Forgot your password?
typodupeerror
This discussion has been archived. No new comments can be posted.

Kernel 2.4.1 Released

Comments Filter:
  • No, he's describing a problem (wayward DLL updates by applications) that has been largely solved in Windows 2000. Apps aren't allowed to arbitrarily update system libraries. A whole library management system has been put in to handle apps that try to do so.
  • Actually, I like it for another reason.

    No more having to listen to Hans Reiser take temper tantrums on the kernel mailing lists. I've never read more juvenile, whining, self-centered, arrogant, and childish posts in all my life. Rather than admit that maybe it wasn't appropriate for his code to be in the development kernel at a time when they were trying to freeze it, he whined, complained, and insulted everyone. Unable to acknowledge the technical reasons, he started accusing the core developers of favoritism, eliteness, and eventually just decided that there was one big conspiracy against him.

    Although it saddens me that his code made it into the kernel, because of his behaviour, at least I won't have to read any more of his bitching.

    It definitely gives me some insight as to why some people won't use GNU or GNU/Linux because RMS is the "spokesperson."


    --
  • actually the ftp.*.kernel.org is not updated ...
  • by orabidoo (9806) on Tuesday January 30, 2001 @07:20AM (#470776) Homepage
    This is either a troll or a case of bad misinformation. Odd version numbers are development (unstable), but that's the middle number, not the final one. so 2.4.1 is stable, because 4 is even. the 2.5 series will open in a few months time, and will eventually lead up to 2.6 or 3.0 or whatever the next stable branch is called.
  • ...that's to keep brain-dead people from using debian :)

    --

  • by tsa (15680)
    It's ten times bigger than the kernel I started with -- 1.2.13!
  • The only differance is Linux isn't attaining global domination with massive marketing tie in's, illegal monopoly leveraging, and an airtight legal disclaimer wrapped around a shoddy rip-off implementation of other people's creations.

    Other Msft bashes include last Sunday's Dave Barry [herald.com] column about his BSOD'ing pc, and this morning ABCNews has this blurb about "If Msft can be hacked is anybody safe". Hahahaha, silly mass media. That's like asking, if McDonalds food isn't so good, can ANY restaurant make a good meal? I mean, they're the largest restaurant chain in existance! They MUST have the best food. Duuuuh. Public opinion will turn eventually, and no $200,000,000 advertising campaign can keep their grand self delusion going forever.
  • The how come I'd doing Video recording Linux?

  • 2.4.1 is a formal stable release, not development.
    The middle number of the 3 is even for stable releases, odd for development releases. The last digit is used to indicate (small) incremental upgrades and patches.

    New development features will go in the 2.5.x (or 3.1.x if we really want change) tree.
  • how about iptables?
  • The kernel docs explicitly state that if you're using PGCC, you're nuts. There's a reason those changes haven't been rolled into gcc yet. Incidentally, there's no such thing as egcs anymore. And finally, there's no sense in reporting bugs in the CVS tree without a corresponding fix or a very very well-narrowed problem definition--it's still in devel, and "it doesn't work for x" is not a big help in getting things fixed.
  • I've been digging for VIA IDE info too since I upgraded my mobo/cpu yesterday morning. I got a MSI K7T pro 2A (ata 100). Defaulting to DMA at boot up with 2.4.0 made my HD go BOOM. Check out signal ground for the most recent VIA IDE drivers:

    http://www.signalground.com/article/3157986091

    They worked like a charm for me.
  • Now I'm waiting for it to be put in distros. I guess I find it annoying how some distros (Mandrake, RedHat) go crazy if you change their kernel. However, they are better for getting a desktop to JUST WORK out of the box.

    Heh. I'm currently running kernel 2.2.17 on this box that started it's life as RH 5.2. Sure, I've upgraded stuff via RPM and source tar.gz:s quite a lot during the last almost two years, but it works. ;) I haven't even used the distribution upgrade options available via the newer distributions.

  • The only problem I had getting 2.4.0 to work with my RH 7 workstation was with USB...I had to tweak one line in the /etc/rc.d/rc.sysinit file, and that was because 2.4 was actually displaying my USB character devices in /proc/devices, whereas 2.2.14 did not.

    The one thing that will save you many headaches when upgrading your Red Hat kernel is to modularize as much as possible. This is the way that Red Hat's system initialization scripts like it, and it's probably good practice anyway.
  • Huh? Can you point us at where the "linux community as a whole" critize MS for putting out regular patches and upgrades?

    If anything, what I've seen most of is that they are critizised for being too slow to issue patches and upgrades, and also for releasing products that are claimed to be stable even when theres tons of outstanding bugs.

    And 2.3 got up a lot higher than 12 or 13, but 2.3 is a development kernel - it's was not intended for normal use - all Linux releases 2.x.y with odd "x" are development kernels.

    Even though I despise Windows, I'd be happy if MS did follow Linux' model and released patches more often, and didn't try to take their customers money for buggy crap while claiming it to be a "stable" version.

    It's getting quite annoying with trolls like you who try to present the "hypocrisy" of the linux community, while presenting opinions that are by no means prevalent as if they were the opinions of the whole community.

  • So you feel that the world should be deprived of potentially very useful code just because you don't like its author's personality?

    It's not like the world is being deprived of "potentially useful code." People have been using ReiserFS for ages now, and most distributions come with either a alternative kernel compiled with ReiserFS, or ReiserFS patches to the mainline kernel already installed.

    Second, it's easy to obtain said patches. It's not like there's one guy, in the middle of the Sahara, that you need to send an SASE to in order to get the source code carefully inscribed on dried clumps of sand.

    I'm not defending Mr Reiser's behavior, but don't you think that the code should speak for itself? I have no idea what he is like as a person and I don't really care as long as he writes good code.

    Don't you think Mr. Reiser should let the quality and/or utility of the code speak for itself, rather than putting all sorts of pressure on the kernel developers to include his code NOW NOW NOW (or, if possible, yesterday)? For a long time, the kernel developers didn't put his code into the kernel because they had different plans. They wanted to make the foundations better for all journalling filesystems. Yet, rather than acknowledge that there might be a technical reason for the lack of inclusion of his code, Hans took a fit and accused anyone and everyone of being part of a conspiracy against him.

    That doesn't sound like someone who lets code quality stand on its own.

    I'd be perfectly willing to let his code stand on his own - as long as HE would let his code stand on its own. The second Hans Reiser made the character of the other kernel developers an issue, his own character became an issue as well.

    --
  • by Elvii (428) <[ten.tsacmoc] [ta] [5791divad]> on Tuesday January 30, 2001 @07:27AM (#470792) Homepage
    If you're like me and are just now starting to use reiserfs, they you might not have the tools to make, check, etc, the filesystem, and they *don't* come in the kernel patch. But seeing as it recommends that you use utils that match your version of reiserfs (3.6.25 in 2.4.1) then here [namesys.com] are the recommended utils from namesys for 2.4.1 kernel.

    bash: ispell: command not found
  • Now I'm waiting for it to be put in distros.

    SuSE Linux 7.1 [suse.com] ships on Feb. 12, with kernel 2.4 available during setup.

    --

  • Here's the change log:

    final:
    - Al Viro: core file hardlink attack avoidance fix

    pre12:
    - Get non-cpuid Cyrix probing right (it's not a NexGen)
    - Jens Axboe: cdrom tray status and queing cleanups
    (etc., etc.)

    Does this mean that there is only one change in 2.4.1? Or are all those changes (listed under pre12 through pre8) included in the final?

    Some of the changes under pre# would affect me (APCI, AGP, r128, eepro100 updates), but the one change under final would not (as far as I can tell). I'll probably still install it, but can anyone read these change files? Are these all changes to 2.4.1, or are these just the last few change cycles?

  • I noticed that 2.4.0 didn't have ip_masq_quake and all those other ip_masq_* modules. Does this mean that 2.4.x doesn't need them any more, or that they haven't been written yet?
  • 2.2.18 rocks? Super! Stick with it! There's no penalty for those who don't upgrade.

    I had to go to 2.4 because I need iptables. Aside from my packet filter machine, I'll probably stick with 2.2.18, just because I've got it finely tweaked and everything works perfectly.

    So here's to no "upgrade now!" thumbscrews!
  • Excuses...Excuses... Everybody's got one.

    I thought about leaving that statement out of my post, but the fun of tweaking all the Linux trolls got the best of me. :)
  • Well, stick a BIG "it depends" on what you say above, because I think you're missing the point. In Linux, device drivers are a part of the kernel. That's why, when you compile a device driver, you need to compile it as part of a kernel, even if you intend (and it's not easy if you do this) to insmod it into some other environment.

    But regardless, the kernel's device drivers are a part of the kernel. When a new version of the kernel is released, it contains, built in, source included, the latest versions of those device drivers. When device driver fixes are done, they are released as part of kernel releases. The central point, that Linux is updated once every month or two, or sometimes several times a month, because of the desire to release updated device drivers, is generally true.

    As for the remaining comments, I'm trying to figure out what you're implying or believe I've said. They look like Linux advocacy, which is fine, but it's not exactly relevent. It remains the case that when Microsoft release updated device drivers, they'll release them with no apparent kernel version change. When Linux releases them, they will. As such, the absurd comments I was replying to suggesting that frequent releases of Linux are a bad thing and contrary to Microsoft's own practices are false. In Linux, version numbers include updates to device drivers. In Windows, version numbers are seperate to device drivers.
    --

  • Ok, having read all the responses to this, I'm at a loss. Could someone please point out:
    • Which part of the above is "Funny"?
    • Which part advocates or opposes microkernel or monolythic kernel design?
    I was making a simple point. Updated device drivers will result in new kernel releases in a monolythic environment, by definition, and wont in a microkernel/otherwise seperate device space kernel design. There is nothing either good or bad about either implied, unless someone really agrees with the notion that version numbers Must Not Be Changed.

    Help me out here, it can't be that difficult to understand, can it?
    --

  • by Tridus (79566) on Tuesday January 30, 2001 @06:44AM (#470827) Homepage
    Yes, all of the changes are included in the file. the pre# just shows when it was changed. Its also useful to know what changed when if you are using the pre versions.
  • by TrentC (11023) on Tuesday January 30, 2001 @10:20AM (#470828) Homepage
    ...you missed this section in the kernel notes:

    2.4.1:
    As of 2.4.1, the kernel now conforms to the Open Naming Specification. According to the specification, any or all of the following pronunciations are considered "correct"; feel free to add your own!
    • LEE-nooks (early Linus' pronunciation)
    • LYE-nicks (early Americanized pronunication)
    • LIH-nucks (later American pronunciation)
    • RED-hat LIH-nucks (pronunication commonly used by newer Linux users)
    • DEBB-ee-uhn guh-NOO LIH-nucks (pronunication used by GPL zea^H^H^H^H^H^H^Hadvanced Linux users)
    • WINN-dohs TOO THOW-zand FIVE (pronunication used by Microsoft after the proper "behavior adjustment" and re-education)

    I hope this helps.

    Jay (=
  • by Mads-Martin (82002) on Tuesday January 30, 2001 @06:44AM (#470829) Homepage
    When are these Gnome guys gonna get their act together and make their Gnome counterpart called gernel? We can't have KDE is the only one with an ernel implementation.
  • The pre changes listed are 2.4.1preX changes. Everything listed on that page are things that changed from 2.4.0 to 2.4.1, so no, you'll get more than one change.
  • Please send it to the mailing list, with a detailed report, and a log from your /var/log/messages with the part that went wrong.

    Someone will look at it over there. Mailing to Linus won't help you since he got nothing to do with AMI Megatrend..
  • Oh, I (and lots of other people) wish there was such a converter. I emailed PowerQuest about adding ReiserFS support for Partition Magic - but so far, I didn't get any answer from them
  • (today's obligatory obscure anime reference:)
    Nah - reverse the g and the n: Nergel - close enough to Nergal for me =)
  • Umm, I can't figure out if you're serious or trolling or ignorant.

    Linux drivers can, and are updaed independently of the kernel all the time. In fact, you can unload an old driver and load in a new driver all without rebooting. Your users might not even notice. No, it's not really a 'microkernel', but it shares a lot of a microkernel's features in that regard.

    Windows requires a reboot every time you loaded a driver, last time I checked. Also, drivers conflicts and incompatibilities are rampant. One of the whole points of Windows 2000 was for Microsoft to start making sure that the drivers vendors released were lots less buggy so their system wouldn't BSOD as often as it used to.

    Near as I can tell, while Windows NT/2000's might be based on a microkernel, it has no effect on the system as a whole aside from the additional technical marketing buzzwords people can use to describ it.

  • you know according to linus' our words a large patch like ReiserFS shouldn't have made it in [linuxcare.com]. He didn't want to add and large fixes, or features. just small hacky fixes.

    Another thing is that apparently 2.4 was failing to boot on i386 machines [linuxcare.com]. It had something to do with the CPU's cache not being large enough i belive.

    after looking at the change log think this might be it..


    Fix UDF writepage() page locking


    anyone know for sure?

    -Jon

    Streamripper [sourceforge.net]

  • While Microsoft isn't liable for the same reason that no software company is liable (years and years of crappy developers have made users come to *expect* buggy software) "liability" goes against the freedom of Linux. If you want Linux to remain free and democratic and all that shit, then you can't enforce product quality. That's why the LSB is such a bad idea. It forces vendors to conform and release good (or at least uniformly mediocre) products.

    PS> Yea I know /. is the "I don't get subtly" capital of the world.
  • Uh, have you read some of Linus Torvald's and Andre Hedrick's linux kernel mailing list posts? These guys are Grade "A" assholes, and thier code is just awesome.

    Andre Hedrick (ide guy)is one strange cookie, here [slashdot.org] is the slashdot interview with him. Read his answers. Make any sense? Should this man be committed?

    Regarding that hard drive copy protection:
    Choices... by cnladd
    I apologize for the open-endedness of this question, but I have to ask it anyways. :)
    If this copy protection were to become mandatory, I can definately imagine the effects that it would cause. But what effects - both long and short term - do you feel this would cause?

    Andre:
    Sorry, I do not feel anything! If you wish to know what I THINK, then I will answer the question. The very nature of asking people how they feel about an issue allows one to wrap it in fuzzy language, and this is how we got into this mess. So THINK DAMN-IT do not FEEL, this is silicon and not flesh!
    Think about all the software you own for backup -- WORTHLESS in a CPRM environment. OPEN wallets!!!!
    Ever had a morning where you were not kissed and told "I love you," when the night before you SCREWED so wildly that you could not remember?
    GOOD MORNING!!!!


    Have fun using thier code, I know I do, and I get a real kick out of reading the kernel mailing list, too.
  • You really don't have to, since reiserfs is resizable (though, for that matter, so is ext2 though only offline). Particularly easy if you have LVM and the ext2online patch applied (in which case all of the following should be possible without any disruption in service whatsoever):

    Shrink your ext2 filesystem. Create a reiserfs in now-free space. Copy data over. Delete ext2 filesystem and resize reiser-based fs to fill remaining space.

    Okay, done. You don't have the ext2online patches? Means you'll have to use parted or PartitionMagic for the first step... a shame, that. You aren't using LVM? You'll have to move the ext2 partition to the back of the disk as well as resizing it (something parted won't do but PartitionMagic probably will) as a reiserfs partition can only online-extend from its tail.

    Still quite doable.
  • Everybody knows that only the 31337 H4X0RZ (or 'leet haxors) call it LIN-U-UX. It only the script kiddies would ever think something different.
  • Try using sndconfig next time you're on a Red Hat or Mandrake box, and listening to the sample wave:

    "Hello, this is Linus (rhymes with `Guinness') Torvalds. And I pronounce Linux (rhymes with `cynics') Linux (still rhymes with `cynics')".

  • Fix UDF writepage() page locking

    This might be referring to MicroUDF [also called UDF], the filesystem used on DVDs [as well as other devices] and integrated into stable kernels later than 2.2.16 and 2.4.0
  • No, no, no! I'm on a winblows computer at school! I can't download it the instant it's out - I won't be part of the /. effect against kernel.org!

    If you've read the ChangeLog, you'll notice that there's a change for almost everything. Regardless of what you use your computer for, and update will probably help you out. Remove all your 2.4-ac patches and tack on the 2.4.1 series patch, then get the next AC patch due out in five minutes (Alan Cox has a LOT of time on his hands based on how fast his patches were released when 2.4 came - as many as 2 a day!).

    The problem with capped Karma is it only goes down...

  • Fully modularized code also means that in BeOS, my NVIDIA drivers are a mere untar away on any BeOS version, while in Linux, I almost invariably have to recompile the drivers. Face it, the Linux driver interface is broken. The sheer fact that NVIDIA has to go through so many hoops to release a binary driver shows just how broken it is.
  • My point was that the person I was replying to was wrong, the Debian version is not the same as the kernel version. If it were, then Debian would end up using the 2.3-devel kernels with their next release ;)
  • by rich22 (156003) on Tuesday January 30, 2001 @05:17AM (#470878) Journal
    I was talking to an MSCE buddy of mine yesterday, and he swore up and down he is running Linux 7.1 now.
  • Then that's more or less irrelevant. a .0.X release is nearly the same as a service pack. Tons of stuff breaks during a .0.X release (such as drivers, ahem) while service packs rarely (if you're lucky, some people aren't) break anything.
  • by macdaddy (38372)
    I see a PPC update from Cortd for pre10. God I sure hope they got all the PPC fixs merged in. Unfortunately it's very rare when we PPC users can take a stock kernel from kernel.org and make it compile correctly with all the fixes. Maybe it's just me but it seems that Linus never waits for our fixes. They merge everyone else's crap in at the last second which almost always breaks some PPC thing. Then they release it before the PPC developers can get it fixed. It's really annoying to feel unimportant. This doesn't just affect Mac users but the RS/6000's, Amigas, embedded machines, even Tivo! Like I said, I sure hope they merged in all the PPC fixes this time.

    --

  • Except that they can. A much higher percentage of Win95 software runs on Win98 than 2.2 software runs (perfectly) on 2.4. The same is true in reverse (for the most part, discounting apps that use '98 specific APIs.) Of course, if your talking about NT and '95, you're on crack. Complaining that NT doesn't run all '95 apps is silly. They're totally different OSs.
  • For your specific example. In the closed-source world of Microsoft, Sun, etc...

    First of all, since the compiler is not bundled with the OS it could be released a few months after the OS. In the meantime making the beta available for people to use and test.

    In fact, both companies have done this quite frequently.

    Your characterization of closed-source companies while entertaining tends to ignore facts and history.

    As far as the RedHat making it work better than it should... Need we remind you of that RameN Worm? :)
  • by Royster (16042) on Tuesday January 30, 2001 @07:05AM (#470888) Homepage
    2.4.0 was released very early in January, not "just last week".

    If MS released a patch or hotfix a week after a major release, it would be because of a horiffic bug or security hole. The *reason* for the release is important.

    2.4.1 adds functionality. ReiserFS is a journaling filesystem that is now in the official kernel. That's good news.

    Linux has had problems with 2.x.1 releases being les stable than 2.x.0. Linus intends this not to be the case.

    Producing patches often is a good thing whether it fixes a bug or adds functionality. That dosn't mean that any patch or upgrade can't simultaneously represent an embarassment for a developer. That depends entirely on the reason for the release.
  • So 2.3 will use a devel kernel?
  • You caon't have it both ways. Linux is either about Open-ness, or quality. Sometimes the two coincide, sometimes they don't. Take your pick.
  • by Shoeboy (16224)
    http://www.cnn.com/TECH/computing/9810/09/nt5.idg/ [cnn.com]
    I'm not sure if it meets your definition of a "fucking" source, but here's an article from CNN in 1998 that discusses in passing how NT 5 was supposed to ship in 1997.
    --Shoeboy
  • 'Cause if you don't release it officially, you'd never get it tested. Bet you didn't know M$ has about a million different versions of Win95 (and 98 too - don't forget Service packs on top of THAT)!

    The problem with capped Karma is it only goes down...
  • by Anonymous Coward
    Incidentally, there's no such thing as egcs anymore.

    Oceaneia has always been at war with Eurasia.
  • Oh, nuts! And I'm literally in the middle of a 2.4.0 install right now!


    Chelloveck
  • A commercial company like Microsoft wouldn't do this because they have the financial responsibility of selling their goods.

    I take it you have never bothered actually using any of Microsofts products.
  • www.ethernalquest.org/linux-2.4.1.tar.bz2 No gurantee that it will stay up tho =) (will be up in like 5-10 mins)
  • The mirrors won't be overflowing because half the linux community is at the Linux World Expo in NYC for the next few days. Too bad I'm stuck in !@#%$%$!#!@#$ school, otherwise I'd be using my ticket and be on my way there right now =(
  • The -preX designations represent interim test versions that were released between 2.4.0 and 2.4.1, ex: 2.4.1-pre6

    --
  • That would be true if it were a compiler error. However, fatal internal errors should not arise, even on code deemed incorrect by that compiler. They should just generate an error message, describing the syntax in the code it didn't like.

    When the compiler explodes, leaving pasta all over the walls, there is clearly a glitch that wasn't trapped.

  • The CVS tree says otherwise. :)

    But, yeah, you're right, bugs -are- going to crop up in snapshots. That's natural. That's why it's in CVS and not a final release.

    On the other hand, IMHO, if you're on the leading edge, you're going to want to know what other leading edges yours is compatiable with. It's less stressing and more knowing the lay of the land.

  • I got a good signature on linux-2.4.1.tar.bz2 that I downloaded earlier today. I can't verify the tar.gz signature since I don't have the file.

    Certainly their signing isn't totally bad. I'd suggest you try downloading from a different mirror and see if it comes out clean. You might also want to check your gpg version--I have 1.0.4.

  • I liked this comment on the poll:

    "Read as an english word linux is pronounced lie nucks. Read as a finnish word linux is pronounced lee nucks. Saying lynn (like the name) ucks is blatantly wrong. But, pronounciation of a word from another language is up to the speaker, and most people who speak english can't pronounce anything at all. So, although about 63% of the posters are wrong, it's up to them what they say."

    Probably the best thing to do is to realize it's a microcultural thing. People in one company will call it LyeNicks, while the company across the street the people call it Linnukz.

    Instead of dorkboy saying "I cannot have a serious conversation with someone that calls it LyeNicks", he should show himself as properly educated to the fact that people pronounce it differently depending on something obscure like the phase of the moon when they first heard about Linux.

  • Huh? I've got Mandrake 7.2 and RH 6.2 systems that run just fine with the 2.4.0 kernel. I didn't have to upgrade anything on Mandrake, and RH only needed a new modutils (which I got by pulling the modutils-2.4.1 source RPM from mandrake's cooker and rebuilding).

    It took almost no time to upgrade either, and they work fine.

    1st Law Of Networking: Loose ends are bad, termination is good.

  • by sharkey (16670) on Tuesday January 30, 2001 @08:33AM (#470924)
    2.4.x uses netfilter for packet mangling. Here is the relevant portion of the FAQ: http://netfilter.kernelnotes.org/netfilter-faq-1.h tml#ss1.4. [kernelnotes.org]

    You can use ipchains to control the filters, NAT, etc., I believe, but iptables is the new user-space tool du jour. The page for the project is http://netfilter.kernelnotes.org/. [kernelnotes.org] I can't make a personal testimonial yet, I haven't had much chance to play with iptables/netfilter/2.4.x firewalling.

    --
  • No, my comment is completely correct. Releasing new versions of the device drivers in the kernel means releasing new kernels. The fact you can take a 2.2.18 driver and load it into your 2.2.8 kernel doesn't change how device drivers are released.

    I can't understand why people have such a problem with the concept. I've had 5 replies now to my posting by people who don't understand that device drivers are part of the Linux kernel. I suggest that those still having problems with the concept go read the change logs. If releasing updates to device drivers wasn't part of releasing new kernel versions, you wouldn't see anything about device drivers in those change logs now would you?

    Well, would you?

    Incidentally, video card device drivers are not usually part of Linux. That's a FWIW. Usually the support is via XFree86 instead.

    But for the umpteemph time: Go read the original article. Now read mine. Now see how it explains why Windows is updated just as often as Linux, only because of the different structural differences, the updates to Linux are just more "obvious".
    --

  • Except that the term "stable" is a bit of a misnomer, as anyone who ran 2.2.0 knows.


    I think the real versioning system is "anything less than x.x.20 is unstable, and anything with an odd middle version number is even more unstable."

  • you just have linux 2.4 tho .. not 2.4.1
  • Any particular reason the kernel distribution system still relies exclusively on manual selection of mirrors by hostname? What's wrong with an automated load-balancing setup?
  • Not really... the 2.2 branch hasn't even reached 2.2.20 yet, and it's been really stable for a while. basically, among "stable" (even middle number) kernels, you have to go by reputation (or do some heavy testing yourself, if you have the time and resources). there have been cases where a stable revision (within the 2.2 branch, or earlier within the 2.0 one) was worse than the previous one. in the 2.2 branch, 2.2.5 was already quite stable. my feeling with 2.4 is that it has been handled better than 2.2; I wouldn't be surprised to see something like 2.4.2 shipped by major distributions and widely deployed.
  • I'd far rather wait until it's ready than have incomplete and buggy releases rushed out to meet a predetermined schedule, which is what those "professional" software houses do far too frequently. Not having a rigid schedule is one of the benefits of open source, not a liability.
  • I should know. I'm one of the conspirators that started egcs.

    The egcs and FSF gcc projects merged in April 1999, and put out the first combined release (2.95) at the end of July 1999.

    The GCC (not EGCS) snapshots represent ongoing development and occasionally something breaks. Please report such things, but don't stress about it; you should expect to find bugs in the snapshots.

  • That's Microsoft Certified, but I use Linux and Mac OS X throughout the day as well. Not all MCSEs are MS'centric. Some just need it (actually, the study, not the testing and certification) for the job they're doing at the time.

  • by iamsure (66666) on Tuesday January 30, 2001 @09:12AM (#470950) Homepage
    I have now downloaded from ftp.us.kernel.org, and from ftp.kernel.org.

    I did the following:

    # gpg --keyserver wwwkeys.pgp.net --recv-keys 0x517D0F0E

    then I did:

    # gpg --verify linux-2.4.1.tar.gz.sign linux-2.4.1.tar.gz

    gpg: Signature made Tue 30 Jan 2001 02:56:09 AM EST using DSA key ID 517D0F0E

    gpg: BAD signature from "Linux Kernel Archives Verification Key "

    And finally, just to be sure, I did:

    # gpg --list-public-keys

    /root/.gnupg/pubring.gpg

    pub 1024D/517D0F0E 2000-10-10 Linux Kernel Archives Verification Key

    sub 4096g/E50A8F2A 2000-10-10

    Which looked valid to me.

    Is the new kernel not signed properly, am I not doing this properly, or is something VERY BAD happening?

  • Even though I despise Windows, I'd be happy if MS did follow Linux' model and released patches more often

    I guess you're happy then? [microsoft.com]
  • by Eil (82413)

    I count seven minutes between the initial annoucement and the "go ahead" post. :P
  • That's how I've pronounced it for the last 6 years. I find it difficult to have a conversation with these newbies who insist on calling it LIN-ix.

    Linux means LINus's UniX. In his original tongue, Linus pronounces it like Lee-nukes. Properly 'americanized', it is LYE-NIX/Line-ix.

    There's a lot of Linux newbies out there that think that the correct pronounciation IS LINN-IX. Incorrect: Lennox makes air conditioners and refridgerators.

    Funny that this guy is so confused that he is mistakenly informing people to pronounce it the WRONG way. :)

  • by redhog (15207) on Tuesday January 30, 2001 @05:27AM (#470965) Homepage
    pre8:
    - ReiserFS merge

    Nice, nice, nice. Finally they agree.

  • I have a similar system and 2.4.0 worked great. The slight boost in kernel performance combined with reiserfs patch makes this Athlon box cook.

    But my PII266 laptop is quite another story. 2.4.0 wouldn't detect the pcmcia network card, sound would not work AT ALL (despite it being a 100% SB clone), and the weirdest thing by far was the fact that I would find my BIOS settings cleared every time I rebooted. And this was on a (fairly standard) mobile 440BX chipset. I had to revert to 2.2.18 and everything's been fine since.

    So, I think that I have to concur with people who warned folks like me not to upgrade to 2.4 unless they had a good reason to. :P
  • okis, its up now =)

  • Because automatically directing to a server based on location tends to take away people's choices. (This can be worked around, but I'm referring to what is common practice these days.) What happens when the mirror is not up to date? Or the page is gone (404'd)? Or the server is down?

    I personally like to select my own mirrors so that I can choose who I download from. In other words, I will almost always choose the server of a for-profit corporate entity than that of a university or non-profit organization when downloading something as large as a kernel tarball.

    Plus, the people who run the kernel.org page reason that anyone downloading and building their own kernels ought to be smart enough to figure out to get to a mirror.
  • by squiggleslash (241428) on Tuesday January 30, 2001 @05:28AM (#470972) Homepage Journal
    Er, no. Windows is a very modular system (both "Old Windows", ie 3.1 - Me, and "Modern Windows", NT-2000, the latter family being based on something that was originally a microkernel) As a result, minor changes, such as device drivers, which are released all the time, have no effect on the kernel version number.

    Linux is a monolythic kernel. This means that low level system components such as device drivers are part of the kernel (even if they're loaded seperately, such as with modules.) As a result of this, releasing updated device drivers or other updated components means releasing a "new" kernel, even though the central structure and code is unchanged.

    New device drivers are released for Windows on a periodic basis, and other low level components are often changed without the user even knowing - popular applications from IE to Office regularly "update" the user's operating system to whatever was latest at the time that product was released.

    The fact that the version number of Windows doesn't change doesn't mean it doesn't change just as regularly, if not more often.
    --

  • by Shoeboy (16224) on Tuesday January 30, 2001 @05:28AM (#470974) Homepage
    Right right right,
    and professional houses ship things late all the time too, NT 5 was supposed to ship in '97.
    But I think some criticism of the slowness of kernel updates is self inflicted.
    Don't you recall the heady days of 1998 when all the linux advocates were boasting about the speed of releases and proclaiming open source as a magic bullet that would lead to instant release and completely bug free code?
    I think that the linux community is now suffering from a backlash brough on by having had too many advocates who were assholes, see Nick Petreley for a good example.
    If you don't want to face heavy bitter criticism, don't make outrageous claims.
    And for the love of God, read the linux-advocacy-howto.
    --Shoeboy
  • It has now been confirmed on multiple (more than 3) : Hosts, distros, ftp clients used to download, mirrors, AND versions of gpg.

    The .gz sign is a bad signature.
  • Nope. Checked. :)

    It has now been confirmed on multiple (more than 3) : Hosts, distros, ftp clients used to download, mirrors, AND versions of gpg.

    The .gz sign is a bad signature.
  • Woohoo! .bz2 works! Allright! Thank you! I couldnt upgrade til I was sure it wasnt 0wned.. :)

    Thank you very much.

    Remember, it always pays to be paranoid.
  • Well, you can leave the hostnames so people can manually choose "ftp.au.kernel.org" if they so wish, but set it up so the generic ftp.kernel.org load-balances to the nearest mirror (for those who decide not to pick a mirror themselves).

    Of course I suppose if they can handle all these requests piling up on the main ftp.kernel.org site then that's their business...
  • by Zigg (64962) on Tuesday January 30, 2001 @05:35AM (#470998)

    What is this mysterious "kernel" package I keep hearing about? Have we finally come into the new world order where all operating systems run on one kernel that does not need a name? God bless America, I knew we'd standardize someday.

    Note: this post is sarcastic, just in case you're wondering.

  • There is accelerated support for the GeForce2 in Xfree86-4.0.2. Also a new .9.6 version of the drivers came out pretty recently from NVidia and I was able to get them working. I wasn't with .9.5 due to incompatibilities with my AGP bus.

    Don't hold me to this, but if you download the cvs of Xfree86 there is most likely a dri module for the GeForce2 by now (since they got the accelerated server, I bet they are workin on the dri module in due course).
  • by AntiNorm (155641) on Tuesday January 30, 2001 @05:36AM (#471002)
    2.4.2 - The answer to life, the universe, and everything
    (Well, sort of)

    ---
    Check in...OK! Check out...OK!
  • by nels_tomlinson (106413) on Tuesday January 30, 2001 @05:37AM (#471009) Homepage
    I'm sure that the post this answers was intended as irony, but for just in case some newcomer is reading this: when the 2.4.0 kernel was released, there had been people running it for many months, sometimes for many months without rebooting. One of the wonders of opensource is that you don't have to wait for the release, and so the release can happen when the product is ready, not when the business plan calls for it.

    For the other side of this, consider Redhat 7.x. Their business plan called for a release when the compiler they wanted wasn't ready. In the closed-source paradigm, they would have called it ready and shipped bugs. Since the compiler is GPL'd they had to explicitly ship a beta compiler, and we got some fair warning about those bugs which we wouldn't have gotten from Microsoft or Sun. By the way, Redhat has done a wonderful job of making that work far better than it should, to judge by the reports of people who have been using it. In the usual closed-source, proprietary course of events, a closed source vendor such as Sun or MS would have denied the bugs, threatened customers to try to hush things up, and the folks who laid out big bucks for the bugs would have had to pay for an upgrade.

    How is Linux ever to become a commercial success/serious platform if development takes years? Same way it's been getting there all along, I guess, by being so much better than the stuff that's rushed out the door to keep the marketing department happy.

  • 2.4.0-test9: works
    2.4.0: broken.
    2.4.1: still broken.

    Just a little bit of a problem for people who sensibly are using a megaraid card for their root filesystem.

    I just mailed Linus...

  • http://www.linuxdoc.org/HOWTO/Kernel-HOWTO.html
  • I don't know if it's all the advocates fault. It hink it may play more into the whole open source methodology from it days back as a grass roots effort. Spin, spin, spin, worse than politicians was the name of the game. Now that the methodology has stabilized somewhat Open Source fundamentals and methodology are being hit with the same issues and problems that BIGCO software development shop is hit with. Poor project management, over zealous targets, too much release functionality and too little testing. Too bad that most of the benefit of participating with open source projects is now lost and the "pirate ship" mentality is being lost. Most open source efforts that are widely seen via the press, like the 2.4.1 release is being tarnished by the same poor management practices that tarnish mainstream products like those from Microsoft to use your example.

    Did I just lump the open source management community into the same bucket as Microsoft? Guess I did. Around /. I know that's quite a politically incorrect thing to do, but it is accurate. Software development problems are software development problems, regardless of the passion of the developers. Open Source management needs to wake up and smell the coffee or the open source projects like this will all get the reputation of untested, underdeveloped hackware.

  • by jd (1658) <imipak@yaCOLAhoo.com minus caffeine> on Tuesday January 30, 2001 @06:00AM (#471026) Homepage Journal
    I've not checked 2.4.1 yet, but many of the AC releases and -pre releases will NOT compile under PGCC or the EGCS CVS snapshots. Something -very- subtle has changed that will cause internal errors in these specific compilers.

    ("Stable" EGCS releases are fine. CVS snapshots older than 2-3 weeks ago seem to work, also, but no guarantee that the binaries'll actually do anything useful.)

    I've reported the bug to the EGCS developers, as internal errors are definitely a compiler bug, EVEN IF it's also a kernel bug.

    Having written all this, I'm now wondering if I'm the only Slashdotian insane enough to use bleeding-edge software compiled with other bleeding-edge software on production machines...

  • It's a fair comment, with an element of truth in it.

    However, 2.4.0 was the major Linux oddity. True open sourcers follow the creed 'Release Early, Release Often' (RERO). Linus probably deservedly got a lot of flak for the long time delay between 2.2 and 2.4. In the end I think 2.4.0 got released just in order to try and impose a 'feature freeze' and stop various developers trying to get their new whiz bang feature in before the shutters came down.

    There is a conflict between RERO and another Open Source aphorism which is 'Its Ready when Its Ready', which means that we as a community don't like to give out release timetables of longer than a few weeks; in other words we try not to speak of vapourware.

    I have no objection to rapid patches, which is not something MS is known for, although thankfully MS quality appears to be improving with the latest version of IE and Windows 2K.

  • You can have it both ways, because Windows and Free Software have different development models.

    A GNU/Linux kernel upgrade is a different beast than a Service Pack. An SP often breaks existing software without warning, or changes functionality. A new kernel does all these things, and sometimes more, but its part of a (somewhat) known development path, and its largely an optional thing. I have machines running 2.0.36 happily, and 2.2.18 as well. Since all machines in question are uniprocessor and have fully functioning hardware drivers, I have no need go to 2.4. I'll wait until there's a need or maybe I just want to play around. The same cannot be said of SP's which often roll up critical security fixes and performance hacks.

    Now, I'll agree to statements about the pace of kernel development being... off (patches every other day for a week after months between patches?) but for the most part, you're comparing apples and oranges.
  • 2.4.1 includes new features, like a journalling filesystem, not just fixes for existing stuff. Also, 2.4 (and 2.2) works a HELL of a lot better on my hardware (ASUS A7V + Athlon + GeForce 2 GTS) than Windows 2000 did (which would lockup hard within an hour, every time). I had to install new drivers for everything plus service pack 1 to even get it remotely stable. Of course, on the same hardware, I can run repeating concurrent kernel compiles with -j4 and not see a single crash for a week.

    Plus, I didn't even have to pay for Linux.
  • If I had any mod points left, I'd give you one. However, let me take a slightly different POV just for the sake of discussion.

    The OS philosophy is different than the MS philosophy. The MS philosophy is that of a traditional (closed) software shop. Consumers expect that when the doors open and a new product is released, that it be "good" and usable in all ways. We don't want to have to keep updating, patching, and fixing it. Isn't that what they were doing behind those closed doors, after all? In the OS world, everything's open for peer review. We know exactly how the process is coming along, and can participate in it. We don't have to wait for "perfection" because we understand that everything is just another step in the process. In fact, waiting for the "finished product" would mean waiting forever, because there's always one more feature or one more performance tweak to get in.

    What I'm aiming at here is that the MS's of the world release products, but the OS world releases features. It's simply the difference in the philosophy of development between the two.

  • by phaze3000 (204500) on Tuesday January 30, 2001 @06:07AM (#471039) Homepage
    get it from:

    ftp://ftp.COUNTRYCODE.kernel.org/pub/linux/kernel/ v2.4/

    Where country code is your country, eg uk for uk, us for the us, nl for holland etc etc.

    I everyone keeps downloading from the main site then it creates problems for the mirrors, which believe me is a bad thing

    --
    • ppp still doesn't work on my laptop using a >2.4 kernel (i *know* how to compile it, thanks.)
    • my network card is still unrecognized
    • my sound chip (Maestro2E) start whizzing when the laptop becomes hotter than usual...
    • I don't have some free space to reiserfs-ize my root
    and...

    my 2.2.18 rocks damn' well... :-)
    Why should I change ?
    To discover the evil of technological inflation ?
    No, thanks, I am not missing the windows community ?
    --

Pause for storage relocation.

Working...