Slashdot Mirror
Linux Kernel 2.6.0-test6 Released
lamont116 writes "The latest in the series of beta kernels was just released by the fine folks who have given us Linux. Enjoy!" The Changelog has a hefty 240K of miscellaneous changes... LWN has an overview of the updates.
Well, since most likely some varation of the 2.6 kernel will be out when Longhorn is released, this puts Linux in a good position with 2.6's better memory management and IDE support and whatnot...
- Linux Kernel 2.6.0-test6 (full) [32MB]
- Linux Kernel 2.6.0-test6 (patch only) [1MB]
It still seems wrong to improve performance through a Slashdotting, but the more the merrier!UNIX? They're not even circumcised! Savages!
I expect Arjan to have updated Red Hat packages soon at http://people.redhat.com/arjanv/2.5/
p age=kernel for the information for making these kernels work on RHL.
Remember though that some things have changed between 2.4 and 2.6 that can't just be worked around by installing new packages. (USB module names, some mount points, that kind of thing.) If you want a clean boot you will have to change some of the init scripts, and this will break booting 2.4. So it's a bit all or nothing at the moment, and I recommend people who aren't convinced it will do everything they need it to do (I couldn't get my network card working under 2.6) stay on 2.4 until it's released proper.
Check out http://thomer.com/linux/migrate-to-2.6.html and http://www.fearthecow.net/index.pl?section=guest&
Actually, while I can hardly wait for 2.6.x to stabalize for many reasons, desktop use is not one of them.
2.6.x will being some very real improvements, but in the big picture of where gnu/linux is still lacking on the desktop, linux (the kernel) isn't really one of the problems.
-- MartinG To mail me: echo kewyjlcxyzvjfxbqwh | tr bcefhjklqvwxyz
It does make Linux more responsive on the desktop too, although many of the performance improvements will be most noticeable to those running clusters.
i.e. the unnecessary blank spaces and newline charaters.
New year Resolution: Don't change sig this year
No, you don't need to be using Gentoo.
Since you tried out a development kernel you seem to be a developer or at least part of quality assurance. Therefor you are not allowed to ignore bugs like that and go back to the working version. Instead you have to find out the source of the problem, write a nice bug report and file it.
If I had the time at the moment to find the source of the problem, I would... However, that's not an option today. I have, however, sent along a nice bug report describing the problem and my system configuration.
Dinivin
the newer -mm patches (for -test4 and -5) are supposed to fix this. These are the Andrew Morton kernels, you can find them under /pub/linux/kernel/people/akpm IIRC. They also claim to fix the "make RPM" target (haven't tried that one myself yet tho).
/dev/hda losing its interrupt. I switched off ACPI and all power management since I default to 24x7 SMP use.
Might want to have a look at those, since he's the next "stable" maintainer.
Also (In reply to a previous post) I had similar probs with
Yes, some rc script hacking is required.
Overall: this stuff blows my mind, I thought just SMP was fast when it came out, but this is in another league. FWIW, my base system is RH9 with updates from RH.
C|N>K
I've just switched from -test5 to -test6, and the difference is remarkable.
Xmms doesn't skip, Mozilla doesn't snag, even during a kernel compile and SETI@Home running in the background.
Looks like Con Kolivas's interactivity patches are definitely something.
Never ascribe to malice that which is adequately explained by incompetence.
Two weeks ago I paid SCO compliance fee for the 2.4 kernel. If I download and use the new edition, would that entail another $699, would it be free, or is there some upgrade fee somewhere in the middle?
Until recently this would all have been fine but now I have my new Sharp Zaurus SL-C760 I am actually concerned about other architectures appart from x86 ;-) At the moment for just generic ARM support in 2.4 you need a large patch from Russell King and then more patches for the Zaurus specific stuff. The last rmk patch was a month and a half ago for 2.6.0-test2 and as far as I know, no one is even working on porting the Zaurus
specific stuff to 2.6.
I don't really know what the arguements are for the present development model where most of the non-x86 architectures are kept separate from the mainline development but I really don't see how it can be a good idea. I guess I don't see what the difference is between individual subsystems, for instance, and support for different architectures. In both cases individuals or teams work on their own but in the subsystem case everything gets merged back in, by the time the kernel it declared stable, whereas for non-x86 architectures this never happens.
It seems to me that given the large size of these architecture patches, their maintainers must spend most of their time just updating them to keep them in sync with the new kernel versions, rather than actually fixing bug or adding new features. Also the fact that ARM users cannot test the latest kernels because there are no rmk patches for them can only lead to a "negative feedback" situation which will hurt kernel development. In general anything that unnecessarily fragments kernel development cannot be a positive thing.
That way we can get a whole lot more testing done with very little disruption on nornal system?
I tried to make one but I couldn't get it to boot, so maybe someone better qualified can try and if successful post Torrent file.
Help fight continental drift.
>> They've never even announced a date for Longhorn. What on earth are you talking about?
I quote from one of their Press Pass documents they have online at microsoft.com:
"Over the course of 2004 you'll see a couple of releases in the betas for "Longhorn" and we'll see that coming to market in 2005.
Now, I'm sure that many of you have heard about or wonder about the possibility of whether we're going to do something before "Longhorn," is there an interim release, and that's something that I don't expect us to do. Currently we have some additional releases that are coming out as follow-ons to the XP Media Center Edition and the Tablet PC Edition so we've got some great advances and fit and finish and addressing additional international marketplaces with new handwriting recognition, new guide data for Europe for the Media Center and so on.
So you'll see some good incremental moves there but really the weight of the company, the weight of all the people in the Windows client division and across the platform's division, the weight of that effort that we're doing is around "Longhorn" and that's what we're focused on and we hope to get you all really pulling the same way so we can come out with a huge wave of excitement for the industry when "Longhorn" ships in 2005."
(quotes and italics mine.)
Several online sources have credited varying target dates for Longhorn, but all generally agree that 2005 will be the earliest that it will be available to the mass market:
Longhorn Betas in 2004, GA in 2005 (ENT News)
Analyst Pegs Longhorn Release at 2006 (ENT News)
Microsoft Pushes Back 'Longhorn' Release (Open Tech Support)
A Longhorn Delay? Not Quite (WinInfo)
Microsoft announces Longhorn release date (CNet News.com)
(I have no affiliation with these sources.)
Microsoft themselves show that 2005 is the target date of Longhorn in a slide picture. (png image)
>> As if that matters. Those are early, early alpha shots, as we all know. All the cool builds with the 3D acceleration are in a different Microsoft lab anyway that hasn't had one of their builds leaked.
First, can you prove there are "cool builds" that will have 100% of the suggested features in a shipping version of Longhorn? Secondly, the history of Microsoft's software release stategy has been plainly made clear numerous times. Hype, Fear, Uncertainty, Doubt, Buzzwords-o'-the-day, Gross Appropriation (oh sorry, I mean Innovation), and just downright unfullfilled expectations litter the computer timescape.
For a company that has been in the software business of well over 25 years, one would think that with their talk of innovation and "exciting new time-saving" features, we would all have an operating system on our machines that used voice controlled openGL hyper speed interfaces, smart enough to do your work for you while you browsed sites like slashdot. Isn't the entire point of using a computer to save time and be more productive? This sadly is not the case today .
>> I'm not really sure what breathing room you're talking about. The developer preview of Longhorn is coming out later this year. The list of features Longhorn already boasts is staggering, and I doubt within two years that ANY Linux projects will come close. We'll still be stucking using X11 with a hacked on desktop simulator, business as usual.
user@host$ diff