Slashdot Mirror
Linux 2.2.17 Released
Paul Weaver was among the hoardes to note that Linux 2.2.17 has been officialy released at the usual places. So take some time out from trying to compile 2.4 test releases and update the boxes that need stable kernels.
← Back to Stories (view on slashdot.org)
Just bought some of Intel's i815 motherboards with on-board ethernet. Unfortunately, 2.2.16 does not have a driver for the Intel® 82562ET 10/100 Mbit/sec Platform LAN that comes built into the motherboards (optional of course). From what I read on newsgroups support could be added with a patch or two to the 2.2.16 kernel since the chip is based on popular ethernet cards using tulip drivers, but unfortunately I don't possess enough Linux Karma to get that to work yet. Anyway, now that 2.2.17 is out hopefully RedHat will be releasing some new RPMs soon and I can get my ethernet to work, and if they don't.. well I'll just have to get down n' dirty and make RedHat and my 2.2.17 kernel play nice ;)
JOhn
Campaign for Liberty
I've been running 2.4.0-testx for quite a while, without any problems. It's easier than trying to patch 2.2.16 to cover everything!
...Corruption in the goat herd Flesh crumbles in the real world.
That's an awfully serious accusation to be levelling against the likes of Alan. Please list reliable sources, 'cuz this is the story of the year if you can prove it.
Given that Redhat funded development of the TUX webserver, which requires 2.4 for it's fundamental operation, I'm not gonna put much stock in it though...
Yes, the scaling issues have been fixed in 2.4.
:)
The scheduler has been completely reworked as has the SMP stability.
The 2.4 kernel although does not have everything in it, is the first kernel to truly address some enterprise level concerns.
It even includes RAW devices for databases
Get your PostgreSQL here: http://www.commandprompt.com/
I'd like to start this by saying that I am NOT speaking for Red Hat in this, but for myself.
/from it/ to the 2.2.x series, so that it doesn't fall too far behind.
I started working at Red Hat about 2 months ago, and I can say that since I have been here, I've been continually impressed by the extent to which everyone works to make sure everything works. No one EVER talks about steering development 'backwards', as its damn hard enough keeping it going forwards.
Now, the 2.4.0 kernel is a BIG deal, it is a big change, and it will save the world, clean your laundry, get you dates, clear up your acne, etc, etc. But it is not finnished. Period. There is more to do on it, and while work continues on it, some code is backported
But should Red Hat wait until 2.4? How about KDE 2.0, or Gnome 2.0, or Gimp 2.0, or Jargon 3.0? How long do you wait, for what? They can't, they have to make sure that the software that they put the stamp on is as uptodate as it reasonably can be, while simultaneously being as stable, and as compatible.
And how can you seriously believe that Red Hat is purposely steering development to slow the advent of the 2.4 kernel? Did you see the Slashdot article about Tux? That is a 2.4 kernel based webserver, something that is VERY cool, and it'd be great for Red Hat if it worked out of the box. But it doesn't, cause 2.4 is not finnished.
If you really have issues with the speed of the development cycle of 2.4, help the kernel developers.
-- Crutcher --
#include <disclaimer.h>
-- Crutcher --
#include <disclaimer.h>
fluxrad said:
Metrol replied:
Well, you're both right on some points but wrong on others. fluxrad is right that a SP is not the same as a kernel upgrade, and Metrol is right that a Linux kernel update doesn't update the whole OS. (FWIW, Metrol seems to have a better idea of what's going on here... not sure if sarcasm is getting in the way here. :-)
A Windows Service Pack upgrades the kernel as well as a large number of system DLLs and utilities. It's effectively a point release of the OS by another name. (eg. In a different world, NT4SP6 might've been named NT 4.6. My personal theory is that changes in version number need to go through more approval processes than a "service pack" might at a large number of big, stodgy companies.) The major architecture of the OS doesn't typically change between SP's, but the SP touches just about everything.
In contrast, a Linux kernel upgrade touches just the kernel. No system libraries, no utilities, no user-land drivers, just the kernel. It is not an OS-wise upgrade.
This bears repeating: By itself, the Linux kernel is not an OS. The Linux kernel plus a usable userland environment is.
I think the main reason 2.2.17 is out as compared to some of the other revs of Linux 2.2.x is that 2.2.17 has been a long time coming. And to think I just installed 2.2.17pre20. Anyone have diffs between 2.2.17pre20 and 2.2.17final?
--Joe--
Program Intellivision!
Well, a better "summary" is at Alan's page here.
Oh, wait a second...
I heard the 2.2 series had serious scaling problems under heavy load, but I can not remember the cause. Does anyone know if these problems have been addressed in the newer 2.4 series? This really worries me, considering 2.0 outperforms 2.2 under heavy load in many cases. Could someone a little more knowledgable about the situation reply...
Why do I see people who have clearly been reading slashdot for less than a year saying that a stable kernel release isn't slashdot material? There was a time when 10% of all the slashdot news was kernel releases.
Besides, what harm does it do you? Shut up and stop bitching at Rob 'n' crew.
--
The reason new features are being added to the stable series is largely to keep the 2.2.x kernels able to support recently improved features (new and updated hardware drivers, tuned algorithms, etc). If it weren't for this, then we'd be stuck with a kernel that required massive patching to support any newer hardware or improved features. Additionally, since the changes are relatively not very intrusive, and are pretty thoroughly tested in the -dev tree, the people running production boxes on 2.2.x kernels can be confident that the kernel is still solid, even with the new features.
;)
The reason 2.x.x *appears* to be incrementing more rapidly than 2.4.x in some areas is because all the cool new stuff (RAID code, USB, PCMCIA, etc ad infinitum...) has been in development (2.3.x) for a long time, and has been pretty thoroughly wrung out. In other words, all the initial "hard work" has largely been done. It's finally been deemed stable in the 2.4 series, and is now being backported for the benefit of those less adventurous souls (like me) who prefer a well-used, known-good kernel (2.2.x) on production systems, but want the benefit of some new features. The same thing happened with backports of 2.2 features into the 2.0.37+ kernels, even after 2.2 was released. Instead of a massive overhaul, with an entirely new kernel architecture to deal with, you get the shiny new stuff (new hardware support, bugfixes, and tuneups) and still get to stay with the tried-and-true kernel you're used to using.
That's likely why RedHat is developing their next distro to 2.2.x, *not* 2.4. After all, 2.4 is still considered a moving target at this point, and is extremely difficult to develop to (Alan Cox mentioned this in a recent diary entry). If you've got an entire distro to worry about, best to keep it with the currently stable and well-known 2.2 kernel, instead of a constantly-changing 2.4-test kernel.
Besides, if you really want to be cutting edge, go get slackware or rabbid squirrel and build your own distro with whatever software versions you want. This is Free software, after all -- if you've got an itch, scratch it!
Does anyone know if this release has full support for ATA/100? I really want to put linux on my shiny new 800MHz Tbird, however, I really don't want to waste my time downloading over my 28.8 modem only to find no support for my computer.
Seriously, Win2k and Win98 do not support ATA/100 without 3rd party drivers. It would be nice if Linux could get ahead of the game with this.
A slashdotter who didn't build his own computer is like a Jedi who didn't build his own lightsaber.
---
I say Rob should pull out a week's worth of Classic Slashdot, from a couple years ago, and post it somewhere. We could then compare the content of THEN with the content of NOW.
/. for a few years now, and I prefer to read a new kernel announcement, where I can pickup tidbits like "it screwed my ext2" or "won't boot on my 8-way Celery bawx" than read about how Bill Gates just hired Larry Ellison to clean his indoor pool!
I've been reading
Absolutely. Many uf us "geeks" run NT on some systems (not really by choice, but hey). It would be nice if Slashdot reported on NT Server / Win2000 service pack releases. Microsoft's website is soooo difficult to navigate that having that info here on slashdot would be a big asset!
---