2.6.13 Linux Kernel Released 464
LynuxFre@k writes "Linux Torvalds announced the release of the 2.6.13 Linux kernel. He noted that there was a major change to the x86 PCI code, and that while all bugs from the change were believed to be found during the release candidate phase, it's possible that some devices may have problems. From this release on, it is intended that major changes only be merged into the kernel within two weeks after a major release. The rest of the time will be spent fixing bugs, with the goal of both increasing overall stability and decreasing the amount of time between major releases. Download the latest Linux kernel from a kernel.org mirror."
New release strategy (Score:3, Interesting)
Re:How about a stable ABI? (Score:5, Interesting)
A) It's been years since I've recompiled a kernel, and I've only compiled a few software packages in years. I use Linux daily at work, and exclusively at home. It may not be as easy to install software as on a Mac, but a good distro is equal to Windows.
B) I agree, but at the same time I find it rare that I have to drop to a command line to do normal computing tasks. I still go there daily, but by choice.
C) I can usually find anything I need online without having to post to a message board myself. However, I do agree that it needs significant improvement. I wouldn't expect non-technical people to search online for their answers.
By the way, you should find other examples to "prove" your technical skill. Ripping videos and using Photoshop aren't too "technical" in nature, especially here. Alternatively, don't try to prove it, just leave it assumed. Note: I'm not calling your knowledge into question, just your examples.
More kernel crashes as of late? (Score:5, Interesting)
Did ATAPI/AHCP on SATA become true on default? (Score:3, Interesting)
As of now no SATA DVD drive works well unless you change one line and recompile the kernel.
So many systems are now built as SATA-only (yes, the IDE ports are completely unused), stock kernels break all live-CD distributions - none of them will boot
Re:kernel bug fixes (Score:5, Interesting)
I care about the results, and so far the 2.6 tree has produced a grand total of one kernel that actually works for me (2.6.11). And the obvious cause, rightly or wrongly, seems to be Linus messing around with the development process.
Or how would you like them to do it? "We will do things this way, and by god, we will do it like this untill the end of time! Even if better ways of doing this comes along, we will not change our ways!"
How about "We will change things only when the alternative has been shown to be unambiguously better on a smaller project, and only when changing major versions". I believe in experimentation but the kernel is such an important project that a bit more conservatism is called for.
Re:Why is Linux so great? Please share your reason (Score:3, Interesting)
The problem is in how they're delivered and the lack of 1) a common packaging format and 2) lack of a common "package namespace" (ie: xorg can be called xorg in fedora and xserver-xorg in debian, that makes dependencies fail and can be only fixed by using a common packaging framework where developers and not distros makers package things)
But Linux continues being much better than windows in some areas. For example, you've to download the
Tannenbaum's Revenge? (Score:3, Interesting)
I also hope it's not going to get modded down to the seventh level of hell, as I'm about to (gasp!) express disagreement with Linus.
First of all, I am vaguely concerned about the Linux kernel development. It's been a long time since there's been major improvements under the hood. I've had Linux desktops freeze on me. In the past, that never happened. Ever. I don't know which kernels are trustworthy anymore. I've read something to the extent that stabilizing kernels is now considered the Linux commercial vendors' job. Excuse me, but WTF?
In the meanwhile, while we Linux types wave our dicks around and gloat over how great we are, the guys at Redmond are happily making it possible to change video drivers in their OS on the fly, and to unload a crashed driver without taking down the system. Will it work? Probably not 100% well right away, but trust me, they WILL make it work or they'll die trying. And Windows 2000 is proof that they can certainly do things well when they put their minds to it.
And Linux is about to become the unstable OS choice and it seriously pisses me off.
A very long time ago, Linus Torvalds and Andrew Tanenbaum had a since famous argument about the core structure of the kernel.
Linus's argument was, if my memory serves, that it all boils down to pushing bits around, and that you should as well push the bits in the simplest way.
And this is where I disagree.
Kernel development is about pushing around the bits that will push bits around. Those are the bits you want to push around in the simplest way. The goal is not simply to produce a good kernel, it's to produce a maintainable, efficiently improvable set of source that will compile into a good kernel. Otherwise, the end product you get is a good kernel for its time that will be a bitch to drag into the future.
Perhaps the state of the Linux kernel development today is but Tanenbaum's schadenfreude.
Assuming that kernel improvements have indeed, as is my admittedly fragmentary view, slowed down worryingly, I find myself wondering if, simply, now is when Tanenbaum should be speaking up, rather than all those years ago. The structural needs then were simple: few consumer devices, reduced architechtural diversity. Today's aren't. And there is STILL no 'just-works' way for third parties to deliver drivers to their customers. The least worse they can do is deliver sources to the kernel maintainers and hope that 1) they will be accepted, and 2) there won't be too many months between now and the moment their customer's OS uses that kernel. Or they can ship scripts that will compile glue code between their driver and the currently running kernel, and hope that the customer has a freaking compiler installed. I'm sorry, I can compile drivers and upgrade kernels manually, but neither are acceptable solutions for the mass market.
In fact, I'll go out on a limb and predict that unless the kernel's structure and development processes are rethought to take into account the use of an OS as a three-party system -- the OS vendor, the user, and the commodity/paraphernalia providers -- Linux will never be a significant player in the desktop market.
Thought on that? Please, please, please prove me wrong. I'm a long time Linux user, I did in my time the mandatory contributions to the kernel that allows me to speak up and bitch now, and from what I can see the future is not looking well for the Linux kernel. So please prove me wrong. Thanks.
Re:2.6 a year and a half old but... (Score:3, Interesting)
What is this nonsense ? You don't want Linux to support hardware ? Linux can not be usable by just supporting standard hardware you know ?
A large number of organizations (as well as Debian Stable and Redhat) still use 2.4. It's pretty pathetic
Isn't that a good thing ?!!! You sound like a troll. What is pathetic with that ?!! And FYI, RH does not use 2.4 as default anymore, and their 2.4 kernel included backports from 2.6. Debian Stable supports both 2.4 and 2.6, you're several weeks late.
Maybe instead of spending time switching policies, kernel developers should be consulting with end-users to find out why we're not using 2.6
What for ? Are you stupid ? Clearly, you are not representative of Linux users : we use what is provided by our distro, and they supported Linux 2.6 since a long time ago. Most distros do actually support Linux 2.6, and FYI there was a race about which distro would support it first when the kernel was out. I think you can not understand that because you think a distro can't support 2.4 and 2.6 at the same time. This thinking must come from the people that want you to believe that apps running on 2.4 are incompatible with apps running on 2.6. I tell you : these people are MS shills and they are wrong.
Aside from security patches, any effort on 2.4 development/maintenance needs to stop. It's a brain drain, and active maintenance is encouraging people to be lazy in upgrading.
You have to let go of this Windows mentality that you are forced to upgrade even when it works. Linux 2.4 still works pretty well, it was still running circles around every version of Windows for example. To this day, MS still have to take very old RH releases with Linux 2.4 (without 2.6 backports) to compare to its latest Win2003, because of that.
So why should people change when it's not broken ?
Right now 2.6 is a lame-duck kernel, and if they keep trudging on and release the next stable without looking at why 2.6 isn't the defacto kernel of choice today, Linux will be rather fubar'd.
Please stop making a fool of yourself. Even the more specialised distros like Geexbox use the latest Linux 2.6 refinements. You are 3 years late at least ! Like most MS shills. I don't imply that you are one BTW, but you are pretty clueless like them.
Re:Sorry (Score:1, Interesting)
Re:More kernel crashes as of late? (Score:2, Interesting)
Linus needs to branch the tree and get the hell OUT of 2.6 and give us our stability back. Seriously, I don't know why everyone is so scared shitless about incramenting a (minor) version number every year or two. Kernel stability needs to be an absolute priority. We already have enough issues with X, and Window managers and other software crashes in the Linux world, but with no solid foundation to build on, we have a situation not that far from Windows 9x.
Re:How to tell if you are a linux fanatic. (Score:3, Interesting)
Does Microsoft make the keyboards themselves (I doubt it) or do they just outsource the production of them or license their trademark to the company which produces it?
An Unstable Linux (Score:2, Interesting)
The issue is that it is not Linux that is unstable, but X (XOrg or XFree). I recently installed a multitude of distributions recently*. X crashed on all of them, many times frequently. All occurred within the first few days of installing. Bear in mind that not everyone has another computer hooked up to the Internet where one can SSH into the machine and kill X. For some, the crash of X is the crash of the computer. CTRL + ALT + BACKSPACE doesn't always work. This is my biggest issue with Linux or UNIX variants.
I do propose a solution: a patch or replacement to X in where it does not run as root at all (to the uninformed, running X as a user still has parts run as root). If this is not possible, then revise it as such:
1) Include only the minimal, absolute necessary code required to run as root
2) A small, and as a result less complex, code will make it easier to reduce bugs and increase stability
3) Make this root code standard across platforms (Linux and other UNIX variants) so no modifications which add to code size are required, again reducing code and enlarging the audience that can review the code
Strip the code that runs as root the *barest* essentials and let all functionality run as a user. Long story short, whatever can't run as root can't crash your computer. Therefore, eliminate or make it as small as possible (significantly less than what X runs as root today).
Are there working projects available that I am not aware of? Recently, I have heard that OpenBSD has something akin to what I am talking about. Is this accurate? What of GNU Hurd? If I remember correctly it implements some of this (at least to my limited micro kernel understanding); however, is it even usable yet?
* I was let down by the new Debian Stable (stock install 3.1ra) (1 of 5 distributions I evaluated). It's wonderfully easy now and set up everything out of the box (mp3 and video support included which many users have been clamoring for from other distributions), but X crashes very frequently when switching to a VC and randomly crashes a lot in general.
Re:More kernel crashes as of late? (Score:4, Interesting)
Have you tried reporting these crashes? I can't find anything about ifconfig triggering crashes. They can't test everything themselves, because they don't have every hardware configuration, so it's important for people who do to tell them when something is wrong.