Slashdot Mirror
Debian May 1 Release Delayed
andrew writes "Anthony Towns, Debian's Release Manager, posted this message regarding the status of the expected May 1st release of Woody made reference to in this slashdot story. In short, he says: "So, it's April 30th (for most of the planet, anyway), which probably means folks are beginning to get mildly curious about whether woody'll actually be ready for release tomorrow. The answer is a definite 'kind-of'. Which is to say, 'no'.""
How do you delay May 1st?
The only surefire protection against Microsoft infections is abstinence. - The Onion
Yes, woody will become the new stable. Sid will remain unstable, and there'll be a new testing tree which hasn't been officially named yet (though the suggestion of "sarge" seems to have stuck in many developer's heads).
Delay of a release date is always a terrible thing, especially for the poor release manager, who, in this case, sounds like things got a little out of his control. Perhaps it's the peril of working on free software, and having volunteers instead of cubicle drones.
;p
Of course, the delay will net the Linux community something positive - a better Debian. Well, maybe not for the l33t d00ds out there who can take charge, and manually bonk around and get all their own security updates... but for the sysadmins, and the desktop supporting IT people.
What I'm wondering is why games are often the most delayed. If anything, a patch to a game won't be the most terrible thing you could do. But Neverwinter Nights, Duke Nukem Forever, oh, and that steaming John Romero pile... Every Blizzard game ever made! Hmmmm. Maybe they don't want us to have so much fun too fast.
If you would have read all of the article, you would have noticed that they were also delaying the release because they didn't have their procedures ready yet to release security updates on woody and potato at the same time. I think we should applaud them for remembering how important security is (Maybe msoft should take note?).
Currently there are 47 release-critical bugs;
woody will presumably released when these bugs
are closed... so help debuging !
I'm expecting to see a lot of 'Debian sucks, it's out of date before it's even released', but I think this is a good thing. Releasing a distribution before it's ready can be disasterous (RedHat's gcc 2.96 anyone?)... I'd rather have a working, secure, stable distribution a few days later than have a highly experimental one with all sorts of hidden defects right now.
I never really expected woody to go on May the 1st but still am obviously disappointed. However, getting over my own selfish wish to have new toys to play with - this demonstrates why debian is good. The guys preparing it have to deal with the same problems every other distributor deals with, except they seem to be obsessive about not releasing shoddy work just to meet a deadline. Given the enormous pressure to release they must be under from the community I reckon that takes guts and they should all be commended for it. (Doesn't stop me being desperate for woody though does it? :-))
Carpe Daemon
... as iso images.
People already running Debian/Sparc or Debian/Alpha will be able to update using apt-get.
The only problem will be that you can't install new machines using woody (but you can install potato and then do a apt-get dist-upgrade to get a recent OS).
Jo
--
Hi! I'm the infamous
reading the tone of aj's message, he seems to be blaming various members indirectly for the delay. surely if he is the "woody release manager", he:
a) wouldn't have let these issues which have been known for months only crop up now.
b) should have known earlier than the day before to announce the delay.
so if you consider the delay of woody to be a failure, i wouldn't blame the anonymous (yet cited) individuals who checked in code late. i would blame the process that resulted in these events.
I don't agree with that, the Millenium Bug went off at the right time on a few systems, did prove unreliable as it was expected to work elsewhere too, but they did get it out (well done guys).
Those timestamped virii seem to do pretty well, announcement goes out before hand and it does work for some people, of course for others it gets broken by another piece of software that just seems designed to break it (which doesn't seem very good software practice to me).
And of course Unix, which is delivered EVERYTIME you install it as Jan 1 1970, which is very impressive, that means you install it BEFORE you need it, even if you decide to install it after you need it!
An Eye for an Eye will make the whole world blind - Gandhi
I'm wondering, does anyone have any record/memory of how late other debian releases were past the promised date?
evanchik.net
Alot of other distro release x86 first and Sparc/Whatever later on. Why can't Debian do that?
... not many people can even name that many to begin with.
Debian supports _11_ architectures - a few weeks ago a friend of mine dug up an old sun he had in his basement. We installed Woody. It works exactly the same as it does on my x86 machine, that's awesome.
In one of the last XFree stories, the Xfree maintainer mentioned that he will not treat non-x86 people like second class citizens. Now, I partially agree with you, I'm an x86-only person myself, but think about it, 11 architectures
Man has been building homes and temples for millennia, yet budgeting and estimation for these projects remains nowhere near an exact science. Hardly any project can come in on time and within the budget -- that's just the nature of business. I'm continually amazed at how people in our field seem to think that this "phenomenon" is unique to software development.
On the other hand, Debian and Mozilla are two projects that are always notoriously late.
The projects aren't late. The scheduled release dates were just early.
The masses are the crack whores of religion.
I remember waiting for potato. There was no announcement about the release date, but a daily statistics of release critical bugs was published on debian.org. Often you could read a sentence like this:The number of release critical bugs is on the rise again.
Really, it's only the people using ISO's that even care. And they don't even need it. It's easy enough to install a base minimum system using the current release, then change one line in your sources file, run apt-get dist-upgrade, and magically you're using Woody. I'd venture to say that most people currently running Debian did this exact thing. And those very people won't gain much, if anything, from an official release.
So who exactly are the great hordes who are out there demanding that this new, wonderful product be released? Do they even exist?
Last time (2-3 weeks ago) I tested the Woody ISO:s you could choose a couple of different kernel-images for installation, three of them were 2.2.X images and one was with 2.4.18. I installed with 2.4.18 and it worked like a charm. I don't think there's anything wrong with the ability to choose. If you install it with the default 2.2 kernel you can apt-get the 2.4.18 packages and have that kernel installed instead.
Quite frankly, I fail to see what's news about it. There has never been a formal announcement of a May 1st release deadline, just a message in which the release manager went out on a limb: "So, to go out on a limb: Debian 3.0 (codenamed woody) will release on May 1st, 2002. Actually, as always, it'll release when it's ready: if we find that the software doesn't meet our expectations on April 30th, you'll find me on the ground writhing in pain with leaves, bark and wood all over the place [1].
(2) Because its a rather frivalous reason. Alot of other distro release x86 first and Sparc/Whatever later on. Why can't Debian do that?[1] I'm going out on a limb, remember."
Because Debian doesn't treat non-x86 users as second class citizens, and because the developers already have enough versions (stable, testing, unstable) of their packages to worry about without different archs having different versions.
Oh well I'm sure they will get it worked out in due time - until then I'm sure more and more people will begin to think of Debian as a dead distribution rather than as an active one.
Debian's release is going to be dead alright. Dead stable that is, which is exactly the goal of a Debian release. Anyone who gets a woody from a daily fix of "latest and greatest" versions can run woody (testing), or unstable and doesn't have to care about releases. Releases are for folks who require stability.
They really don't have anybody to blame but themselves I mean they are the only ones shipping a distro that still uses the 2.2 kernel. There are sound reasons for shipping with a 2.2 kernel as the default kernel; check the archives for the debian-boot and debian-cd lists. In any case, 2.4 kernels are supported, just use the "bf2.4" flavour of the installation system.Yeah.
;)
Developers all over the world must be waiting with bated breath to release all their new cool features the week AFTER the Debian release
-- What do you need?
-- Gnus. Lots of Gnus.
For most purposes, Woody has been pretty stable for months. All this new date means is that "Woody" becomes the officially released "Stable" Debian distribution.
Debian is a little behind because they insist that all software be packaged and configured in a consistent way. It makes for a more stable and upgradeable system.
Debian has high quality standards, which contributes to these kinds of delays.
Trading off a few weeks of bleeding edge currency for stability seems well worth it to me.
This is good news. I've been a Debian fan for a while, but being able to point to this posting as a proof of just how serious they take security (serious enough to delay a release) will make it much easier for me to push Debian in my work environment.
Assorted stuff I do sometimes: Lemuria.org
The last thing I want to read about is debate over when it's time to "release the woody." That is just nasty, and there is no place for such filth on the Internet.
Karma: Good (despite my invention of the Karma: sig)
There have been several pieces of software which have been relased and then patched within weeks of the release... infact, wasnt XP one of those products?
As long as the delay is reasonable, and there are good reasons to delay (which I'm sure there are) then dont complain!
debian are doing us all a favour by not releasing something their note sure of quite yet
"Ten years from now, they could do it in a few seconds." -- The Racketeer of the Hellfire Club, 1993, Phrack 42
The first 90% of a project takes 90% of the time, the last 10% takes the other 90% of the time
Last time I tested 2.4.18 the USB support was severely broken (uhci doesn't work, usb-uhci lockups right when inserting the module (or the other way round)). It worked fine in 2.4.17... Another example: half a year ago the fresh installation of Mandrake took my data for a breakfast. The reason was USB combined with devfs, it was actually quite reproducible. Alan Cox said something along "we have a reason why we don't enable devfs in Red Hat" after I reported the problems on linux-kernel.
I don't know whether Debian also quality-assures the kernel and makes own patches as the commercial distros do. If they don't, their concerns about stability are quite valid.
The 2.2 default with the ability to choose seems like a wise solution right now.
I wouldn't be so sure of that. You could run testing rather than stable, for one thing. Also we've made significant improvements in the project's infrastructure (the stable/testing/unstable split with mostly automated propagation from unstable to testing; good autobuilders) which significantly increase the chances of "woody+1" being released within a much shorter timeframe than potato->woody (which also was lengthened by the legal and technical resolution of crypto-in-main).
If you read Anthony's message, woody is now decoupled from unstable -- essentially as if it were a stable release. Since the only thing the release is waiting on is rbuilder (as I read the message), I think people can start uploading new stuff to unstable again if they want.
Daniel
Hurry up and jump on the individualist bandwagon!
The problem is, many people HAVE to run testing because stable is too old. It's not a serious issue on my home PC, but I can imagine trying to get my boss to switch to Debian and telling him he has to use a "testing" distro because we need a web browser newer than mozilla M18. Or, worse, having to go to "unstable" because some security update didn't get into testing. If, in fact woody+1 is released faster, that will help a lot. Stable would still be old, but much more useable.
Slow down, cowboy! It has been 4 hours since you last posted. You must wait another few hours.
NOTE: THIS IS MY PERSONAL INTERPRETATION OF EVENTS AND NOT AN OFFICIAL STATEMENT ON BEHALF OF DEBIAN!
For people who didn't read or failed to comprehend Anthony's message, here are the relevant parts:
On the upside, woody itself is ready to be released. The only outstanding
changes that need to be made are the standard security fixes that need
to be made throughout the lifetime of stable anyway.
Unfortunately, that's exactly where we've dropped the ball: the security
team presently don't have the resources to handle security advisories
for woody.
...
the final automatic run of the testing scripts was today, and will
be reflected in the next mirror pulse. From this point, we'll have
manually approved security updates to some packages, and very little
else, until release.
This translates to the following: woody is now being treated as if it were a stable release. The only thing that it doesn't have at the moment is support from the security team.
The reason it is not being released as stable is that it is significantly harder for the security team to support than potato (due to almost-doubling the number of architectures), and "over the next week or so", technical solutions to this problem will be implemented. If you can live without this for a while (I don't know how long this will take to resolve, but it sounds like a few weeks is an upper bound), you can install woody now. Otherwise, you might want to wait a bit.
Daniel
Hurry up and jump on the individualist bandwagon!
some of the speed issues are rather maddening. Consider: I work very closely with the debian maintainer for
nano, in fact I'd say we are friends. He has done his best to get a particular nasty issue, in fact the problem was annoying enough that it required a fix upstream (on my end). But even though two official releases have gone by since the fix was put in upstream, it may not in fact end up in the first release of woody, four months later. I have used debian for probably 5 years now, but I have to wonder if source distros like gentoo have the right idea about making the user decide how to compile his or her package which severely cuts down the burden on the package maintainers. I guess it all comes back to how to balance the burden of upstream/package maintainer/end user...
v2sw7CUPhw5ln6pr5Pck4ma7u7LFw0m6g/l7Di5e6t5Ab6TH.
I honnestly don't mind it if Woody is a few weeks late from the ETA, especially if it's about making the build more consistant between all architectures and to ensure the security patches will be uploaded in a timely manner.
What I do mind is Woody being delayed, only a few weeks from when packages like KDE 3.0 and Gnome 2.0 would become stable enough for inclusion. Meanwhile, at the moment, Galeon and Mozilla don't build cleanly on all platforms, not to mention XFree86 4.2 ...yes, Branden explained that he must first smooth the process for all architectures and I agree with him, however...
What makes Debian support by makers of non-free packages so absent is because Debian stable distros are always 2 years behind everybody else, in terms of what version of glibc, XFree or kernel the stable distro is installing with. There are two solutions I can think of for that:
Otherwise, if we're gonna wait a few more weeks, we might as well give KDE 3.0 and Gnome 2.0 (not to mention XFree 4.2) enough time to slide from unstable to testing and be included with Woody. Nobody that needs Linux in a production environment can afford to wait 2 years for those to be released, at a time when they are just upgrading to Woody from their already much deprecated Potato. When it comes to that, the solution will be to crossgrade to Suse or Red Hat, if a desired package is not available the day Woody makes it to stable and becomes a priority upgrade on everyone's TO-DO list; Debian will be no more in yet a few production environments, if it looks like it's gonna be obsolete at birth again, the same way Potato was.
As for those who feel like saying Blah! Just point your APT sources to unstable, you'll always have the latest!, don't.
While testing is almost sufficiently stable for a production environment, it is a constantly moving target that would need to be upgraded every couple of days; this is simply impractical for a production environment, nobody has that much spare time on their hands at work.
Then unstable is, as its name implies, unstable; I've often had computers become partially incapacitated for a few days, because some new package was uploaded without its updated dependencies, making APT stop the upgrade process right after unpacking a few packages.
The solution to the perpetual Debian release lag is simple: release always, release often. Allowing new packages based upon existing libc or xlib to be released within the lifespan of a distro - not just bugfixes and security patches - is a must, at the very least.
Software is not supposed to be about how to work around a useability issue. - Ken Barber
Read the message from Anthony Towns. The only real problem is getting a mechanism in place to automatically build security updates for the 11 architectures supported by woody when the need arises. The architectures currently in potato are not a problem, just the additional 5 added since. The release will be delayed until this mechanism is in place.
This is a very sensible decision, and should be applauded.
Although Sid was a character in Toy Story, it also stands for Still In Development. Unstable will always keep the name Sid, because it is the one where new versions of actively developed software land first. You can read more about that in the Debian FAQ.
--
Promoting critical thinking since 1994.
I was hoping to celebrate two things on that day. Now I can only look forward to cheering the first year with my beloved. Life is so cruel! Damn you Debian for letting me down! *sobbing*
j/k
Why bother.
Meaningless misleading statistics on how many packages of each architecture are downloaded every day from gluck.debian.org (a main http.us.debian.org rotation server) are available here. Someday I'll get the rrdtool feed working.
this will result in worldwide protests.
police will have a hard time, calming down the debian-users.
join the revolution
Now I can only look forward to cheering the first year with my beloved.
You've had a Real Doll for a year now? How time flies. Post a review!
The reason why it works for conventional software projects is that the new people need time to tune in to the project, and existing people are tied up to handholding the newcomers. For Debian, this just does not apply.
You must not read debian-devel. New maintainers (and even people who are already official maintainers) ask basic, everyone-should-know-that-already questions there on a regular basis, tying up time for everyone else (even if it's just the time to discard the message). It's probably part of the reason that many people have unsubscribed from -devel. Not to mention the effort the rest of us have to put into reporting (and fixing!) bugs in their packages.
You could argue that it's less of an issue for Debian, but it's certainly not irrelevant.
Oh, and one more note: it was a member of the new-maintainer queue who was responsible for filing 80 frivolous release-critical bugs, simply because he didn't know better. (see debian-devel ) Bringing people up to speed involves educating them about the social aspects of Debian ("don't file tons of bugs at release time that we have to spend time ignoring") as well as the technical ("package maintainers can't change the priorities of their packages"), and people weem to have as much trouble grasping both.
Daniel
Hurry up and jump on the individualist bandwagon!
When Potato became the stable distribution, everyone said that they should have waited just a tiny bit longer, because Linus was supposed to release the new 2.4 kernel any day now. It turned out that the 2.4 kernel was delayed, and if Potato had waited for it, it would have been far too long.
/etc/apt/sources.conf file to read stable. Leave it at testing, and you'll get the new KDE and Gnome very soon.
There are thousands of packages in Debian. If any single package upgrade is not ready for the release, it's not fair to the other packages and their maintainers to make them wait. If you want to have more up to date stuff on your system, then when Woody becomes stable, don't change your
If tits were wings it'd be flying around.
On my Debian system, I don't install any kernel packages or headers at all. I get the source and compile it myself, in the same way that everyone else does (not the Debian way - compiling the source through the packager).
This works just fine. Woody has absolutely no problems with the 2.4 kernels, so if you want to go that route, you should be fine.
If tits were wings it'd be flying around.
I think this is a bit of equivocation.
The project manager definitely said that they were planning release of May 1. Call it unofficial or official or whatever, the statement was made by the person nominally in charge of the project.
I'm using Red Hat right now, and one reason (the main reason is that it actially installed smootly on my machine, unlike mandrake, slack or debian) is that I'm a little put off by the seat of the "seat-of-their-pants" way the project seems to be run.
I'm not trying to be insulting. I'm sure that the project is filled with fine programmers, but for users and potential users of the system, there seems to be a lack of accountability.
evanchik.net
In one of the last XFree stories, the Xfree maintainer mentioned that he will not treat non-x86 people like second class citizens.
Its more than that; from Branden's (Xfree86 maintainer) posting...
In woody, we support 11 architectures: alpha, arm, hppa, ia64, i386, m68k, mips, mipsel, powerpc, s390, and sparc. For how many of these machine architectures do Slackware, Mandrake, or Red Hat have 4.1.x, let alone 4.2, available? [emphasis mine] XFree86 themselves don't test or prepare distribution tarballs for several of these architectures. Debian is the de facto portability laboratory for XFree86 on Linux. Sure, I'll grant you that a lot of people, the kinds with the overclocked Pentium 4's and the latest GeForce card, really don't care about portability, or supporting architectures they've never heard of. But portability is important to me and it's important to Debian. I refuse to treat non-i386 users like second-class citizens.
Now that's class, and that's why I'm going to kiss a little backside and give all of the Debian developers/maintainers a big virtual pat on the back and say "thanks for all of the work you guys have done, both on Woody and in Debian in general."
Jay (=
(A perfectly happy Debian user who doesn't mind one whit that Woody will take a few more days...)
Yeah, but don't they have more people working on the packages as well?
BTW, thanks for the dates. They really help to put things into perspective.
I'm wondering if maybe debian ought to consider lessening the scope and ambitiousness of its releases, if only to keep them a little more current.
I now await the being repeatdly told that apt-get will solve all problems, including war in the Mid-East.
evanchik.net
someone *please* mod this up
evanchik.net
Well, your boss has to make a choice. He can have software that is old but extremely stable, or software that is modern but less stable. It is in the nature of software enginering that stability improves over time and with patches.
At least Debian gives him this choice. Install a shiny new RedHat if you prefer; just don't complain that it is not as solid as Debian Stable.
~~~~~ BigLig2? You mean there's another one of me?
experimental packages for debian and a short howto:
6 4
i'm using it right now actually. not the most stable thing in the world, but it's alright.
http://kde-look.org/content/show.php?content=14
my blog
Please, stop comparing RPMS to apt-get. Compare rpm to dpkg all you want, but apt-get is completely different. If you want apt-get functionality on Red Hat use up2date, if you want it on Mandrake use urpmi. These tools have existed in their respective distributions for several releases now and work just as well as apt-get.
Whenever a new release of Debian approaches, there are always people wanting it sooner and people wanting it later (so the very latest version of their favorite package can be included). Because Debian is released so infrequently, people fear they will be stuck with old software. As suggested, if Debian released every six months, like clockwork (and FreeBSD), then both parties would be satisified. There is always a new version just a few months away, so there is no need to worry if your favorite package didn't make it in this release. If Debian does not drop its "cathedral" developlment approach for shorter incremental developmenet.. well, we'll see ya'll in 2004 when a Debian stable released finally includes KDE 3.0, Gnome 2.0, and Mozilla 1.0.
cpeterso
Personally, the best way to put my mind in a fast deep hack is a quiet work environment with cool, slightly humid air, lots of natural colored light, and a feeling like I'm slightly isolated but able to interact with my coworkers. Poorly working, loud air conditioner units are bad. Flourescent lights are bad. A quarter of the flourescent light bulbs are worse. Cubes with only two walls are bad. Cubes with a second worker are worse. More than that and you've ruined the entire point of hiring more people. Don't bother putting a worker at an open desk against the wall when you run out of space. You'll get more bang for your buck by not spending it. Cubes with four walls, even if one is a half height wall where the entry opening is are better. The size of the cube is less relevant as long as its well lit, ventilated, and quiet.
And on a related note, snacks and drinks are good, too. Personally, I'd love to see free snacks and drinks available in the breakroom. (oh, yeah, having a breakroom's good too... it should be very well lit and ventilated and have non-tinted windows to the outside world) I know free snacks and drinks can be a problem, since it can get abused too easily, but what about a system which accounts for who got what? Maybe take that magstripe on the back of my id badge and allow me to swipe for free cokes & snacks. That way, if someone's taking an inordinately high amount of cokes, you can talk to them (or sign them up for coca-cola's anonymous).
Anyway, enough soapbox for now.
but 'woody'll' is not even remotely proper. You can't make it into a contraction. 'woddy will' would be the correct way to say it.
You must not read debian-devel.
:-)
Why are you forbidding me?
It's an idiom. "I take it you don't read debian-devel" is a synonym.
Anyway, I may have been a bit hard on NMs; I worked with one recently who was very competent, and some (well-known) long-time maintainers still ask basic technical questions on -devel.
But I do feel that adding new people adds to the general level of poorly designed packages, and to the level of ignorance about various bits of policy (a lot of new packages were leaving out menu files at one point, although that seems to be less of a problem now) Even if handholding on lists doesn't take time, cleaning up after these people does.
Daniel
Hurry up and jump on the individualist bandwagon!
So if the product's late, obviously the team is just on strike....
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
If the release is delayed for a short period of time, ( >1 month), I would agree that this is not a big issue.
But the delay does bring into relief the very long development lags in debian. Two years is a long time for a version upgrade. Take a look at any other distro two years ago, and they don't stack up against more recent distros.
On the other hand, the forthrightness of the explanation for the delay is pretty commendable.
But debian really should do something about making for frequent "Stable" upgrades. I believe this could be done without impacting security.
I suppose the biggest problem is debian's huge scope. Not only 10000 packages, but 11 platforms. The old sytem of development that started with far fewer platforms and packages is not scaling well with the project's increase in scope.
So either the process needs to change or the scope should be lessened. Security and stability should not be compromised.
evanchik.net
Led Zeppelin's "Physical Graffiti," a double LP that is deffinatley one of the best works of the band, was delayed something like 8 weeks for the cover. That most have sucked for the Zepp fans in 1975.
Here's the catch: It had the most expensive cover ever made to that date. And it was soo damn cool (an apartment building that had pieces of paper with differnt pictures in the windows of the building that you could slide around to change. On the front AND back)
I don't know if that's a good analogy, but... hey, anything for a good time!
Da comp cant tell u da emotional story.It can give u da exact mathematical design,but whatz missin is da eyebrows. -FZ
Why can't they speed things up a little by moving unstable to testing at the same time as they move testing to stable? ie -- there would always be the following three distros:
stable -- current stable release
testing -- almost feature-complete next release
unstable -- new features get added here
Then, as soon as testing is declared stable, unstable moves to testing, and a new unstable is created...
Maybe that would make things a little faster, as it's basically biting off less at a time... unless of course, they already do that...
rr
Quidquid latine dictum sit, altum videtur.
If you don't want to spend time fixing someone else's bug, use an alternative software package.
<SARCASM>Yeah, that's a great way to improve the quality of Debian.</SARCASM>
Daniel
Hurry up and jump on the individualist bandwagon!