package you don’t have installed, you can try http://packages.ubuntu.com as well.
Edited 2007-03-01 21:11
re: “One of Arch’s strengths was the ease of building packages. I wondered how to do the same thing in Ubuntu. The entire thing seems a lot more complex than building an Arch package – There are 3 forms of building a package for Ubuntu, each way consists of writing several files, where the absolute minimum is 2 files (please correct me if I’m wrong here).”
Install checkinstall from the repo and do ./confgure and then checkinstall ; that will create a deb for you and install it (later you can easily uninstall that).
I fire the GUI up. Wow, that’s a mess.
I then look at the generated smb.conf file – It is loaded with garbage.
I Then try printers – In here, the GUI is so basic it is also rendered useless
Unfortunately, when the system boots up again, I am not presented with a GUI at all.
Wow… I’m tempted to install this superior-to-Vista OS. It’s definitely ready for the desktop! [/sarcasm]
<em>Wow… I’m tempted to install this superior-to-Vista OS. It’s definitely ready for the desktop! [/sarcasm]</em>
Mmm… I’m tempted to try this Vista OS you mention and see if it supports my linux filesystem, or if it allows me to install on anything but the first partition and dual-boot without destroying the MBR, and allows me to configure multiple SATA disks to be used like one big disk as you can do with LVM, please no more C: D: thing, that’s too ’80s, too DOS; we are in 2007 you know. I’m also willing to see how does it manages my memory and how easy on resources this wonderful OS must be. I’ve also heard that it even doesn’t touch the disk at all, so it really improves disks lifetime and power consumption. I’m pretty well informed about its much improved security and that it’s virus-free, yet being released only a month ago and nobody is using it. It must be a panacea of OS, and it only costs how much? 800 bucks for an usable desktop? But hey, you’ve got easily configuration for SMB shares.[/sarcasm]
Edited 2007-03-01 20:47
if it supports my linux filesystem
google ext2ifs
or if it allows me to install on anything but the first partition
sure it does.
and allows me to configure multiple SATA disks to be used like one big disk
sure it does, diskmgmt.msc -> dynamic disk
I’m also willing to see how does it manages my memory
in a very efficient manner, with technologies like SuperFetch.
800 bucks for an usable desktop
Actually $100-200, for a _really_ usable desktop, not like one described in the article. [/irony]
Edited 2007-03-01 20:59
google ext2ifs
Yeah great, ext2. Or ext3.. without journalling! I used ext2ifs for a while but I got sick of running e2fsck all the time. If it was a little more stable, maybe. I ended up going back to fat32 for crossplatform storage.
usable desktop, not like one described in the article.
Some “power user” not finding things the same as in his distro of the last 3 years doesn’t make it unusable. I mean what’d he complain about really? Samba had a lot of options? Upgrade wasn’t fire and forget?
Actually $100-200, for a _really_ usable desktop
Paying money for a usable desktop? That’d be OSX then..
I mean Windows can be easier for a new user than Linux for sure, I don’t recommend Linux for everyone. Not like I can recommend Windows either though, from what I’ve seen average Joe is quite succeptible to malware on Windows and doesn’t know what to do about it. Oh well, Linux is easy enough for me, and Windows easy enough for you 
Edited 2007-03-01 21:34
ext2ifs
This is like saying it’s acceptable for Linux to support any Windows filesystems, so long as it’s FAT32. The Linux community bent over backwards to reverse engineer NTFS with no low-level documentation or cooperation from Microsoft. The least Microsoft can do is take our clean, open source implementations of ext3 and reiserfs and implement a basic Windows port. Or do they expect the Linux community to do this, too?
diskmgmt.msc -> dynamic disk
What the hell is that? I haven’t been using Windows very much lately, but I wouldn’t even know where to begin from this explanation. Is this a registry key? Is a *.msc a kind of configuration utility? Not easy enough for a Linux user to figure out.
in a very efficient manner, with technologies like SuperFetch.
You mean in a more aggressive and wasteful manner? There’s nothing efficient about SuperFetch. Anticipatory paging is only a good idea if you have a small amount of memory, lots of disk bandwidth, and no more than (3*NCPUS)/2 threads simultaneously awake on average. This is an unusual situation for a modern desktop system. Otherwise you’re wasting precious disk bandwidth and evicting possibly useful pages from memory when it isn’t necessary, all to try (possibly unsuccessfully) to keep the current thread runnable when there’s other threads that also want to run. It’s a tradeoff that could end up improving or decreasing performance, not an efficiency improvement by any means.
diskmgmt.msc -> dynamic disk
“What the hell is that? I haven’t been using Windows very much lately, but I wouldn’t even know where to begin from this explanation. Is this a registry key? Is a *.msc a kind of configuration utility? Not easy enough for a Linux user to figure out. ”
Perhaps he didn’t explain it well, but it is not much different than the standard response to a question about Linux either.
All he is saying is open the disk management management console, by typing diskmgmt.msc, in the run box and then convert your disk drive to a dynamic disk. this allows you to set up raid, JBODs, you name it, just like LVM. An easier way to do it is just right click on my computer, choose manage, and then click on disk management. either way, it’s easy to do in windows.
Edited for sloppy typing
Edited 2007-03-02 00:33
Actually $100-200, for a _really_ usable desktop, not like one described in the article.
The article is from someone moving from one Linux distribution to another, not from someone moving from Windows. The fact is that when you use a distro, you get used to its way of doing things and its quirks.
I moved from Mandriva to Ubuntu, and at first there were things I was used to doing on Mandriva that confused me on Ubuntu – however, that doesn’t mean the desktop is not usable (never mind the fact that the article’s author is an advanced user).
Personally, Kubuntu’s printer interface (i.e. KDE’s) is very usable, and easy to find.
So unless you have something of value to contribute about either Arch Linux or Ubuntu, I suggest that you abstain from bringing Windows into it. Doing so would be off-topic (not to mention trolling).
The fact is that when you use a distro, you get used to its way of doing things and its quirks
Things mentioned are not quirks, but a serious lack of polish.
Personally, Kubuntu’s printer interface (i.e. KDE’s) is very usable, and easy to find.
Agreed, that’s why I prefer KDE for my Linux boxes, overall it seems more flexible to power user.
So unless you have something of value to contribute about either Arch Linux or Ubuntu, I suggest that you abstain from bringing Windows into it. Doing so would be off-topic (not to mention trolling).
Yeah, I got it. When Linux/MacOS users flood Windows topics with their “insightful” opinions about how superior their OS’es are — it’s ok and they got modded up quickly. But just mentioning Vista in Linux thread is offtopic, trolling and crime.
<em>Yeah, I got it. When Linux/MacOS users flood Windows topics with their “insightful” opinions about how superior their OS’es are — it’s ok and they got modded up quickly. But just mentioning Vista in Linux thread is offtopic, trolling and crime.</em>
Usually comments from Windows fanboys are not constructive and full of false information, by lack of knowledge or senseless fanboyism. While there are a lot of linux fanboys, I usually see more informed comments from Linux users than from Windows users. That should point you where the problem may be. Either on Windows or on their users.
Things mentioned are not quirks, but a serious lack of polish.
I disagree. The author approach Ubuntu expecting to find ways of doing that were similar to Arch Linux, which is why he was confused (also, going from KDE to Gnome *will* make people think interfaces have been simplified to the point of uselesness.
Agreed, that’s why I prefer KDE for my Linux boxes, overall it seems more flexible to power user.
Well, we agree on that at least.
Yeah, I got it. When Linux/MacOS users flood Windows topics with their “insightful” opinions about how superior their OS’es are — it’s ok and they got modded up quickly. But just mentioning Vista in Linux thread is offtopic, trolling and crime.
Let’s look at this from a logical point of view. Either such trolling is good for the discussion, or bad. The fact that it’s modded one way or another is irrelevant. If it’s bad in Windows threads, it’s similarly bad in Linux threads. Excuses such as “yeah, but other people do it” don’t change this.
So you’re welcome to mark Linux posts as off-topic in Windows thread (when they really are), but you also have to accept the fact that your own off-topic Windows posts in Linux threads will be modded down as well.
Look at it this way: you can either choose to elevate the debate, or contribute to the flamefest. If you choose the second, expect me and others to call you on it.
…I feel this is very correct in many aspects.
Arch was definitely much easier when thinking about the ingenious package manager, pacman. However, why am I using Ubuntu now? Well, I got tired of having to fix things that broke every big upgrade. And I wanted to have a set of things to «just work» out of the box.
I became an Gnome-user coming from Fluxbox. I really enjoyed it. However, lately Firefox and Gnome has eaten so much of my memory that Linux kills them often. Or my machine becomes unresponsive for several minutes. I also found myself hating the mouse again. So now I use Ion3 in Ubuntu.
Come to think about it, I don’t use the polished and really nice Gnome in Ubuntu any more. May be I should go back to using Arch? Using Ion3 I don’t really need to fix much if I go back to Arch. Very interesting.
Thanks for the review. It made me miss Arch.
Hello from Israel. A difficult place for a linux user.
Shalom weedow. why do you say… “A difficult place for a linux user.”?
From a fresh Edgy, all I did was right-click a folder, install samba, ready! althought I have to log out and log back in; problem with Nautilus.
I have to agree, Ubuntu’s configuration files are messy and often in strange places. This is unfortunate since things in Ubuntu does indeed break without any particular reason (like you got dropped to the shell).
Previously there was a discussion about whether linux is ready for the desktop. I think that problems with upgrading and such must be fixed, but I’m sure they will be.
At the time I answered no, because of the fonts, and no one agreed with me until they saw a screenshot the fonts of a default Fedora install. Today I still answer no and bring a screenshot from an _improved_ Xubuntu 6.06 install:
http://files.myopera.com/tsg1zzn/albums/114948/freetype.png
Feel free to flame me, but it doesn’t get any better with antialiasing. Other serif fonts are much worse.
Edited 2007-03-01 21:33
1. Do not try to get current release, but older one.
2. Do not find out you could have installed KDE-based distro from the start.
3. Bitch at length about how doing everything the hard way doesn’t work seamlessly.
I think you actually didn’t really read the review open minded.
That the update from version X to Y worked flawlessly was a plus point for me.
It’s something I’m used to in Arch, but my gf had Ubuntu, did dist-upgrade, and it broke. It’s nice to hear this is not common.
So I think this is a better review than “I put the CD into the drive and set my BIOS to boot from CD. Hard work! But then all was great! Here are the screenshots!”
On the other side, I think this review lacks some structure and should have handled some more aspects. You don’t really get a feeling about the differences of the two distros.
The Edgy release did have quite a few major changes and only a 4 month window to get them right (6.06 to 6.10), hence the name Edgy :-D. The most common problems I saw with upgrades were caused by having removed the ubuntu-minimal, ubuntu-standard, or ubuntu-desktop meta packages.
I would suggest that anyone who upgrades Ubuntu to make sure that those meta packages are installed before the upgrade so you’re sure to get all the utils needed to run your system.
Edited 2007-03-01 22:19
Full ack for that! If one wants Kubuntu 6.10, why would one install Ubuntu 6.06? Dist-upgrades are risky AND time-consuming under ANY OS (Linux distros, Windows…), even though all my Dapper->Edgy updates went fine. There was lots of really bad propaganda when MS released the Vista update versions which require an XP actually *installed* (not just the update check). Here’s the guy who does the same thing with Ubuntu by choice!
And, just as an example, back in the days when I used (the then KDE-based) Fedora, getting a Gnome desktop running the way it should involved serious configuration work, much more than just doing > debconf kdm
Most other points also have been pointed out already in other posts… Like, he could have done some research to find “checkinstall” as the tool he wants, etc. And I also don’t see how EVERY package configuration script should be able to guess all new settings which have to be set on update without user interaction. I should really try Arch if it does that. Quite some revolutionary AI seems to have gone into their package system.
And as for the question whether Ubuntu is a distribution for novices, I’d like to point out that I do not consider myself a novice (first Linux PC in 1995), and I feel most adequately served by Ubuntu.
Cheers!
I’m sorry, but I have to respond to that.
I did install Ubuntu using an older version, from which I basically upgraded to Kubuntu 6.10.
A clean upgrade is a necessity in the Linux world. One cannot reinstall his operation system from scratch every 6 months when a newer version is out.
Using dist-upgrade is an implemented feature. If it fails to work, then there is a bug waiting to be fixed.
It cannot be excused as something that should not be done. It was performed using the native tools. This is part of what package management is about.
I’d have to counter that the notion of downloading an entirely new iso and reinstalling the entire OS just to install kde is somewhat absurd.
Perhaps OT to some extent, but i wish to add some of my experiences.
With regard to the printers, i successfully shared a printer on my network with a couple of clicks… i think he maybe he is expecting an arch methodology to apply to Ubuntu… doing all the samba stuff probably isn’t necessary.
In my instance i have both windows and other Ubuntu machines/ dual-boots on the network.
In either case Ubuntu comes network friendly with the other computers showing up, regardless of whether they are windows or Ubuntu. Furthermore, installing a (HP) printer was very simple, Ubuntu comes with the HP ‘control center’ so all the advanced options of align cartridges, clean heads etc, are present… and networking the printer as i said is as simple as a couple of clicks.
I bought the printer knowing it was well supported by both Ubuntu and Hewlett Packard.
All is not perfect on Ubutnu, but you go a long way to mitigating issues with a little research into hardware support. The effort pays big dividends when using GNU/Ubuntu, with regard to _your_ ultimate usability expierience.
Arch Linux seems to do many — possibly too many — things its own way. I recently installed Arch (0.8 beta2) to see how it’s shaping up, and it kinda surprised me that Arch saw my hard drive as sda while all other distros see it as hda. I also wondered why Arch installs so much stuff under /opt. Also the init system is strange but I don’t mind that because Arch boots faster than most other distros.
Now, viewing the list of distros that are LSB compliant (http://www.linux-foundation.org/en/LSB_Distribution_Status ), I’m not really surprised that Arch isn’t mentioned here. Can any of you Arch Linux users explain why Arch developers seem to think that it’s not important to comply with standards?
This diverging from standards is the main reason why I don’t have Arch installed on my computer any more. (Another reason is that Arch has too small package repository, IMO.) Arch has many excellent qualities but, personally, I just find it too weird to really like it.
I like Frugalware better (it’s like Slackware with Arch’s pacman and a bigger package repo) but Frugalware needs to improve their installer before it can actually become popular.
(s|h)da change was due to the introduction of a new PATa infrastructure (more precisely, move of IDE to libata) in kernel 2.6.19 (IIRC). If you stick to the old infrastructure, there should be no change there. I don’t know whether this change of names is due to the upstream (i.e. kernel) change or udev/Arch tricks.
Arch uses a BSD-style initscripts, unlike most of the distros, who use SysV-style ones. They’re simple and fast.
There has been pretty much discussion about /opt in Arch – I remember some arguing that Arch actually _is_ conforming to LSB in that regard, more than other distros. Well, seems to be controversial to say the least – I, personally, couldn’t care less where KDE or GNOME sits. Since my first contact with LSB, I thought it was kinda out of place.
As for packages, I see ~3200 packages on Frugalware ftp (they might be hiding elsewhere). Currently Arch official repos host ~4000 of them and another 4000 pieces of software are available in AUR as PKGBUILDs. Thus, your point is not valid here.
Rereading your first words, I think it’s actually kind of on the contrary – it’s all those different distros offering their own GUI tools for all the possible tasks that are doing it their own way. Once you get familiar with one, you’re lost in another. Arch tries to follow the KISS principle here, which makes it more transparent – once you learn how to handle it, you will know how to manage on other distros (or even other Unix-likes) as well.
And if you mean “shape up” as “mature”, not much to see there. Those versions number are just bumped when there’s some major improvement to the installation media. Arch matures together with the OSS community, 1.0 version ain’t gonna be a breakthrough in any way.
Honestly, I don’t know why I’m advocating for Arch so much 😉 Maybe because it has never let me down.
(s|h)da change was due to the introduction of a new PATa infrastructure (more precisely, move of IDE to libata) in kernel 2.6.19 (IIRC). If you stick to the old infrastructure, there should be no change there. I don’t know whether this change of names is due to the upstream (i.e. kernel) change or udev/Arch tricks.
I have a sourcemage gnu/linux installation with the 2.6.20 kernel and it still sees my hard drive as hda.
Arch uses a BSD-style initscripts, unlike most of the distros, who use SysV-style ones. They’re simple and fast.
Agreed.
As for packages, I see ~3200 packages on Frugalware ftp (they might be hiding elsewhere). Currently Arch official repos host ~4000 of them and another 4000 pieces of software are available in AUR as PKGBUILDs. Thus, your point is not valid here.
I didn’t add the AUR repos when I tested Arch, they’re not enabled by default. Still, you might have a point there. If I choose all the official Arch repos on their web site (http://www.archlinux.org/packages/search/?repo=all&category=all&q=&… ), I get about 2750 (11 x 250) packages.
On the Frugalware web site I can’t do a similar package search but they have 2 DVD ISO images (4.1G + 1.9G) or, alternatively, 11 CD ISO images worth of packages. (http://www4.frugalware.org/pub/linux/distributions/frugalware/fruga… ).
The bottom line for me is that I can find most of the software I need from the Frugalware repos but not from the (official) Arch repos.
Rereading your first words, I think it’s actually kind of on the contrary – it’s all those different distros offering their own GUI tools for all the possible tasks that are doing it their own way. Once you get familiar with one, you’re lost in another. Arch tries to follow the KISS principle here, which makes it more transparent – once you learn how to handle it, you will know how to manage on other distros (or even other Unix-likes) as well.
No, I meant that I find myself a bit lost on the command line in Arch when compared to the other distros I’ve used — the file system hierarchy is different in some aspects (their use of /opt, for instance).
Honestly, I don’t know why I’m advocating for Arch so much 😉 Maybe because it has never let me down.
Yeah, Arch seems to be a fast, reliable and decently up-to-date distro. I also like their KISS philosophy. But I just find some aspects of Arch a bit weird — it all comes down to personal preference, I guess. If Arch would comply more closely with standards, I would like it even more than I currently do.
It appears to me that your major hangup is the use of /opt for, as the Arch Packaging Standards call them, “Large self-contained packages” (see http://wiki.archlinux.org/index.php/Arch_Packaging_Standards#Direct… ). If you’ll give the LSB a read-through, though, you’ll see that, actually, this is the preferred use of /opt (see http://en.wikipedia.org/wiki/Filesystem_Hierarchy_Standard ). Thus, I don’t feel it’s fair to accuse the Arch developers of not being “standards compliant”. Technically speaking, Debian and its derivatives violate the LSB by not using RPM, if you want to be nit-picky
Also, a straight-up and simple package count (using Pacman) gives me 8,035 packages in Arch’s repos (including the ‘community’ repo). Frugalware gave me 3,296. Even without including the ‘community’ repo, Arch still has 5,426 packages available. Hopefully this resolves any confusion over repo size (although I encourage everybody to check my numbers).
Of course, I’d be interested in which packages you can’t find in any of Arch’s repos. I’d recommend you give the AUR a look-through and, if you don’t find what you like, submit something (making packages for Arch is so ridiculously easy, I almost prefer it to binary installs… almost).
Regarding the hda->sda debacle, I believe that this is, indeed, due to an underlying change in the kernel structure (see http://www.archlinux.org/news/272/ ). Maybe SourceMage is doing some magic (no pun intended)?
And finally, I’ve found most of the devs and maintainers to be helpful, affable, flexible folks. If there’s something you don’t like, the community seems pretty open to debate and criticism (assuming you don’t come out swinging with accusations, presumptions and foregone conclusions).
Arch *does* comply with standards, arguably more closely than any other major (or minor) distro. I think you might just be confusing “standards” with “your comfort zone”, in which case, the entire OSS world isn’t complying with Microsoft’s “standards”
Give it another go. I think you’ll like it once you get used to its eccentricities (like any other distro).
It appears to me that your major hangup is the use of /opt for, as the Arch Packaging Standards call them, “Large self-contained packages”
It’s just that very few distros install Mozilla and/or the DE’s under /opt. In this sense I find the Arch solution non-standard and inconsistent.
Regarding the hda->sda debacle, I believe that this is, indeed, due to an underlying change in the kernel structure
Maybe you’re right. It seems to be a recent change in Arch and if I see a similar change happening in other distros, then I’ll admit that it’s not specific to Arch alone.
Of course, I’d be interested in which packages you can’t find in any of Arch’s repos.
There were several pieces of commonly used software that I couldn’t find from Arch repos using pacman’s search function. Unfortunately I’ve already forgotten what I searched for. Terminus-font, at least, was one of those that I missed — it’s my favourite font for terminals.
It’s just that very few distros install Mozilla and/or the DE’s under /opt. In this sense I find the Arch solution non-standard and inconsistent.
Again, as I said, just because Arch is different doesn’t mean it doesn’t comply with standards. By the definition of the LSB, large, standalone, 3rd party add-ons should be installed in /opt (take a look at the link I posted earlier, as well as any number of discussions on the FHS).
The “standard” you’re applying is as arbitrary as saying that, because Firefox doesn’t render pages like IE5, it’s not standards-compliant. Arch *is* complying with the standard, arguably better than the big boys, which seems to be the source of your gripe.
Perhaps what you meant is that it’s not “familiar”, in which case, I would agree. However, I think Judd and the Arch devs chose this for a very good reason. I and many others (i.e. the entire Arch dev team and maintainership) find this rule very helpful. Need to find a Gnome/KDE/Mozilla-related file? It’s a hell of a lot easier with the /opt layout than the /usr/each_project_has_its_own_whacked_out_structure layout.
I was with Arch in the early days, when the /opt rule wasn’t heavily enforced and /usr was a hellish backwater of symlinks and redundant directories. It was ugly and it *required* an `slocate x` to find what you wanted. We live in enlightened times.
It seems to be a recent change in Arch and if I see a similar change happening in other distros, then I’ll admit that it’s not specific to Arch alone.
It is. A simple Google search will turn up forum discussions (and complaints) about this change in Debian, Knoppix, Suse and many others.
There were several pieces of commonly used software that I couldn’t find from Arch repos using pacman’s search function. Unfortunately I’ve already forgotten what I searched for. Terminus-font, at least, was one of those that I missed — it’s my favourite font for terminals.
Enable the ‘community’ repo in /etc/pacman.conf and perform a search for ‘terminus-font’. Alternately, check the AUR online (here’s the package: http://aur.archlinux.org/packages.php?do_Details=1&ID=4809 ). Note that, in my numbers earlier, I failed to mention the thousands of packages in the AUR that aren’t even in the ‘community’ repo for lack of widespread interest. If you can’t find a package, chances are high it doesn’t exist or it’s changed its name
As always, I’d recommend checking the AUR and the main package search. It only takes a moment and prevents you from being able to claim that Arch “doesn’t have many packages”.
Again, as I said, just because Arch is different doesn’t mean it doesn’t comply with standards. By the definition of the LSB, large, standalone, 3rd party add-ons should be installed in /opt (take a look at the link I posted earlier, as well as any number of discussions on the FHS).
The FHS link you provided doesn’t say that “large, standalone, 3rd party add-ons should be installed in /opt”. Instead, it says that “Add-on application software packages” go under /opt. What is considered “add-on applications” remains a matter of interpretation. Very few distros seem to think that important applications like Mozilla and the DE’s are merely “add-ons” — and hence they don’t install those applications in /opt.
http://en.wikipedia.org/wiki/Filesystem_Hierarchy_Standard
Need to find a Gnome/KDE/Mozilla-related file? It’s a hell of a lot easier with the /opt layout than the /usr/each_project_has_its_own_whacked_out_structure layout.
I just find it easier if the application executables go all consistently under /usr/bin, and so on. It’s a matter of personal preference, I guess. I just don’t like the Arch-specific way.
As always, I’d recommend checking the AUR and the main package search. It only takes a moment and prevents you from being able to claim that Arch “doesn’t have many packages”.
Hmm… Is it possible to search these AUR archives via web browser? If I can find all my favourite apps (or their functional equivalents), I might give Arch another go (despite the weird /opt policy) when 0.8 comes out.
It’s a matter of personal preference, I guess. I just don’t like the Arch-specific way.
That’s fine, I didn’t like it either, when I first used it. Our mutual dislike doesn’t make it any less standards-compliant than any other distro, though (I believe RHEL is the only truly LSB-compliant distro, by the way).
Hmm… Is it possible to search these AUR archives via web browser?
It most certainly is! Take a gander at the main package search ( http://www.archlinux.org/packages/search/ ) and the AUR search ( http://aur.archlinux.org/packages.php ). Bask in the warm glow of package-consumption bliss.
I might give Arch another go (despite the weird /opt policy) when 0.8 comes out.
Go for it. You might even like it. Don’t bother waiting for 0.8, though. Arch has a rolling release system, so the numbers only indicate a new installer or ISO. You could install 0.7.2-base (or 0.1, for that matter) and then update all your packages to their most recent form with a single command (`pacman -Syu`).
It really is that simple
Thanks for pointing out the search engine for the packages in AUR. AUR seems to have many (but not all) of the packages I missed in the repos (current & extra) that were automatically enabled after a fresh Arch installation.
However, most of the AUR packages seem to be located in the “unsupported” repository. That sounds a bit scary. I was able to find a reference that explains what this “unsupported” repo is: “The [unsupported] repository is not really a repository. Unlike the other repositories, it does not provide binary packages. It is used to refer to the collection of PKGBUILDs in AUR which are submitted by regular users, thus the [unsupported] repository is unofficial.”
http://wiki.archlinux.org/index.php/Official_Repositories
The same ArchWiki reference also warns normal users against enabling the official “testing” repository because it “can have name collisions with any of the other official repositories” and “your system may be broken after you update with [testing] enabled”.
Just thought I should maybe share this piece of information. 😉
Just to add a few more cents to this discussion. What * I * personally feel to be the best approach is to enable the current, extra and community repos, leaving everything you want from AUR to the “manual” process, that is:
download it from the site, makepkg it, then pacman -A it if all went well.
It is really straight forward and leaves you with a very solid system. Perhaps not as solid as a Debian or other big distro, but that is only because Arch is always under some ongoing changes and testing bleeding edge technology…
Sometimes things break, but they’re just so simple to fix in arch that it doesn’t even bother me at all.
I have a sourcemage gnu/linux installation with the 2.6.20 kernel and it still sees my hard drive as hda.
I hope that you understand what would happen if Ubuntu would switch from hda to sda within 1 version release….
(hint 99% of ubuntu users with formerly hda devices would scream that kernel panics and can’t boot)
Anyway, since 2.6.19 you can eaither select old way of device naming (separate hda and sda) or new (experimental) way of naming: all pata and sata are sda now.
This has not much to do with Arch but more to do with kernel development. Simply read lklm.
A user abandoning Arch should feel quite at home in Frugalware which also uses pacman but focuses more on providing a smooth desktop experience. Lots of packages in the reps, swift and balanced trade-off between stability and cutting edge. For the little more comfortable geek in you.
It has been my experience that if you use Ubuntu, then you better hope it recognizes all your hardware because manually configuring it seems to be much harder than if you use Kubuntu. Which is probably what you should have used in the first place.
As for network printers, it is normally a 2 minute job to set up in Kubuntu, but unless you know what you are doing it can be a nightmare in Ubuntu. I finally figured it out by booting the Kubuntu live CD and setting up my network printer there. Then I just copied the info I needed and used that in Ubuntu. None of the examples in Ubuntu worked for me. I finally found this works. ipp://xxx.xxx.xxx.xxx:631/ipp
Wow, another install review,.. does he even compare more then uhmm.. 3 things to arch, and they are pretty much ‘installer’ related items.
which would make it,… ::drum roll:: another install review.. yey!!!
I was quite excited as somebody told me recently about Arch and its KISS based architecture. But since I am using Ubuntu for two years (before Ubuntu I was an excited Debian user) I doubted to change a (pretty good) running system. It is nice for me to hear that an Arch user have wrote such an positive review about Ubuntu. So Ubuntu cannot be – at least – that bad in comparison to Arch.
Recently I was quite happy as I heard in the sauna a not very experienced Windows 98 user telling his friend that he installed Ubuntu because Win98 was very slow, insecure and user-UNfriendly (driver problems).
For me it is just another proof that Ubuntu covers quite a wide user experience level bandwidth. It is suitable for unexperienced users that have even problems with Windows. But it offers also the flexibility, a power user expect from an OS.
I am using Ubuntu for development, work and on my web server and it fulfill (at least my) the requirements of each of these areas. And since Canonical offers commercial support, Ubuntu is also suitable for enterprise solutions.
Agree with the article. Sure, Arch is in a different market segment than Ubuntu but it has one other difference: It does what it says on the tin. Ubuntu just doesn’t live up to the hype.
As I already hinted in the subject: if you want to share a folder in Ubuntu go to Administration -> Shared Folders. A more easy dialog does not exist for Samba.
This is something I was wondering too about the article. Why search for synaptic and then a samba gui package before browsing the menus for a preinstalled tool. If he really ran ubuntu for two weeks, I wonder how he could overlook that.
I’m a Slackware and BSD guy who also uses Arch and a little distro called “LinuxMint”. The only thing I have to complain about if you can call it a complaint is that BSD’s rely on Linux Binaries and I wish there was support for BSD in web media content like Linux, Solaris, and Windows. Other than that… I think it’s best to say that there is something to learn from all of them… guess that’s why I have 5 computers running different operating systems… and no… no dual booting. I read all the flames about operating systems, but it depends on how you want to use it. If you find one to your liking then stick with it, You get bored with it, find another one that peaks your interest, but in the end… they all have something to be learned from.
3 Linux distros – 1 BSD – 1 Solaris -> ( No Windows ).
Edited 2007-03-02 11:45
The title seems a little misleading, I thought, that the article would be about a permanent change of distros… but eventually it occured that it’s about an archer plaing around with everyone’s favourite toy. For me this article is especially interesting because I went to Arch from Ubuntu.
Generally I agree with the author of the article, for me too Arch is simpler and more fun to play with.
Hello.
I’m actually still using Ubuntu. I wrote this review after 2 weeks of usage with Ubuntu, after everything was pretty much working to my satisfaction.
I now continue to use it and to explore it further under the hood, to have a better understanding of the distribution.
Hopefully I’d be able to write a more structured article, comparing the underlay of both distributions, when the time comes.
Edited 2007-03-02 14:44
Although almost half a million people worldwide ( my estimation ) have tried Ubuntu, many of the people who use it multiboot between Ubuntu and Windows. On the other hand, most ArchLinux users are ‘linux only’ users.
Just ask in both forums.
This is because ArchLinux is way easier to maintain. You simply have to configure everything once and then update forever.
Another thing is that incrementally upgrading things is a bit safer. This is why I prefer rolling release systems. Having to upgrade all packages each 6 month in Ubuntu scares me a bit.
One last thing is the package management. I know that debconf and dpkg are very powerful tools but creating packages in ArchLinux is still easy and powerful. It supports postinstall/upgrade/remove operations.
The only thing I wish existed in ArchLinux is a gui frontend to pacman.
There are few frontends, most notably jacman. Creating frontends should be even easier once libified pacman3 sees the daylight, which ought to happen soon.
Next one – Samba. I start searching the menus, can’t find anything.
There is a preinstalled tool with a nice and clean gui. How can he run ubuntu for 2 weeks and miss that one?
I Then try printers – In here, the GUI is so basic it is also rendered useless. Not good.
I wished he gave more info about this point. What kind of printer? Is it usually supported under linux? Directly connected or shared over network?
I installed two different printers myself under ubuntu. One was simply plug in the usb cable, click next on the automatically opening wizard and start printing.
The second one was shared by a windows machine. Not as easy as the usb printer. (No automatic detection.
) But the gnome-cups-printer gave me a choice of all connected clients in a dropdown box. Not very hard.
The entire thing seems a lot more complex than building an Arch package
Since I don’t have experience with arch I can’t compare it. However creating a package isn’t very hard. Either replace your make install step with checkinstall or use dh_make, which doesn’t require you to edit any file most of the time. (Though you should tweak some settings.)
For the fun of it i installed arch and it booted from the CD installed 2.6.16 kernel.So after i vimmed /etc/resolv.conf i thought a “pacman -Syu” would be appropiate.In addition i got a kernel panic from the kernel update to 2.6.20
I just wonder what the author means with simple
If you get a kernel panic when booting with a new kernel, you most probably haven’t followed post-install instructions spat out by pacman.
If you get a kernel panic when booting with a new kernel, you most probably haven’t followed post-install instructions spat out by pacman.
That would be it:-)
See, that’s the point. Archlinux is for users who DO care about their system beeing healthy and not just simply do “update & forget”. If any update is to break the system stability it is explicitly explained after upgrading the package with pacman. You didn’t read the message, you learned your lesson. And frankly, that’s also what makes the archlinux so simple (but I don’t mean “easy” here!). Because instead of the package’s complex post install scripts it is _you_ who determine what to change in /etc after the updating so that the system works fine after reboot. And such user interventions are really barely needed at all.
I have had no stablility problems after upgrading my system on a daily basis since I’ve installed it, and that’s more than a year now. Compare to the couple of Ubuntu poor upgrade issues they had in the past.
Edited 2007-03-05 00:20