“The Dunc-Tank project has been the topic of much debate in the Debian community since it was launched in September last year. Aimed at overcoming Debian’s notorious delays in meeting its scheduled releases, Dunc-Tank collected donations to test the effect of funding on open-source software development. It has now been more than a month since the scheduled release of Debian 4.0, codenamed etch. However, even with Dunc-Tank’s funding, etch is yet to be seen. Liz Tay speaks with Debian Project Leader and Dunc-Tank mastermind Anthony Towns to find out what happened.”
Please don’t link to sites with such a sh*tty layout.
It’s just wasting our time.
The main problem with Debian is not (only) funding, but their too chaotic development process who make impossible to create any sort of reliable schedule.
Strangely, if this is a advantage or not, i don’t know. Gentoo apparently is quite chaotic too, but don’t appear to have any serious issues with this. And Debian ‘Unstable’ branch is in fact quite stable, enough even for some production systems.
“Gentoo apparently is quite chaotic too, but don’t appear to have any serious issues with this…”
With Gentoo it isn’t much of an issue of how chaotic they are because they are constantly upgrading the system and have “rolling releases” where your system is upgraded in place. You keep up with the latest as you want. There are no point releases per se.
So, chaotic or not, Gentoo isn’t affected by their development process as much as Debian is with their point releases.
“Gentoo apparently is quite chaotic too, but don’t appear to have any serious issues with this…”
With Gentoo it isn’t much of an issue of how chaotic they are because they are constantly upgrading the system and have “rolling releases” where your system is upgraded in place. You keep up with the latest as you want. There are no point releases per se.
So, chaotic or not, Gentoo isn’t affected by their development process as much as Debian is with their point releases.
You’re forgetting that Gentoo also make releases that can be put on a LiveCD about twice a year (currently, I believe the relevant release is 2006.1, which as they start from .0 is the second ’06 release). It’s only after you install the system that the “rolling release” comes into play – which as I understand it happens with debian-testing and -unstable too. For example, I started with 2004.1, but I’m downloading a new portage tree right now. However, I’ve no idea how much it’s changed from 2006.1 since I never used that install medium, and it’s quite possible that by the time I have finished emerging what I want from my new portage tree, I will have some stuff still around from 2004.1 alongside the latest stuff.
I suspect a lot of the differences between Gentoo and Debian emerge (if you’ll excuse the pun) because Gentoo, though it has a large number of packages, has a smaller number than Debian. Also, Debian’s releases are composed of (potentially) all of the software available for Debian at the time of release, whereas the packages on Gentoo’s prebuilt package CD represent only a subset of everything available.
Just my 2 cents, but it’s probably too early to tell whether Dunc-Tank is a success or failure. So far as I can tell as an ordinary user, Dunc-Tank is not “Aimed at overcoming Debian’s notorious delays in meeting its scheduled releases”, as the story summary has it. It is aimed at providing funding for Debian projects more generally as and when the need arises.
If you read the article, you’ll see that Dunc-Tank has had a pretty rough ride. A small but vocal minority have tried deliberately to sabotage it, and setting it up has proved tricky as well (legal paperwork, etc.).
Even so, I think it’s very important that Dunc-Tank does work out, for two reasons. First, because as the article makes clear, a few folks involved in producing Etch were overworked and exhausted. Overwork and exhaustion are bad news. These folks had every right to something better than that. Indeed, you could argue that Debian would have been irresponsible had it not tried to work something out on their behalf.
Second, because if every time a new initiative is tried at Debian a few disssenters are able to tear it down, Debian has little to look forward to beyond stagnation and low morale.
“Hire good people and pay them what they’re worth” is common sense to most folks, and if special circumstances arise this surely applies to the FOSS world too. No one wants another soap opera of the Sarge release variety.
FWIW, I am typing this on Debian Etch using Gnome. Still some RC bugs around I gather, but it is very fast, creamy smooth and altogether wonderful.
I think the problem has more to do with the fact that they were counting on one guy with a demanding day job to coordinate a release of the largest software project on the planet. I wonder if this guy’s employer has any idea what he’s up against.
“Sorry I’m late, boss… Millions of people were waiting for me to fix the last-minute bugs in 10GB of software for 11 architectures, so it was a long night.”
I have an idea. Why doesn’t Debian become the hackers’ proto-distribution is was always destined to be? Forget about releases, just keep packaging upstream sources and propagating them through the three branches as they stabilize. Let downstream projects freeze the release snapshots, update the GUI installers, roll the LiveCDs, and host the user forums. Make it easy for downstream projects to contribute bug reports, updates, and fixes. Let them decide what to do about firmware blobs.
I would guess that for most Debian hackers, making releases is nothing but a distraction. They boot the minimal iso (or even an Ubuntu LiveCD) and use dpkg to install the current state of whatever branch. They don’t need a release snapshot or really even a GUI installer. They just want to install stable, pull in a package from testing, test it, and bump it to stable. They can do this in a chroot within their polished Ubuntu desktop without harming the sustainability of the Debian project.
I don’t think Debian developers volunteer because they like the installer or the release management (at least I hope so). They choose to contribute to Debian because it is arguably the universal free software distribution (running on a variety of kernels and a variety of architectures). It’s not an innovator, it’s an incubator. It represents ultimate freedom.
It’s a project that should–and could–be structured as a horde of developers working in parallel on a distributed workload. But instead it all comes down to one professional PHP developer who, on nights and weekends, must get the stable branch in just the right state to warrant rolling a ~130-CD release.
“I have an idea. Why doesn’t Debian become the hackers’ proto-distribution is was always destined to be? Forget about releases, just keep packaging upstream sources and propagating them … a horde of developers working in parallel on a distributed workload. But instead it all comes down to one professional PHP developer.”
Lol, call in the management consultants! Maybe the project leader’s room for manoeuvre is constrained and he was simply trying to do the best he could in the circumstances? Just guessing. If this has thrown up “management issues” then they’ll need addressing, I guess.
In the meantime, I hope Debian doesn’t become a proto-distro for hackers. Partly for selfish reasons, as it wouldn’t be so pleasant and easy for someone like me to use. Partly because Debian needs to be present and visible, imho, or folks will lost touch with it and, just as important, the DDs will risk losing touch with their users. The lure of the ivory tower and all that. And, of course, Debian would then forego the huge audience for Debian stable among server users and institutions. Interesting idea, though.
Personally, I wish they’d consider merging unstable and testing, and then put in a turbo-charged tasksel that made setting up a nice desktop edition relatively easy. Perhaps with a couple of packages that set up a good desktop look and feel and pull in all the multimedia stuff. That would “let the cat out of the bag” as they say wrt Ubuntu.
I think that stable releases are quite important for Debian because they help to keep the number of bugs low and the level of quality high. Without releases there probably wouldn’t be much incentive for the devs to squash those hard-to-diagnose and difficult-to-fix bugs that nevertheless can cause a lot of trouble if they’re left unsolved.
Although the release process tends to be slow and painful, people usually feel much better after the release because they get a healthier and more bug-free Debian as the result of this process.
I don’t think that one can do project management when everyone on the project is her/his own boss.
DG
Maybe Debian should follow Mozilla’s lead and set up a privately-owned, taxable subsidiary called Debian Corporation. It will take care of all financial affairs, guard the Debian trademark, and redirect all money back into the Debian project’s schemes.
Also, true to the Debian mantra, it will never go public, never compete on the stock market, and never be held in obligation to any external company. It’s main goal will never be about making money or accruing wealth for itself, but for the Debian project.
Of course, it would be an irony for Debian to go for Mozilla’s model, since Debian and Mozilla had that tizzy over Firefox patching. However, as I heard back then, Debian takes the same legal approach towards its trademark.
Let me preface my statement by saying that I absolutely love Debian.
I think the problem with Debian is all the ego and posturing involved in the project. If they could focus on programming and leave the bickering to somebody else, they’d move along a lot faster.
With regards to “unstable”, I think Debian would do well to change their “stable”, “testing”, “unstable” conventions and focus on creating a good workstation (or desktop) os and then let those features/programs trickle down into their server release (which would equate to stable). I think a lot of people think Debian is behind the times because they think of “unstable” as an alpha or early beta release when it reality, it is as stable as most other “release” distributions.
I’m half asleep. I hope some of that made sense.
You’re probably half asleep, so …
I do really love DEbian, but you cannot seriously compare Unstable to Fedora Core XYZ. Sometimes Unstable is utterly broken for a couple of days – that’s certainly not the case with Fedora or SUSE.
So, yes: Unstable is pretty good – but don’t compare it to other Linux distributions on the market. Unstable is – not always, but often enough – unstable. At times it is quite stable, but only at times …
If I were to donate money to Debian development, what I’d want would be a menu of categorized tasks on the Debian site. Click a task, and it shows you the developer who wants to work on that task while getting paid donations. Then click to donate. There should also be some way to track the dev’s progress (a blog maybe) while also displaying how much is being donated to them (so you could get an idea if it’s money well spent).
This way, the *users* could decide what tasks they think are important, and the developers working on those tasks could benefit.
Edited 2007-01-27 07:21
I think that is an excellent idea as long as the individual projects are given some space to introduce themselves, their importance and their future tasks. I think also there should be an option to help funding individual packages/developers/maintainers or to chose to fund whole suites/branches where the organisation internally distribute the funding among the members (such as “desktop”, “server”, “package-system”, “games”, “scientific” etc). I’d love to help out this way.
Debian incorporates almost the same development model like BSD, therefore you have real quality. But they have one problem, the mix it up with faith. But faith and engineering isn’t something to mix up in reality. It’s of course part of it, but sometimes you have to draw a line to go on with high quality development.
>Until then a man with an experience is never at the mercy of a man with an argument…:-)
False, they are going for quality instead of quantity. And people should prefer an os who cares about quality instead following the hype of some gurus.
Well, it seems obvious to me that the intended goal (release Etch December 2006) of Dunc-Tank was missed. That’s a failure in anyone’s book.
The hard-ass behaviour of some DDs had probably a good deal to do with this, but at least they found more bugs. Who knows if they would have gone the extra mile before Dunc-Tank? The bottom-line is, Dunc-Tank and opponents have helped to improve the upcoming release. In sports we call this “lousy but useful”, poor execution, usable result.
Wether the general setup of the Debian project has to be discussed in this light is another question, that has been asked and discussed many times before and will be in the future.
“Well, it seems obvious to me that the intended goal (release Etch December 2006) of Dunc-Tank was missed. That’s a failure in anyone’s book.
The hard-ass behaviour of some DDs had probably a good deal to do with this”
The public impression that “disgruntled Debian developers” have managed to delay the Etch release seems to originate from Steven J. Vaughan-Nichols’ opinion piece (http://www.linux-watch.com/news/NS3128387759.html ). Unfortunately his opinion is quite uninformed. SJVN obviously doesn’t use Debian and apparently he doesn’t like any of the non-profit GNU/Linux distributions. He seems to have no idea how the Debian release process works and it looks like he didn’t even bother to read the release updates at debian-devel-announce before publishing his misguided opinions.
http://lists.debian.org/debian-devel-announce/
In this interview Anthony Towns points out the actual problem that has been blocking the Etch release: the debian-installer team needs to make yet another release candidate and they had to wait for the kernel team to prepare a new kernel package for that purpose. Those “disgruntled developers” had nothing to do with this delay.
Towns also says that the long-awaited kernel package got finally done on January 5th this year. (http://lists.debian.org/debian-kernel/2007/01/msg00152.html ) The second release candidate for the debian-installer is not yet ready and when it is, it will take a couple more weeks to test it before the final release can be made. Plus all the release critical bugs need to get fixed prior to release, of course.
It looks like Dunc-Tank has helped to bring the number of release critical bugs down, but this didn’t help to get Etch out on time because both the kernel and the installer fell behind the schedule. Maybe Dunc-Tank would have been more successful if they’d been more flexible and guided the money to the most critical show-stoppers? Maybe they should have directed some of the money to kernel packagers and installer developers, in addition to release managers?
But who knows what the actual release blockers for Etch+1 (Lenny) will be? They may not be the kernel and the installer, they might be something else altogether…
“””Unfortunately his opinion is quite uninformed. SJVN obviously doesn’t use Debian and apparently he doesn’t like any of the non-profit GNU/Linux distributions.”””
Quite the contrary. My perception, and I follow SJVN’s writings pretty closely, is that he is angry with the Debian devs for acting like such numb skulls over the last few year, to the detriment of the distro itself.
Most of the “crises” that Debian has faced have been issues that only a Debian developer could possibly understand, let alone perceive as a problem.
And yet, the real crisis that Debian faces is not even acknowledged internally. Namely that the ratio of the relative energies being devoted to bickering on the lists, and to getting actual work done is spiraling out of control.
Personally, I think Dunc-tank funds would have been better spent on mental health services and one on one counseling for some of the developers.
That’s not a joke. Not a troll. I’m quite serious.
Debian is *far* from my favorite distro. And the devs are *far* from my favorite group of people.
But Debian, the distro, is absolutely critical to the community. Half our ecosystem depends upon a healthy Debian.
My perception, and I follow SJVN’s writings pretty closely, is that he is angry with the Debian devs for acting like such numb skulls over the last few year, to the detriment of the distro itself. […] And yet, the real crisis that Debian faces is not even acknowledged internally. Namely that the ratio of the relative energies being devoted to bickering on the lists, and to getting actual work done is spiraling out of control.
Yes, I can believe that you have read SJVN’s writings — what you state here echoes another opinion piece by SJVN, where he speculates that Debian “is a group where far too many of its people are far more concerned with moronic minutiae than they are with development.” He also suggests that Debian is on its death-bed, apparently because all the development in Debian has been stalled and developers just concentrate on fighting with each other. That’s at least the general impression that he tries to convey.
http://www.linux-watch.com/news/NS7543606709.html
Had he actually used Debian, he’d know that at that time Debian “testing” was receiving a continuous flow of program upgrades (currently “testing” is frozen prior to release and it only receives bug-fixes). Also a quick look at the Debian mailing lists would have quickly shown him that 99,99% of the posts there concentrate on technical matters instead of bickering between developers.
But SJVN didn’t bother to check his facts. Had he done so, he would have quickly realized that he had no reason whatsoever to worry that Debian is dying.
Personally, I think Dunc-tank funds would have been better spent on mental health services and one on one counseling for some of the developers.
That’s not a joke. Not a troll. I’m quite serious.
Have a nice day, troll. ๐
Have a nice day, troll. ๐
Thanks for the note of cheer. It’s all grey and snowy here. ๐
I’ll admit that SJVN occasionally goes over the top.
But not often, really.
I seriously doubt your claim that only 1 in 10,000 posts on the Debian lists could be considered bickering. You should have laid off hitting that “9” key so many times.
Like about 3 times. ๐
“That’s not a joke. Not a troll. I’m quite serious.
Have a nice day, troll. ;-)”
I use Debian and I subscribe to several of their mailing lists as well as check Planet Debian every now and then.
SJVN may have been over-egging things, because all journalists do, but imho he is fundamentally correct in his central contention that “online politics” and “infighting” are a serious and ongoing problem at Debian. There’s not enough feeling I can find of a healthy esprit de corps, in the sense that becoming a DD is a privilege that also entails responsibilites, like, bluntly, behaving like an adult. Becoming an official developer on any project is not just a flashy badge which says “I’m l33t and can do what I like”.
Just my 2 cents, but I think your post is rather dismissive and I certainly don’t think it reasonable to call the other guy a troll. Reports of Debian’s death may well be greatly exaggerated, but when these reports arise it is usually because more antics from some immature characters among the DDs have provoked them.
SJVN may have been over-egging things, because all journalists do, but imho he is fundamentally correct in his central contention that “online politics” and “infighting” are a serious and ongoing problem at Debian.
Sorry, I disagree about the seriousness of the problem.
SJVN says that Debian is in trouble, that Debian is dying. And the only proof to these statements is his observation that there are occasional disagreements and heated debates at the mailing lists. But SJVN’s reasoning is invalid and his conclusions are absurd.
This reasoning seems to go like this: Arguing takes time and it has been proven that sometimes Debian developers argue with each other. Hence, the developers clearly spend all their time arguing, which probably means that they have no time for the actual coding. Therefore, Debian is clearly dying because the developers don’t have any time to write code.
But this is stupid. Public discussions are the normal method of making decisions in Debian and disagreements are not really as dangerous as you seem to think. It is only when some developers don’t respect the decisions that have been made together after this open discussion that you can start calling those rebellious developers immature.
Not all Debian developers participate in online discussions — a majority prefers to concentrate on producing code. And not all developers who participate in discussions bother to argue with others. And only a very small minority of Debian developers continue arguing after the public discussions have ended, votes have been cast, and decisions have been made.
If you read the mailing lists, then you should know that this is not such a problem that journalists like SJVN make it seem. The discussions there concern mostly technical matters.
da_Chicken,
Some of us disagree with you and think that the Debian infighting *has* become a real problem to us all.
I’m truly sorry that it offends you.
But the fact that we are concerned is a bit beyond your control.
I doubt that anyone thinks that Debian is going to die.
But even an ill, yet not dead, Debian holds dire consequences for the community in general.
We’re not on opposing sides, da_Chicken.
We are on the *same* side, fundamentally. Even if we don’t happen to agree upon this or that.
Edited 2007-01-27 20:59