| |||||||||||||
The first 3.0 kernel prepatch is out
Linus has announced the availability of the
3.0-rc1 kernel prepatch. "So what are the big changes?
NOTHING. Absolutely nothing. Sure, we have the usual two thirds driver
changes, and a lot of random fixes, but the point is that 3.0 is
*just* about renumbering, we are very much *not* doing a KDE-4 or a
Gnome-3 here. No breakage, no special scary new features, nothing at
all like that. We've been doing time-based releases for many years
now, this is in no way about features. If you want an excuse for the
renumbering, you really should look at the time-based one ('20 years')
instead." Perhaps the most significant changes since our first merge window summary are the addition of
Cleancache and the namespace file descriptors patch.
We'll have a second merge window summary out shortly;
in the meantime, it seems that kernel.org isn't quite ready to show
changelogs for the new numbering scheme - stay tuned.
(Log in to post comments)
The first 3.0 kernel prepatch is out Posted May 30, 2011 3:22 UTC (Mon) by Hausvib6 (guest, #70606) [Link]
So, since Linus mentions that the third number is for stable versioning. What's the next version after 3.0? 3.2?
The first 3.0 kernel prepatch is out Posted May 30, 2011 3:39 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]
3.1 would be the next version
The first 3.0 kernel prepatch is out Posted May 30, 2011 5:43 UTC (Mon) by Hausvib6 (guest, #70606) [Link]
Sweet, then after 3.9, will it be 3.10 or 4.0?
The first 3.0 kernel prepatch is out Posted May 30, 2011 5:50 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]
I think if the reasoning holds, 4.0 would be logical.
The first 3.0 kernel prepatch is out Posted May 30, 2011 5:57 UTC (Mon) by PaulWay (✭ supporter ✭, #45600) [Link]
I think it's more likely to be 3.10, and so forth up to whenever Linus decides that he can't count past that. Like 2.6.9 went to 2.6.10.
Have fun,
Paul
The first 3.0 kernel prepatch is out Posted May 30, 2011 12:17 UTC (Mon) by alspnost (guest, #2763) [Link] Surely the basic point is that the second number has ceased to exist. Exchange "2.6." for "3." and carry on. So we will now have 3.1.x and eventually 3.12.x, where 'x' is the stable patch number.
The first 3.0 kernel prepatch is out Posted May 31, 2011 20:45 UTC (Tue) by rfunk (subscriber, #4054) [Link]
Reminds me of Solaris about a decade ago, when Sun followed Solaris 2.6 with Solaris 7, effectively treating the "2." as silent.
(Meanwhile it was SunOS 5.7 under the hood, having already done a numbering shift when they introduced Solaris to succeed SunOS 4. And yes I'm oversimplifying.)
The first 3.0 kernel prepatch is out Posted May 30, 2011 5:37 UTC (Mon) by ikm (subscriber, #493) [Link]
> No breakage
> kernel.org isn't quite ready to show changelogs for the new numbering scheme
No breakage at all, indeed :)
The first 3.0 kernel prepatch is out Posted May 30, 2011 7:03 UTC (Mon) by ersi (subscriber, #64521) [Link]
What? No major changes? Like the kernel being rewritten in Visual Basic?
* 2.6.<odd>: still a stable kernel, but accept bigger changes leading up to it (timeframe: a month or two).
* 2.<odd>.x: aim for big changes that may destabilize the kernel for several releases (timeframe: a year or two)
* <odd>.x.x: Linus went crazy, broke absolutely everything, and rewrote the kernel to be a microkernel using a special message-passing version of Visual Basic. (timeframe: "we expect that he will be released from the mental institution in a decade or two").
The first 3.0 kernel prepatch is out Posted May 30, 2011 7:30 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]
This is odd.x and not odd.x.x. The rules are different :-)
The first 3.0 kernel prepatch is out Posted May 30, 2011 7:47 UTC (Mon) by MortenSickel (subscriber, #3238) [Link]
Last time I checked, 3 was an odd number... ;-)
M.
The first 3.0 kernel prepatch is out Posted May 30, 2011 8:36 UTC (Mon) by anselm (subscriber, #2796) [Link] Yes, but there's only one ».x« after »odd«.
The first 3.0 kernel prepatch is out Posted May 31, 2011 0:51 UTC (Tue) by Cyberax (✭ supporter ✭, #52523) [Link]
So we should wait until GregKH releases 3.0.1?
Or maybe we should wait for Linux 3.11? :)
The first 3.0 kernel prepatch is out Posted May 31, 2011 6:59 UTC (Tue) by Felix.Braun (subscriber, #3032) [Link]
Yay, Linux 3.11 could be here Christmas 2014!! That would be seriously cool!
The first 3.0 kernel prepatch is out Posted May 31, 2011 11:21 UTC (Tue) by jdv (subscriber, #712) [Link]
Linux 3.11 -- for Workgroups?
The first 3.0 kernel prepatch is out Posted Jun 1, 2011 3:09 UTC (Wed) by fest3er (guest, #60379) [Link] Linux 3.1.4: we'll be able to have our Linux and our pie.
Back to 3-level versions Posted May 30, 2011 8:24 UTC (Mon) by rvfh (subscriber, #31018) [Link]
Can't agree more with this versioning (there was one digit too much), and the reason for the change (20 years).
Thanks Linus!
The first 3.0 kernel prepatch is out Posted May 30, 2011 8:32 UTC (Mon) by salimma (subscriber, #34460) [Link]
Will be nice if the license changes to GPLv3+ with an explicit TiVoization permission, but that won't happen -- well, one could dream.
The first 3.0 kernel prepatch is out Posted May 30, 2011 19:07 UTC (Mon) by elanthis (guest, #6227) [Link]
Will be nice if the license changes to BSD/MIT/X/zlib, but that won't happen -- well, one could dream.
;)
The first 3.0 kernel prepatch is out Posted May 31, 2011 7:21 UTC (Tue) by lkundrak (subscriber, #43452) [Link]
Nice as in "even more vendors coming up with forks with crap closed patches from mainline." If you would like to live in the world where this can happen, please go for cheapo Android tablet or switch to Mac OS X.
By the way, certain parts of kernel are BSD licensed -- DRM for example. I highly recommend you running the closed Poulsbo DRM, just pick a Dell Mini 10 with and old-ish ubuntu and enjoy the non-GPL fun!
Linux kernel 3.0 Posted Jun 3, 2011 18:44 UTC (Fri) by davi (guest, #18853) [Link]
GPLv3+ versus BSD/MIT/X/zlib license.
Will be nice if Linus Torvalds and the rest of the community members who have contributed to the Linux kernel development (being individual or companies), transfer their copyright to me.
So I could sell it to it to Microsoft, for example, and leave this job ;)
Git trees with 2.6 in their name Posted May 30, 2011 9:31 UTC (Mon) by Aissen (subscriber, #59976) [Link] A quick search on http://git.kernel.org/ shows 354 git trees containing 2.6 in their name, including "mainline" i.e Linus' tree. (and HPA's very interesting auto-generated merged tree: linux-2.6-allstable) Is that gonna be changed as well ? As long as people are updating scripts, you might as well make them upgrade fetch/clone scripts (although this would be against good style)
Git trees with 2.6 in their name Posted May 30, 2011 9:46 UTC (Mon) by jku (subscriber, #42379) [Link]
This is covered in the email thread: Git trees are not renamed, a "symlink or something" is possible in the future if people dislike seeing '2.6' in there.
The first 3.0 kernel prepatch is out Posted May 30, 2011 10:34 UTC (Mon) by cesarb (subscriber, #6266) [Link]
I wonder how much userspace is going to break because `uname -r` now would start with "3.".
I would not be surprised if, for instance, "config.guess" style scripts try to parse the kernel version for some random reason.
The first 3.0 kernel prepatch is out Posted May 30, 2011 11:08 UTC (Mon) by jengelh (subscriber, #33263) [Link]
Damn those developer assumptions! They should have used a ">="-"I feel lucky"-style check whenever possible, like PKG_CHECK_MODULES([foo], [foo >= 2.9]) is used in many .ac scripts. If they can't do that even for uname -r, that's a big developer fail.
The first 3.0 kernel prepatch is out Posted May 30, 2011 11:17 UTC (Mon) by arekm (subscriber, #4846) [Link]
The bigger problem for script will be that now it's going to be x.y instead of x.y.z. Some configure* scripts split that by dot and tries to figure out versions.
Generic version checking macros for linux kernel never existed (or any version checking macros) in autoconf land. Everyone did that in own way.
The first 3.0 kernel prepatch is out Posted May 31, 2011 14:03 UTC (Tue) by rriggs (subscriber, #11598) [Link]
It seems to me that the kernel should be versioned such that x.y.0 is the "Linus" release, with any patch level > 0 being from the long term maintenance team.
The first 3.0 kernel prepatch is out Posted May 30, 2011 11:40 UTC (Mon) by robbe (guest, #16131) [Link]
I suspect more scripts to break due to two-component version numbers (x.y) than because of the new major.
The first 3.0 kernel prepatch is out Posted May 30, 2011 14:17 UTC (Mon) by adobriyan (guest, #30858) [Link]
One also wonders how many of those "uname -r" or even uname(2) invocations
were right thing to do and how many of them were developer's lazyness in checking for a exact feature he needed.
The first 3.0 kernel prepatch is out Posted May 30, 2011 11:13 UTC (Mon) by Yorick (subscriber, #19241) [Link]
Changing the version numbering just because a long time has passed since it was last done makes no sense whatsoever. A version number, in order to be useful, should communicate something.
Sometimes a new major version serves as a warning sign: that although little has happened on the surface, deep structural changes and rewritings may have affected stability or compabitility. It could also be used to alert the user of radical new ways of operation, or as an advertisement for large improvements in usefulness, be it functionality, performance or stability.
Merely that a part of the version number hasn't changed for several years isn't really anything that users need to be informed about - they already know that or at least are not affected by the fact.
Of course Linus is entitled to use whatever scheme he likes with any or no justification, but a little rationality would help. When commercial software vendors play with version numbers as an attention-getting PR device even though nothing has really changed at all, we respond with patience and/or contempt. Will we do the same now?
The first 3.0 kernel prepatch is out Posted May 30, 2011 11:36 UTC (Mon) by NAR (subscriber, #1313) [Link]
For the big change I suggest forgetting Linux on the desktop and concentrate on servers and embedded systems. I've just tried the latest Kubuntu on an Intel videocard - the stability reminds me the Win 3.11 days.
The first 3.0 kernel prepatch is out Posted May 30, 2011 11:47 UTC (Mon) by tialaramex (subscriber, #21167) [Link]
We know, sometimes there are bugs. And every bug feels like the end of the world to someone.
A friend's Mac was "broken", he complained "nothing works" and "I need a new Mac". It turned out the deskbar or whatever its called had lost some state, simply restarting that program took his experience of the Mac from "nothing works" to "oh, that's fine now".
I'm actually seeing more Linux on random netbooks and laptops among the layfolk than ever before. That might be a local blip, but so might your Intel Kubuntu experience for all I know.
The first 3.0 kernel prepatch is out Posted May 31, 2011 17:00 UTC (Tue) by robert_s (subscriber, #42402) [Link]
Could you link to the bugs please?
The first 3.0 kernel prepatch is out Posted May 30, 2011 11:45 UTC (Mon) by jku (subscriber, #42379) [Link]
So if project A makes improvements in small steps while project B makes the same improvements in one single release, only project B is allowed to make a major version number change? That doesn't make any sense to me, and does _not_ help me as a user understand the differences between version numbers.
Difference between linux 2.0 and 2.6.39 is massive. If that's not enough to warrant a version number increase, I don't understand what would :)
The first 3.0 kernel prepatch is out Posted May 30, 2011 12:52 UTC (Mon) by Yorick (subscriber, #19241) [Link]
That is a valid observation. Clearly, the users' knowledge should affect versioning. Now, not many people are considering switching from Linux 2.0 to 2.6.39, but your example was of course stretched to make a point.
On the other hand, big steps in version numbering at arbitrary points every now and then do not help either, unless there is somehow a guarantee that two versions with the same major number do not differ in any way that would deserve such an advertisment.
The first 3.0 kernel prepatch is out Posted May 31, 2011 10:12 UTC (Tue) by jwakely (subscriber, #60262) [Link]
> Now, not many people are considering switching from Linux 2.0 to 2.6.39, but your example was of course stretched to make a point.
How many people actually think "hmm, do I want to upgrade my kernel from X to Y?" rather than just taking whatever their distro offers, and maybe upgrading their distro release? Hint: it's not a significant number.
N.B. LWN subscribers are probably not very representative, so "I do" and "me too" replies are unnecessary
The first 3.0 kernel prepatch is out Posted May 30, 2011 13:49 UTC (Mon) by SEMW (guest, #52697) [Link]
The rationalisation seems pretty clear to me. The current situation, where each release significant changes only the 3rd level (below-minor) version is silly. (As jku points out, you get a situation where 2.6.0 and 2.6.39 are vastly different, with no change to even the minor version). The obvious remedy is to switch to incrementing the minor version with each release.
The move to 3.0, in this picture, is just a way of doing this from a clean slate. The alternative, going 2.7, 2.8, 2.9.. is less inelegant -- apart from anything else, the 2.x series is bound up with assumptions about 2.odd being development-only.
The first 3.0 kernel prepatch is out Posted May 30, 2011 19:53 UTC (Mon) by elanthis (guest, #6227) [Link]
Then Linux should just be numbered like Chrome, newer Firefox releases, systemd, udev, etc.: a single version number component, incremented for each release. Yes, in a few years we'd be at Linux 157 or whatever, but that doesn't matter; it would just mean it's the release after 156 and before 157.
In all honesty, "users" don't give a shit about the version number, especially not for a low-level component like the kernel. They care more about the version numbers of the system as a whole (Ubuntu, Fedora, Android, whatever), and even then they only really care about "is this recent or not, and can I Google this number to get an idea of what features it has or is missing."
The first 3.0 kernel prepatch is out Posted May 31, 2011 7:13 UTC (Tue) by jezuch (subscriber, #52988) [Link]
> Changing the version numbering just because a long time has passed since it was last done makes no sense whatsoever. A version number, in order to be useful, should communicate something.
Well, except in mature codebases, where there's absolutely no reason to do anything "big". And where there's no reason to do anything "big", there's no reason to bump the more "major" version numbers. And thus version numbers no longer make sense or mean anything important. And voices in Linus's head say it means it's time for changes.
Linux being stuck in the "2.6.x" version line means one thing only: it's mature (and has been for quite some time). And it's a good thing :)
The first 3.0 kernel prepatch is out Posted May 31, 2011 10:03 UTC (Tue) by Yorick (subscriber, #19241) [Link]
I don't think that is quite true. GCC is very much a mature code base, and yet they occasionally do fairly big internal changes that, while perhaps not always user-visible definitely merit alerting the user. I believe the decision to go from 3.4 to 4.0 was motivated exactly in this way.
The first 3.0 kernel prepatch is out Posted May 31, 2011 20:55 UTC (Tue) by nevets (subscriber, #11875) [Link]
Well there was a big change for this release. We went from 3 digit kernel releases to 2 digit kernel releases. This has already broken some user space scripts, thus it is indeed rational for a change in the major version number. ;)
Honestly, I think Linus missed the boat a long time ago. I've been saying that 2.6.0 should have been 3.0, as the development model changed with 2.6. The 3.0 could have marked the end of the X.<odd/even> development model.
The first 3.0 kernel prepatch is out Posted May 31, 2011 21:12 UTC (Tue) by dlang (✭ supporter ✭, #313) [Link]
the development model didn't change immediately with 2.6.0.
it was several releases after that that Linus made the change. (I don't remember the exact version, but I believe it was after 2.6.7)
The first 3.0 kernel prepatch is out Posted May 30, 2011 11:58 UTC (Mon) by vonbrand (subscriber, #4458) [Link] Methinks this "voices in the head" thing is going too far... Anybody seen Linus' medication?
The first 3.0 kernel prepatch is out Posted May 30, 2011 20:02 UTC (Mon) by drag (subscriber, #31333) [Link]
As long as the voices continue to make sense and don't tell him to try to eat the family dog or anything like that then I don't see a problem with them.
They are certainly smarter then the voices in MY head. A bunch of stupid rednecks, all of them.
Voices Posted May 30, 2011 22:53 UTC (Mon) by tialaramex (subscriber, #21167) [Link]
Indeed.
Linus says "voices in my head" which can just refer to an internal dialogue, which is squarely inside human normal. If he actually _hears_ voices, and if those voices seem independent (of course they aren't but it's the perception which matters) then that's a auditory hallucination and yes, that's not normal. But it's still not an illness and so there's no reason to take potentially dangerous pills so long as it's not causing a problem.
A "voice in your head" which tells you to call your mother is normal
Voices Posted May 31, 2011 19:58 UTC (Tue) by drag (subscriber, #31333) [Link]
> we have drugs which might help.
I am sure that there are all sorts of drugs that will make you hear all types of different voices. I think I'll pass this time.
but it's nice for you to offer.
Voices Posted Jun 1, 2011 1:36 UTC (Wed) by vonbrand (subscriber, #4458) [Link] Warning! Leg strain gauge out of order.
The first 3.0 kernel prepatch is out Posted May 30, 2011 15:01 UTC (Mon) by Julie (✭ supporter ✭, #66693) [Link] [Posted May 30, 2011 by corbet]
Re-winding to May 25th's posts... [Posted May 25, 2011 by jake]
What happened? (You just can't stay away, can you...) ;-)
The first 3.0 kernel prepatch is out Posted May 30, 2011 17:39 UTC (Mon) by corbet (editor, #1) [Link] If you're spending your "holiday" in airports, you might as well post some news...:)
The first 3.0 kernel prepatch is out Posted May 31, 2011 5:00 UTC (Tue) by jmranger (subscriber, #43784) [Link]
To me, this sounds like a variant of Linus releasing new kernels just after the weekly edition is online.
From now on, be prepared for him playing you tricks when you state your holiday plans early...
The first 3.0 kernel prepatch is out Posted May 31, 2011 11:01 UTC (Tue) by dgm (subscriber, #49227) [Link]
I thought it was well established that reading LWN was one of the few things in Linus routine. In fact it's the item just before releasing any new kernel. How else could he discover what's great in the latest RC? I'm sure it helps him get into the mood of writing some hilarious release comments.
The first 3.0 kernel prepatch is out Posted May 31, 2011 1:10 UTC (Tue) by jcm (subscriber, #18262) [Link]
Fix for module-init-tools coming shortly. I suspect other programs and scripts will spontaneously explode though with the new numbering. Looks like Linus has gone for 3.0.0 in the interim.
Jon.
The first 3.0 kernel prepatch is out Posted May 31, 2011 6:57 UTC (Tue) by jcm (subscriber, #18262) [Link]
Pushed it out for Fedora, but there seems to be some problem with hera syncing tonight, so it's not showing up on kernel.org yet.
The first 3.0 kernel prepatch is out Posted May 31, 2011 9:05 UTC (Tue) by ejmarkow (guest, #56170) [Link]
After testing it out, compiling and adjusting Grub respectively, the system crashed after booting up. I notice the directory created in '/lib/modules/' is named '3.0.0-rc1' as opposed to '3.0-rc1' in '/usr/src/' where the source was extracted.
The first 3.0 kernel prepatch is out Posted May 31, 2011 13:38 UTC (Tue) by jengelh (subscriber, #33263) [Link]
And it's in the commit log of the v3.0-rc1 commit referencing tag as to why (namely, because the details have not been figured out yet).
The first 3.0 kernel prepatch is out Posted May 31, 2011 18:27 UTC (Tue) by obi (guest, #5784) [Link]
I think X.Y.Z (Major.Minor.Patch) is a good thing - it's more consistent with the stable/patch releases to always have 3 components.
Also, it's closer to Semantic Versioning ( http://semver.org ) - which I like because it brings some guidelines/sanity to versions and dependencies.
The real reason for the change... Posted Jun 1, 2011 3:13 UTC (Wed) by Alan_Hicks (subscriber, #20469) [Link]
...is Slackware 13.37 where we moved to a $distro_version.$kernel_version numbering scheme. This will allow us to release Slackware 14.0.
The real reason for the change... Posted Jun 1, 2011 10:59 UTC (Wed) by Felix.Braun (subscriber, #3032) [Link]
> Slackware 13.37 [...] moved to a $distro_version.$kernel_version numbering scheme
Wow! It had completely escaped me that this version number could actually be rationally explained that way! That makes it even more leet ;-)
The first 3.0 kernel prepatch is out Posted Jun 1, 2011 3:44 UTC (Wed) by Hausvib6 (guest, #70606) [Link]
I think the next very very very long term stable kernel version will be at least 3.1 because version-wise it:
* sounds brand new (3.y) * sounds polished (x.1) Commercial distros marketing dept will love it.
The first 3.0 kernel prepatch is out Posted Jun 1, 2011 5:03 UTC (Wed) by pr1268 (subscriber, #24648) [Link] Well, I suppose that, as a Slackware user, I'll have to wait four years to try Linux Kernel 3.x. ;-)
|
|||||||||||||