No thanks. Life is too short for that. But here it goes anyway. Rosa: modified in Russia. I don't even want to know what shady things are going on in this distro. Korora: Fedora-based. Not for me. Chakra: KDE. Nope. Ubuntu Gnome: could be my second choice. But Ubuntu lost me, at least for the time being. Sabayon: Gentoo. Nope.
From the article: if I want to use Arch Linux without banging my head with the terminal, I should be able to do that
Look, I got burned by Arch a couple of times and I finally dropped it after 2 years after some crazy nuts updates that broke my system even following the news section.
HOWEVER, everybody does what they want, that's the point of linux and linux distros. You can't think like that, you simply can't. If a distro wants you to do crazy updates, you either do it or you go find another distro. The only thing I can say in your favor is that if somebody makes a derivative for simple folks like us, say Chakra, then it's mother distro community should not ditch on them, it's linux after all.
Knowing little about Arch Linux, I dismissed it entirely, even though a good sysadmin friend uses it exclusively on his laptop.
In my personal life, I first installed Slackware in 1995 but had converted to Debian by 1997 and I've used that almost exclusively until the last two years or so (I have one RHEL6 and one FreeBSD server). I have, of course, used RHEL and Ubuntu previously in my professional life.
About six months ago, I installed Arch Linux on a new Thinkpad W530. I didn't intend to really use it, I just wanted to see what the fuss was about. It's stuck around and I'm still running it and I haven't had the first problem with it.
I'm not gonna be converting everything over to it, obviously, but it works for me on this laptop (which I do almost 100% of my work on).
Well, I hope you're subscribed to the announce mailing list then. Because breaking changes that require manual intervention do occur, and if you don't deal with them in a timely manner you'll be SOL.
Waiting until the last minute to port all my init scripts to systemd was definitely not a good idea.
Yeah, Arch is a great distro for some specific use cases. I loved it when I was in college and had 3+ hours every other week to get my packages un-borked after some update, but nowadays I just don't have the time... Great distro for bleeding edge stuff, though!
They do occur but they're rare. I remember the move from sysvinit to systemd taking about 10 minutes, and I have a 3 year old Arch install that has never outright broken on me after an update.
Frankly I remember going through more pain every 6 months from in early Ubuntu days (although that's a lot better now also).
Filesystem snapshots should make any upgrades almost painless. If it failed, roll back to a pre-update snapshot and in the meanwhile inspect changes from a post-upgrade version (branch) of the filesystem. Maybe even run this version in a VM.
Sad things are: 1) almost no FS supports writeable snapshots and LVM snapshotting is far from being convenient; 2) you have to setup everything by hand.
Do you know of any installer-time options for linux other than btrfs? I don't know of any fs choices, but openSUSE supports btrfs out of the box and even integrates it with the package manager to allow rollbacks. It's pretty convenient and as far as I know you can simply mount an old snapshot (writeable!) and use standard Unix tools to compare them. I'm not sure if that's everything you want, though - a tool that could actually tell you what's changed using the internal snapshot information would probably be much faster and superior to a diff or rsync dry run listing. It's still closer than anything I've ever seen and the ability to simply switch which subvolume (snapshots are just subvolumes) is the default or mount it within an extant fs tree is a win for me.
I would say nothing, but they can be refined. This might get me hate, but the thing with using Windows for work and Linux for my personal projects: the GUI feels much more "refined" for Windows, even though that can be highly subjective...
Nothing. It's what needs to be removed that's the problem. I basically never use a DE because xmonad is all I need, and if it needs to be customised, it's not limited at all.
As a fan of Haskell, I've tried to move to xmonad, but realized that most of what I wanted to do to it was make it behave like ratpoison so I moved back. In particular, retaining my screen muscle-memory, and the "commands in a tree under a single initial character, which can be sent with a command" which nicely keeps everything a couple keypresses away while allowing easy logical groupings and still makes sure that key bindings never get in the way of something that I want to send to an application.
New? More? Really? I beg to differ... I thin it needs the KDE + GNOME + Englightment + etc. Guys to unite and create 1 full-featured DE, bug-free, with at set of SOLID applications (mail, browser, etc.) and 1 unified layer underneath.
But then again, we all know that's not going to happen and licenses is not the only problem here or anywhere else where you see this kind of fragmentation (android?).
First you say "I beg to differ" then in the very next sentence you suggest that a large bunch of people should get together and create a new DE. So which is it?
The GNOME and KDE communities seem to have two completely opposing philosophies.
I'd argue the best way to achieve unification would be to make GNOME 4 a Qt based DE. Qt is just superior and GTK+ only exists because of ancient Qt licensing issues that are no longer relevant. That said, this will never happen because, being C++, it'd alienate a lot of GNOME developers. During the transition you'd also set back GNOME for years.
I think there are so many words written on why the available Linux DEs suck that I don't see how is my (yet another) opinion is gonna help you.
My point is that the real problem of Linux on the Desktop _is_ the Desktop Environment, and I just don't understand why people trying to fix that problem think that assembling a new distribution _is_ the solution.
That's a lot more helpful, but I still don't quite get the conflict.
Building a new distribution gives these people the choices they want (a custom installer, better quality repositories, a decent system-wide configuration interface) - it's a way to create a unique take on Linux. A desktop environment is but one part of that - and it seems those people don't agree that it should be the most prominent one.
All these words written on why the available Linux DEs suck were so far unconvincing from my point of view and I would've loved to see another thorough explanation. Most of the things I've read aren't specific to any Linux DE and mostly "I don't want to adapt to change" arguments - just like "Windows 8 is the downfall of Microsoft Windows" and "Maverick is the worst OS X ever" (or "Firefox changed the default theme!!").
Technically I'd say that there are a number of decent DEs available, all with benefits and tradeoffs. In fact, my biggest gripe with Linux on the desktop isn't that there's no decent DE, my pet peeve is the inconsistent looks between applications written for different DEs/in different Toolkits. Mixing GTK with QT (with motif, tk, swing, e17's libraries and whatnot in the mix) looks bad, feels unnatural and restricts my selection of tools for a job ("Am I going to install that thing? Naaah, it's based on QT"). I don't want more DEs, I'd prefer more progress a la freedesktop / a unified path forward.
Obviously this is all a rather subjective thing, but Linux runs fine on my desktop for a long time already.
I'm intrigued enough by the Timeline view of ROSA that I'm downloading it now and preparing to boot it on the linuxhack machine I've got just for the purpose .. this is something I've wanted in a Desktop GUI for decades, and I'm quite interested in the implementation in ROSA - anyone used it yet, got an opinion based on experience?
For me, a Temporal interface to the work I've done on the computer is the One thing missing from modern computer use interfaces. I've always thought that the primary index key for all data created on the computer should be date/time, and better tools for querying/navigating on that index should be developed for users .. but it seems there is some work to be done on making this interface mechanic sensible to the average user. It'd sure be beneficial to me anyway - I'm often thinking "what is that file I made last week .. name forgotten, but I remember when it was, roughly .." while looking for things - if a Time-weighted interface were oriented towards usability, it would be quite useful, imho.
Thanks for the tip -I'll check it out, and now that I know its nepomuk behind the ROSA Timeline thingy, I can investigate things further without much fuss .. appreciate the tip!
One comment on OA asks why fully free distros left out; Trisquel, Parabola and I'd add gNewSense 3.0. Unless of course the OA thinks these get enough love already!
Sad that Bedrock Linux is not on that list. Also, why bother with Sabayon over Gentoo? If configuring gentoo is too time-consuming just stick to arch or ubuntu/ubuntu derivative.
I'm not sure I'd describe Bedrock Linux quite like that.
Rather, the idea is that you can install packages from just about any other distribution, including things which are typically seen as mutually exclusive. The majority of the system can be comprised of packages from something known to be stable like Debian or CentOS, while still having easy access to things like packages from Arch (including Arch's AUR), retain extremely good library compatibility with popular Ubuntu for precompiled non-distro-provided things like proprietary software, and still get to leverage the flexibility tools like Gentoo's portage provides.
So, yes, you could use binary packages unless you need to build from source - Bedrock Linux can fit that design if you want it to work that way. That's a perfectly fine way to use it. But I would frame that as one of a number of ways to go about things, as the system is quite flexible.
Everything on that list is catagorized as "works-out-of-the-box" or "works mostly out of the box". My guess is the author has no interest in distributions like Bedrock Linux or Gentoo that require initial setup to get the ball rolling, even if the resulting system has benefits over the works-out-of-the-box crowd.
One of the most important things I look for in a Linux distribution or any software project I may rely on for a long time is it's organisational structure. Is there a committee, Benevolent Dictator for Life, sponsoring company etc. It's not exciting stuff but any project wanting people to spend time and effort on it should make sure they have thought about that stuff and let people know what to expect.
It would be nice if short reviews like this made some mention of that aspect of the different distributions.
As a very long time Linux user distributions which I find intriguing are the ones using the "Nix" package manager.
Nix describes itself as a "purely functional" package manager.
As I understand it you don't need to have admin rights to install package (something I always found very, very odd since 20 years or so with both .rpm and .deb based distros: why the heck do I need to be admin to install packages that basically only one user is ever going to use?).
With Nix you can also install simultaneously several version of the same package (even for a single user if I understand it correctly).
I really wish something like that would one day "catch up" in the Linux world and see what gives.
By the way this is in no way a harsh criticism of Debian or anything: I'm a "20-years long" Linux desktop user and a "Debian die-hard" fan :)
But I find the concept behind Nix really intriguing and apparently there are regular releases.
> why the heck do I need to be admin to install packages
This is being solved in mainstream distributions by adding another layer of abstraction, instead of trying to do within "traditional" userspace.
We're rapidly moving towards a multi-VM model (be it full virtualisation or LXC) where all users "have root" inside their own containers.
Not being able to install the packages you want isn't an issue when you can quickly and easily create a VM or container and do it in there, all within your existing userspace privileges on the host system.
It's no longer about who has root on a particular machine. It's now just a distinction between who manages a system and who uses that system.
If you want to manage a system instead of just using it, you can achieve that by having your own entire system by spawning a container or VM, rather than just being limited to your home directory.
The cost in disk space is negligible in today's terms. The cost of memory in a VM is still significant, but LXC eliminates that issue.
Is there anything innovative or just fun in those?
For example, Gobolinux (now dead) tried to restructure the filesystem. NixOS (pretty obscure) tries to innovate with package management. Vyatta (dead, don't miss this one) tried to bring Cisco-like look-and-feel CLI for GNU/Linux based routing. And so on. What's fancy in those mentioned distros - I mean, except for the packages they have installed and/or pre-configured by default?
A long, long time ago when I was learning C++, I remember the Gobolinux folks were always happy to make new packages ("recipes" as they were called) for me and add them to the repo. In particular, I remember them kindly adding Code::Blocks. The people behind it were truly generous with their time, and Gobo had, in my opinion, absolutely outstanding ideas.
It's nice to see Sabayon mentioned. I've been using it at work for a few years now. I'd used Gentoo for a long time at home previously, and wanted something at work that felt like it, but didn't have time for the tedious installation process, nor the long builds if I suddenly needed to pull in a large number of packages for other purposes. These are the very issues Sabayon purported to address with an out-of-the-box setup and a binary package manager. I tried it, liked it, and have used it since.
The package management tool (Entropy) is nice. It feels a lot like Portage, but also handles a few other tasks (configuration updates, mirror selection, etc.) within the tool itself. It's a nice compromise between Portage and the binary package managers you find in most popular systems.
While there are occasionally other fun linux distros to toy with for a few days, I have never found any compelling reason to stray from Debian after it has served me on the desktop and at the server level for sixteen loyal years.
I have been an Ubuntu user since 2005 and been horrified by the Gnome 3 and Unity desktop disasters and I've been stuck with Ubuntu 11.04 because it's the last release with Gnome 2. For the past 2 years I have been looking for a new desktop environment but failed to find something that would fit my basic needs until I recently tried XUbuntu again, this time version 13.10 and I must say, I'm blown away. This is what my current desktop looks like:
No bells, no whistles but a plain a simple desktop for a developer to get stuff done. I absolutely love the latest Xfce release but more importantly, I have a stable distribution that's been around for quite some time so I'm confident I can use this for the next few years to come.
61 comments
[ 2.6 ms ] story [ 140 ms ] threadLook, I got burned by Arch a couple of times and I finally dropped it after 2 years after some crazy nuts updates that broke my system even following the news section.
HOWEVER, everybody does what they want, that's the point of linux and linux distros. You can't think like that, you simply can't. If a distro wants you to do crazy updates, you either do it or you go find another distro. The only thing I can say in your favor is that if somebody makes a derivative for simple folks like us, say Chakra, then it's mother distro community should not ditch on them, it's linux after all.
In my personal life, I first installed Slackware in 1995 but had converted to Debian by 1997 and I've used that almost exclusively until the last two years or so (I have one RHEL6 and one FreeBSD server). I have, of course, used RHEL and Ubuntu previously in my professional life.
About six months ago, I installed Arch Linux on a new Thinkpad W530. I didn't intend to really use it, I just wanted to see what the fuss was about. It's stuck around and I'm still running it and I haven't had the first problem with it.
I'm not gonna be converting everything over to it, obviously, but it works for me on this laptop (which I do almost 100% of my work on).
YMMV but IWFM.
Waiting until the last minute to port all my init scripts to systemd was definitely not a good idea.
Frankly I remember going through more pain every 6 months from in early Ubuntu days (although that's a lot better now also).
Sad things are: 1) almost no FS supports writeable snapshots and LVM snapshotting is far from being convenient; 2) you have to setup everything by hand.
and likewise when I'll care to learn Lua I'll move onto awesome-wm. Which will probably never happen.
But then again, we all know that's not going to happen and licenses is not the only problem here or anywhere else where you see this kind of fragmentation (android?).
I'd argue the best way to achieve unification would be to make GNOME 4 a Qt based DE. Qt is just superior and GTK+ only exists because of ancient Qt licensing issues that are no longer relevant. That said, this will never happen because, being C++, it'd alienate a lot of GNOME developers. During the transition you'd also set back GNOME for years.
Why do you think that? What DEs did you play with and what were the problems? Why do you feel that new distributions aren't needed/useful?
Can you expand your point and create a real argument?
My point is that the real problem of Linux on the Desktop _is_ the Desktop Environment, and I just don't understand why people trying to fix that problem think that assembling a new distribution _is_ the solution.
Building a new distribution gives these people the choices they want (a custom installer, better quality repositories, a decent system-wide configuration interface) - it's a way to create a unique take on Linux. A desktop environment is but one part of that - and it seems those people don't agree that it should be the most prominent one.
All these words written on why the available Linux DEs suck were so far unconvincing from my point of view and I would've loved to see another thorough explanation. Most of the things I've read aren't specific to any Linux DE and mostly "I don't want to adapt to change" arguments - just like "Windows 8 is the downfall of Microsoft Windows" and "Maverick is the worst OS X ever" (or "Firefox changed the default theme!!").
Technically I'd say that there are a number of decent DEs available, all with benefits and tradeoffs. In fact, my biggest gripe with Linux on the desktop isn't that there's no decent DE, my pet peeve is the inconsistent looks between applications written for different DEs/in different Toolkits. Mixing GTK with QT (with motif, tk, swing, e17's libraries and whatnot in the mix) looks bad, feels unnatural and restricts my selection of tools for a job ("Am I going to install that thing? Naaah, it's based on QT"). I don't want more DEs, I'd prefer more progress a la freedesktop / a unified path forward.
Obviously this is all a rather subjective thing, but Linux runs fine on my desktop for a long time already.
For me, a Temporal interface to the work I've done on the computer is the One thing missing from modern computer use interfaces. I've always thought that the primary index key for all data created on the computer should be date/time, and better tools for querying/navigating on that index should be developed for users .. but it seems there is some work to be done on making this interface mechanic sensible to the average user. It'd sure be beneficial to me anyway - I'm often thinking "what is that file I made last week .. name forgotten, but I remember when it was, roughly .." while looking for things - if a Time-weighted interface were oriented towards usability, it would be quite useful, imho.
[1] http://zeitgeist-project.com/experience/
Rather, the idea is that you can install packages from just about any other distribution, including things which are typically seen as mutually exclusive. The majority of the system can be comprised of packages from something known to be stable like Debian or CentOS, while still having easy access to things like packages from Arch (including Arch's AUR), retain extremely good library compatibility with popular Ubuntu for precompiled non-distro-provided things like proprietary software, and still get to leverage the flexibility tools like Gentoo's portage provides.
So, yes, you could use binary packages unless you need to build from source - Bedrock Linux can fit that design if you want it to work that way. That's a perfectly fine way to use it. But I would frame that as one of a number of ways to go about things, as the system is quite flexible.
It would be nice if short reviews like this made some mention of that aspect of the different distributions.
Edit: I guess you can think of it like "arch with an installer and configured desktop environment".
Nix describes itself as a "purely functional" package manager.
As I understand it you don't need to have admin rights to install package (something I always found very, very odd since 20 years or so with both .rpm and .deb based distros: why the heck do I need to be admin to install packages that basically only one user is ever going to use?).
With Nix you can also install simultaneously several version of the same package (even for a single user if I understand it correctly).
I really wish something like that would one day "catch up" in the Linux world and see what gives.
By the way this is in no way a harsh criticism of Debian or anything: I'm a "20-years long" Linux desktop user and a "Debian die-hard" fan :)
But I find the concept behind Nix really intriguing and apparently there are regular releases.
This is being solved in mainstream distributions by adding another layer of abstraction, instead of trying to do within "traditional" userspace.
We're rapidly moving towards a multi-VM model (be it full virtualisation or LXC) where all users "have root" inside their own containers.
Not being able to install the packages you want isn't an issue when you can quickly and easily create a VM or container and do it in there, all within your existing userspace privileges on the host system.
It's no longer about who has root on a particular machine. It's now just a distinction between who manages a system and who uses that system.
If you want to manage a system instead of just using it, you can achieve that by having your own entire system by spawning a container or VM, rather than just being limited to your home directory.
The cost in disk space is negligible in today's terms. The cost of memory in a VM is still significant, but LXC eliminates that issue.
[1]http://crunchbang.org/
For example, Gobolinux (now dead) tried to restructure the filesystem. NixOS (pretty obscure) tries to innovate with package management. Vyatta (dead, don't miss this one) tried to bring Cisco-like look-and-feel CLI for GNU/Linux based routing. And so on. What's fancy in those mentioned distros - I mean, except for the packages they have installed and/or pre-configured by default?
I do wish it took off.
The package management tool (Entropy) is nice. It feels a lot like Portage, but also handles a few other tasks (configuration updates, mirror selection, etc.) within the tool itself. It's a nice compromise between Portage and the binary package managers you find in most popular systems.
http://i.imgur.com/BK2leWF.png
No bells, no whistles but a plain a simple desktop for a developer to get stuff done. I absolutely love the latest Xfce release but more importantly, I have a stable distribution that's been around for quite some time so I'm confident I can use this for the next few years to come.