Linus on Kernel Version Numbering 416
walshy007 writes "In a recent thread it was asked what it would take for an 'unstable' 2.7 development tree to be created, to which Linus replied:
'Nothing. I'm not going back to the old model. The new model is so much better that it's not even worth entertaining as a theory to go back. That said, I _am_ considering changing just the numbering. Not to go back to the old model, but because a constantly increasing minor number leads to big numbers. I'm not all that thrilled with "26" as a number: it's hard to remember. I think the time-based releases (ie the "2 weeks of merge window until -rc1, followed by roughly two months of stabilization") has been so successful that I'd prefer to skip the version numbering model too. We don't do releases based on "features" any more, so why should we do version _numbering_ based on "features"?'"
Re:Excellent notion (Score:5, Informative)
An [microsoft.com] excellent [microsoft.com] point [microsoft.com], sir [microsoft.com].
Why versioning based on features? (Score:3, Informative)
Well, at least with the kernel, it gave me an idea of whether I ought to expect a program to run with few problems, require recompiling to work with few problems or require porting and compiling to work with few problems.
Re:A suggestion (Score:5, Informative)
Why bother with Linux?, get a proper OS. You know, one that doesn't make you create everything yourself, and hide stuff with obscure names in obscure locations, unique for the developer who shat it out.
BeOS died long ago. And AmigaOS runs on specialized hardware. And OS X is UNIX-based so it does the "hide stuff with obscure names in obscure locations" and runs offically only on specialized hardware. And don't even get me started with Windows... So basically, all competition for OSes died after Windows 95. So either you get a UNIX-like OS such as Linux, or you get Windows.
Re:A suggestion (Score:5, Informative)
Besides, it's not like they don't want to support all the hardware available, it's win-only hardware manufacturers that are the main obstacle towards better hardware support in linux.
Re:Linus... (Score:2, Informative)
that's also why there are no drivers...
*no* drivers? This is +5 insightful? WTF?
Linux has a very clear advantage in drivers for older equiptment (e.g. printers and scanners like mine - 10 years old, work perfectly in Linux, support dropped in Vista).
It takes a bit of time to get support for new hardware sometimes and wiFi support is patchy (but improving) for well known reasons. But the sheer volume of Linux drivers for hardware of all ages beats any particular Windows version easily.
Re:Linus... (Score:5, Informative)
This "no drivers" myth continues to be quite pervasive.
There are lots of drivers, as long as you don't want to install something non-mainstream and you're okay with the binary blob drivers for the NVIDIA and ATI/AMD graphics cards (although, my understanding is that the ATI/AMD front is changing and AMD is pushing the specs out the community now).
There are very good drivers for hardware in just about any class. Scanners, printers, digital cameras, webcams, video capture, bluetooth, USB, you name it.
The trick is that you have to buy hardware that is known to work well and be supported on Linux. You might have to buy stuff that's a bit behind, too. Here's an example: the Epson Stylus C120 has a release date of August 2007. The Gutenprint driver for the C120 just appeared within this last month or so in 5.2 Beta releases (I think it's been available in the CVS for sometime). That means distros that keep up like Ubuntu will probably start supporting it in their next releases.
So you had to a wait a year. Big deal. In that year, the list price dropped from $89.99 to $69.99.
If you're one of those people that just HAS to have the latest hardware NOW, you're probably a gamer and should use Windows anyhow.
Re:Excellent notion (Score:3, Informative)
Well actually, 2.6.26 is "Rotary Wombat", and 2.6.25 was "Funky Weasel is Jiggy wit it". Not quite Ubuntu's codenames, and nowhere near as publicized, but equally strange.
Re:customers rely on "features" (Score:3, Informative)
Unfortunately, the alphabetical naming scheme for Ubuntu "only" goes back about two years. Before that, they were more random. For example, which came first, Hardy Heron, or Hoary Hedgehog?
I was wondering how long it would be before somebody mentioned Ubuntu, though, because Ubuntu already uses a date based version number scheme. The current version is 8.04 LTS, released in April, 2008. The other versions released in the last two years are 7.10, 7.04, 6.10, and 6.06 LTS. The date based version has another advantage in that it is easy to see whether a given release is still supported. The LTS (Long Term Support) versions are supported for 3 years after release (5 years for server edition), and the rest are supported for a year and a half. Based on that, we can tell that out of the above versions, 6.10 is no longer supported.
Re:Excellent notion (Score:5, Informative)
Major and minor numbers have their place, too. They tell me something about the amount of change. I'll update from 2.6.25 to 2.6.26 without a second thought, as I expect nothing important to have changed. I'll spend a few minutes on the Changelog when I go from 2.6 to 2.7 because I expect a couple of minor things to have changed. I know that going from 2 to 3 will be a major update and might result in all kinds of incompatabilities, so I'd better make sure all my apps are ready first.
Kernel development no longer works that way--the current model is, every 2-3 months a new kernel is released, then there's a couple weeks to merge new work, then the rest of the time until the next release is spent tracking down regressions.
There's never going to be another long-lived development branch: having years where all the real development went on in a kernel that nobody actually used caused all sorts of problems: bugs would pile up because the development branch wasn't getting enough testing, and distro's had to backport a lot just to be able to distribute "stable" kernels with the features and hardware support their customers needed.
So the kernel version is *always* going to start with "2.6.". Hence the thought that maybe the version numbering doesn't explain the new process as well as something like "2008.07" might.
And as for incompatibilities, they shouldn't happen. You should be able to drop a new kernel into an old system and everything should work--if not, report it as a bug. There's a few exceptions where some interface is dropped or change, but normally the assumption is that it's something that won't cause a problem for people--so if it does, speak up, they need to hear from you.
(Of course, the above only applies to real userland interfaces, not to internal kernel interfaces. If you're trying to run a bunch of proprietary out-of-tree code (like the proprietary nVidia driver) inside the kernel, then you're on your own.)
Re:A suggestion (Score:4, Informative)
If Windows had a kernel numbered 1.33.7
Funnily enough, the build number of Windows XP is 2600 [wikipedia.org].
Re:It also reflects... (Score:4, Informative)
In the good old days, I could depend on 2.5 being bleeding edge and 2.4 being ready for production. Since 2.6, I either depend on a distro to track the kernel versions(which I do at home), or watch the kernel forums to see which should be the next "stable" kernel version (which we do at work).
Haha, that's not what happened at all. The vendors forked 2.4 and backported all the fancy new 2.5 features. So most users were running something that was nothing like the 'official' kernel. Meanwhile 2.5/2.6 went into dev hell because it wasn't relevant for the people paid to develop Linux.
Odd/Even was a nice simple system in theory, in practice it didn't work out so great.
Re:YYYY.DDD (Score:3, Informative)
Ok, but comparing two numbers in X.Y.ZZ form is much, much quicker than reading the release notes for every version change from 2007.047 to 2008.198. Plus you wouldn't have a way to figure out the approximate number of releases between the two versions.
Generally the point of version numbers is to know which is newer, or to be able to provide a range for compatibility ("my program FooBar is compatible with 2.6.20 or newer"). Switching to a date-based system does not assist with either need.
Re:I like version numbers that tell me something (Score:2, Informative)
Well that is to some extent the point Linus was trying to make. How many people that use Linux currently have ever used a 1.x release? Therefore the 2 in 2.x.x is irrelevant. They're going to stick with the .6 branch in Linux now, so the 2.6 in 2.6.x is also irrelevant. Remember that "major changes" are going to happen in the 2.6 branch as well now, so you would still need to know that the big change happened between 2.6.14 and 2.6.20. So really what we have is Linux 26 or what have you. If we're going for incremental numbers then using a date is as good as some arbitrary number. Although when all is said and done I would actually prefer just to use Linux 26.x and just go from there. Nearly all of us use whatever the distro tells us to use, and really all you need to know is where your kernel lands in the age/stability/cutting edge time line. It's good enough for emacs right?
Re:Excellent notion (Score:3, Informative)
Oh please! Nobody in a position to be making decisions based on kernel version should be put off by a number like 2.6.26. Can you honestly see a PHB in ever being allowed by the developers to be in a position to say, "No, I don't think we should go with this kernel, lets go with another one."
I've already seen those sorts of decisions made on multiple occasions. In fact, I'd be surprised if it *hadn't* happened at anything except the smallest businesses.
Never underestimate the decisions that will be made by a egomaniacal micromanager who doesn't trust his staff.
Re:A suggestion (Score:3, Informative)
But Windows NT kernel does not have version numberin, (for developers there must be) but OS has the stupid NT x.y numbering and marketing names XP or Vista. So user does not know that XP is NT 5.1 and Vista is NT6 and next one is NT7 (Windows 7) and so on.
C:\>ver
Microsoft Windows [Version 5.2.3790]
C:\>
Re:A suggestion (Score:2, Informative)
Well actually, Windows has version numbers too but it is not shown to the user unless you ask for it (eg. winver), but to be fair, the version of the linux kernel is usually not shown to the user either unless you ask for it (eg. uname).
Where version numbers tend to be relevant is usually related to drivers and hardware compatibility. This is also the case with the linux kernel, especially if you are dealing with the vanilla one.
It is useful for users to have some sort of numbering to also report bugs. I would hate to have to provide support when it is impossible to determine the environment which is causing a problem.
Re:Another Suggestion (Score:2, Informative)
Actually it seems like Vista now implements something called XPS for printing, the GDI path is now legacy.
Much to my disappointment they've published the specifications and made them available royalty free, rather than patenting a few vital bits and releasing the rest to printer manufacturers under NDA. Pussies.