Slashdot Mirror
Guessing Linux 2.6.0 Release Date
thorgil writes "Guessing about the linux-2.6.0 release date is hard, but here is a new angle (pseudo-scientific): I made a graph (gif) based on errors/warnings from John Cherry's (OSDL) compile statistics for linus' linux bitkeeper tree.
My guess is around 12th October, 2003. What is your guess and more important, why?"
is that you have way too much free time on your hands.
It's ironic that slashdot would run a story about linux today at all. But what really surprises me is that Slashdot would continue operation today, even though they allegedly support the Online Demonstration Against Software Patents.
/. staff to immediately shut down operations and support the
I would urge the
demonstration, unless they really don't care about open-source software at all.
Maybe a great open source businessmodel?
1) Do free stuff.
2) ?
3) Call your local bookeeper and gamble on kernel 2.6.0 release-date.
4) Profit!
To quote a famous game developer: "When it's done."
The question that really count is when will the first stable version of 2.6.x be out. I mean 2.6.35 or such...
But I don't think the "it compiles, let's ship it" is the criteria for releasing 2.6.0 A better way is to look at Andrew Mortons must-fix list. When most items are fixes, it can be released. ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/ must-fix/must-fix-6.txt
Should it be a linear best-fit? I'd be guessing that the number of errors/warnings will only approach zero? Much like tracking bugs.. On second thoughts, errors will more than likely hit zero but warnings we can live with.. :)
Anyway, interesting stuff
Oct, 12th is in about 6 weeks. So, because every IT project takes twice as long as you think, my guess is around Nov, 30th.
....Excuse me, but
For those of use that are running the 2.5/2.6beta kernel, what should we do when we do find bugs?
Since when do compile time errors and warnings reaching zero mean that there no more bugs in a program? Most bugs are those the compiler doesnt complain about.
What are you waiting for?
This should have been a poll. Now, it just leads to endless ramblings.
So now we can start keeping eye on the schedule !
I can already see the Washington Post's spaculative banner headline:
"Linux Major Update late for two weeks ! What's up ? Is Linus falling behind ?"
Give me my daily dose of SCO!
this really reminds me of the 'wheel of fish' ...
q> when is linux-2.6.0 released?
a> ummm.. blah blah
q> absolutely wronggg! youre so stupiiid!
class he-man extends man!
Fair enough. My guess for the release date?
CowboyNeal.
Endless ramblings?
But we wants our endless ramblings. This is slashdot.
Otherwise, I might have to do some actual work.
I don't have one
PROGRAMMERS DRINKING SONG:
...
99 little bugs in the code,
99 bugs in the code,
fix one bug, compile it again,
101 little bugs in the code.
101 little bugs in the code
(Repeat until BUGS = 0)
!
... according to your approach, MS Longhorn shall never be released. Your theory works real fine!
The Andrew Morton must-fix list dates back from May! No use...
I'm in a Unix state of mind.
you should stop using Debian-stable! ;-)
"Testing? What's that? If it compiles, it is good, if it boots up it is perfect." -- Linus Torvalds
So now we're guessing the release date based on when it will compile without errors, eh?
When the kernel itself is declared "released" is irrelevant to most people. If you really want the latest and greated, you can always download whatever the current version is, whatever it's called, and use it.
What's important is when most distro companies (other than bleedinge edge Gentoo and "we don't need no steenking 2.x kernels" Debian) will start building their distributions around 2.6-final instead of 2.4. For that, it's quite obvious at this point: The spring refresh cycle. (The fall cycle may have a few optional pre-release kernels, but the real action will be the spring.) Sometime in the April timeframe we'll see Red Had, Mandrake, and SuSE releasing 2.6-based versions. Hopefully they'll also have funness like KDE 3.2 and so on by then, which are just as important to most people.
When Linus says "ok, I'm done, let's work on something else" isn't important. When Red Hat says "we'll give you a support contract on this now", THAT'S important.
--GrouchoMarx
Card-carrying member of the EFF, FSF, and ACLU. Are you?
but DEC-25 is OCT-31, isn't it ?
Trolling using another account since 2005.
Maybe I do... Oh wait... Yeah I do...
Warning: This sig contains a small bug. ==> *
I don't think it will happen during the next 6 weeks, there are still some major things to be done. IDE still does not work as module (some circular dependencies in symbols), ISDN is still somewhere between the no longer working old model and the not yet complete CAPI support, just to name two.
seriously... that shit doesn't fly at most commercial software companies. why are people checking in code with compile warnings? why aren't they compiling the code and fixing the problems before checking in the code?
>> Does it still use the SCO licensed code in it ?
'still' would mean that it did in earlier versions. This has not been proven, and based on all of the information we have seen (as in code presented in that PowerPoint slideshow) does not lead to the conclusion that the 2.4 kernel contains any code that is 'owned' by SCO.
Because that's my birthday!
follow the documentation in the kernel source on how to submit bug reports.
It will be the 21st century..
(I want to die quietly in my sleep like my Grandad - not screaming in terror like the passengers on his plane..)
"You lied to me! There is a Swansea!"
"When it's done." I think this is a good "guess" for two reasons:
a) it's 100% accurate.
b) It didn't cost me precious hours of my life to come up with this answer.
I'll now continue to invest my time in more important stuff...like reading slashdot.
(Hey! They say it's "Stuff that matters!")
But then again, gifs and pngs are still better for line art. chunking the image into 8x8 blocks doesn't make sense for charts and low color images. In these scenarios, it's better to just use gif/png.
> Next bet: when will be 2.7 tree be opened?
;)
Perhaps after 2.6.0 stable is released?
If it takes a long time to get 2.6.0 stable developers will start to complaint that they are not allowed to make new features. Then there are 2 options: feature creep in the new 2.6 testseries (a new MM 8) ), or start a 2.7 tree. The thirth:
Hofstadters law:
"Everything takes longer than you expect, even when you take into account Hofstadters law."
Douglas Hofstadter, "Godel, Esher, Bach", ISBN: 0465026567
Think I'll be forgiven for not realizing that was today? I expected OSDN to shut down, but no...
You can't judge a book by the way it wears its hair.
just in time for revolution(s) ..
I fuse with Mercer every single day...
according to the FSF gif will be patented technology in at least one country (Canada) until Wednesday 7 July 2004.
... in all the vacation you mentioned, he is able to guess the correct 2.6 kernel release date.
".Sig Stealer" was here
How much influence has SCO on the developers, e.g. make them response to the SCO FUD instead of fixing bugs in the kernel? That's also a sort of "denial of service" attack.
Now when it's not released on October 12, the press will start writing about the "delayed" new version of Linux. Remember the pre 2.4 days?
didn't Linus said that 2.6 was being released when x86 code was stable?
And other archs maybe would have to wait some minor versions?
Considering this and the graph predictions, my guess is 3-4th week of September.
I'm a chainsmokin' alcoholic sociopath, so-ci-o-path
Rather than say that there must be 0 such compiler warnings/errors, I'd like to see what statistics would be produced if used on a 2.4/2.2 kernel. At least then we'll have something to compare the numbers to.
I think the RULE project might be able to help you.
http://www.rule-project.org/en/
They are creating a installer/distribution for Red Hat that is made for low-end hardware.
Go here to create your own Slashdot dis
"I'll take Linux Kernel release dates for 200"
Sanity is the trademark of a weak mind. -- Mark Harrold
I'm with you, though. I think Linux and its users would be much better off if the developers imposed a bit more process on themselves, and didn't rely so heavily on the "keep tweaking and releasing until it seems to be right" model.
It really sucked last year when my 21st birthday was also on the Monday of exams week. OSU's present to me? The toughest exam (EE) I've had there yet, at 7:30am nonetheless. But don't worry, after that, the rest of the week was a blur.
Berto
Not that I think it's the greatest language or anything, but my experience with ADA was that vastly more stupid things that programmers (in this case me) do show up as compile time errors. Almost to the point where if a program compiled it was bug free. Of course it's still possible to have a logical errors, but whole classes of what would be run time errors in C are compile time errors when ADA is properly used with things like range checking.
I'm going to guess it'll be delayed until the SCO code is publicly released so any problems can get cleaned up. Possibly in October we'll get a "gold level" release that would be the final kernal minus SCO fixes, that way if SCO loses it can just be rebranded without losing any worktime and if SCO wins the claim can always be made that as soon as the problem was shown it was fixed and never made it into the "final" kernal. Linus has been accused of being arrogant at times (mostly by folks with rejected code) but never accused of being stupid.
It doesn't matter what you wrap your emotions around, Reality is a brick wall specifically designed to scramble eggs
To quote a famous game developer: "When it's done."
Daikatana wasn't exactly the hit everyone was salivating over...
You can accomplish anything you set your mind to. The impossible just takes a little longer.
Just glancing at the graph, it doesn't look like a linear curve to me. Is a least squares fit the appropriate tool for this? It appears to me like it is leveling out.
2.6.0-test* seems solid enough for daily use, although if you have a laptop with a Synaptics touchpad, there seem to be some problems with the driver for that (I've not been able to get mine recognized but I prefer the USB mouse anyway.)
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
I "fell asleep" last night. And in that sleep I was visited by three ghosts of linux. Well to make a long good story short and not as good, they told me that if i didn't change my ways and always use linux for everything then tiny Tim Torvolds would DIE.
That and 2.6 would be released on September 21, 2003.
Well.. maybe. Or Maybe not. But Definitely not sort of.
You're right, just because it compiles doesn't mean it's done, but if it compiles successfully for a while it does mean that not too much new is going on and it might be ready for a release.
All measurements of this kind have inaccuracies. Do you have a better one? If so, then let's hear it.
USB is still a bit flaky, anything using USB devio could have problems, and ISO transfers don't work.
The correct answer is, when we think it's about ready.
the 2.4 kernel still isn't "done".
thank God the internet isn't a human right.
is that this is a slow news day.
Based on the 2.4 experience with the memory mapper changing horses well after 2.4.0, I'd be careful making predictions.
Also, Linus is now full time at OSDL (+).
Also, Alan will be going back to school (-). Good for him, though.
I'll go out on a limb, though, and say
which is evidently Finnish Independence Day."Provided by the management for your protection."
I'd bet you could do this on one of those online betting sites. I guess you could use that graph to calculate odds for betting.
I don't know enough about the Linux dev process to comment on it, but in software I've worked on there's typically time tacked onto the end of the schedule to do performance testing and tweaking.
- First they ignore you, then they laugh at you, then ???, then profit.
Take a close look. He uses Least Square, but really it looks more exponential, so probably the version will get very close to perfect, but they will NEVER release it
What Would SCO Do?
If he's ready by October 15th he should wait to release it then. That's doing to be a date of massive sucktitude. SCO raises their linux "prices" and Microsoft drops support for older MSN protocols and bans uncertified third-party clients. Also, it's my birthday and I want a good birthday present.
...all forms of online protest, including turning off your server to protest or boycott anything. If you shut off your server as a form of protest, then you owe me licensing royalties. If you run your server, well then you owe SCO royalties since they own all operating systems.
it should be either:
(Repeate until BUGS == 0)
('=' is an assignment operator, '==' is a comparison operator)
or:
while (BUGS > 0): sing(...)
I used to have a sig, but I set it free and it never came back.
That linear fit model is just plain wrong. Problem identification (and therefore problems fixed) are both non-linear in nature, for what should be obvious reasons. Three bugs, roughly three times the chance of finding one per unit time. Of course the kernel is divided into rough sections and those working on networking code are not likely to hack the vm, but this applies equally for each portion of the kernel.
d bugs / d time = bugs
(or if y = bugs and x = time)
dy = y dx
Solving this is left as an exercise for the reader.
I am setting up a market-based delphi poll to predict linux kernel releases. My partner, Admiral Poindexter, and I will be contacting a select number of you to participate in this endeavour. Please have $1000 ready to invest. BTW, if you have predictions on when the King of Jordan will be deposed or when the next terrorist strike will occur, we are interested in that as well. Thank you, and have a Total Information Awareness day.
My guess is that regardless of the state of 2.6.0, that RH X will ship with a 2.6 based kernel around Halloween. I base this on the fact that RH 9 already contains alot of 2.5 code which has been back-ported and the fact that RH usually issues a release around Halloween.
So, does the person who guesses closest receive a free copy of the 2.6 kernel?
cause its my birthday. As good a day as any.
Slashdot is corporate-owned. It's a business. It makes money. They're not going to shut down their business for a day when they could be posting more SCO, "Microsoft hole," anime, and amateur rocket stories.
"Sufferin' succotash."
The 2.4 kernels took 7 months to go from testing to 2.4.0. If the first 2.6 test kernel was in July, the first 2.6 kernel might be in Feb 2004.
Never mind the release date! When are we going to see the next SCO article?
This sig no verb.
NumDays = NumOpenBugs/(bugsClosed-bugsOpend)*time period
Very simple, and suprisingly effective for large projects where the bug finding and bug closing rate is very close to constant over a period of time (lets say a week)
So basically this says, that as long as your bug openning rate is higher than your bug closing rate, don't even bother predicting a ship date, but once you are closing more than you are openning - you can predict a date.
The nice thing about this is all though it is a rough metric, it tends to guess long - and it is something simple you can get any pointy hair boss to buy into (and even script it so that they can determine for themselves what the "ship day" is likely to be)
I have mod points and I am not afraid to use them
Anyone who's taken differential equations will tell you that when the rate of change of a quantity depends on itself, then the quantity will change exponentially.
So maybe we should really be talking about bug "half life?" How long it takes for the number of bugs to drop in half.
Yes, I think a lot of things have been fixed or fixes are underway and mostly working.
And there are also some problems nobody wants to attack...
Wasn't J. Pointdexter working on this problem over at DARPA? Creating a market of people betting on the next kernel release?
"I honestly would vote libertarian if their candidates weren't usually total cooks."--slashdot poster