rolling – Page 2

No freeze of Debian’s development, what does it entail?

April 28, 2011 by Raphaël Hertzog

The main feature of rolling is that it would never freeze. This is not without consequences.

Possible consequences

It can divert developers from working on the release

No freeze means developers are free to continue their work as usual in unstable. Will it be more difficult to release because some people will spend their time working on a new upstream version instead of fixing RC bugs in the version that is frozen? Would we lose the work of the people who do lots of NMU to help with the release?

It makes it more difficult to cherry-pick updates from unstable

No freeze also means that unstable is going to diverge sooner from testing and it will be more difficult to cherry-pick updates from unstable into testing. And the release team likes to cherry-pick updates that have been tested in unstable because updates that comes through testing-proposed-updates have often not been tested and need thus a more careful review.

Frozen earth

My responses to the objections

Those are the two major objections that we’ll have to respond to. Let’s try to analyze them a bit more.

It’s not testing vs rolling

On the first objection I would like to respond that we must not put “testing” and “rolling/unstable” in opposition. The fact that a contributor can’t do its work as usual in unstable does not mean that he will instead choose to work on fixing RC bugs in testing. Probably that some do, but in my experience we simply spend our time differently, either working more on non-Debian stuff or doing mostly hidden work that is then released in big batches at the start of the next cycle (which tends to create problems of its own).

I would also like to argue that by giving more exposure to rolling and encouraging developers to properly support their packages in rolling, it probably means that the overall state of rolling should become gradually better compared to what we’re currently used to with testing.

The objection that rolling would divert resources from getting testing in a releasable shape is difficult to prove and/or disprove. The best way to have some objective data would be to setup a questionnaire and to ask all maintainers. Any volunteer for that?

Unstable as a test-bed for RC bugfixes?

It’s true that unstable will quickly diverge from testing and that it will be more difficult to cherry-pick updates from unstable into testing. This cannot be refuted, it’s a downside given the current workflow of the release team.

But I wonder if the importance of this workflow is not overdone. The reason why they like to cherry-pick from unstable is because it gives them some confidence that the update has not caused other regressions and ensures that testing is improving.

But if they’re considering to cherry-pick an update, it’s because the current package in testing is plagued by an RC bug. Supposing that the updated package has introduced a regression, is it really better to keep the current RC bug compared to trading it for a new regression? It sure depends on the precise bugs involved so that’s why they prefer to know up-front about the regression instead of making a blind bet.

Given this, I think we should use testing-proposed-updates (tpu) as a test-bed for RC bug fixes. We should ask beta-testers to activate this repository and to file RC bugs for any regression. And instead of requiring a full review by a release manager for all uploads to testing-proposed-updates, uploads should be auto-accepted provided that they do not change the upstream version and that they do not add/remove binary packages. Other uploads would still need manual approval by the release managers.

On top of this, we can also add an infrastructure to encourage peer-reviews of t-p-u uploads so that reviews become more opportunistic instead of systematic. Positive reviews would help reduce the aging required in t-p-u before being accepted into testing.

This changes the balance by giving a bit more freedom to maintainers but still keeps the safety net that release managers need to have. It should also reduce the overall amount of work that the release team has to do.

Comments welcome

Do you see other important objections beside the two that I mentioned?

Do you have other ideas to overcome those objections?

What do you think of my responses? Does your experience infirm or confirm my point of view?

Towards Debian rolling: my own Debian CUT manifesto

April 27, 2011 by Raphaël Hertzog

As you might know, I’m of one the people who promote the idea of having a Constantly Usable Testing (CUT). I will post a set of articles on this topic to help me clarify my ideas and to get some early feedback of other Debian contributors. I plan to use the result of this process to kickstart a broader discussion on debian-devel (where I hope to get the release team involved).

Let’s start by defining my own objectives with Debian CUT. I won’t speak of the part of Debian CUT that plans to make regular snapshots of testing with installable media because that’s not where I will invest my time. I do hope other people will do it though. Instead I will concentrate on changes that would improve Debian testing for end-users. I consider that testing (in its current form) is largely usable already but there are two main obstacles to overcome.

Testing without freezes = rolling

The first one is that testing is no longer testing during freezes. The regular flow of new upstream versions—that makes testing so interesting for many users—is stalled because we’re using testing to finalize the next stable version. That’s why I’d like to introduce a new suite called “rolling” that would work like the current testing except that it never freezes. Testing would no longer be a permanent suite, it would only exist during the freeze and it would be branched off from rolling.

Rolling should be a supported distribution

The second obstacle is not really technical. We must advertise rolling as a distribution that ordinary people can use but we should make it clear that it’s never going to have the same level of polish that a stable release can have. And to back up this assertion, we must empower Debian developers to be able to provide good support of their software in rolling, this probably means using “rolling-proposed-updates” more often to push fixes and security updates when the natural flow of updates is blocked by transitions or other problems. Some maintainers won’t have the time required to provide the same level of support as they currently do for a stable release and that’s okay, it’s not worse than the current testing. The goal is to treat it on a best-effort basis and to try to gradually improve the situation.

My goal for wheezy

This is the the minimal implementation of CUT that I would like to target in the wheezy timeframe. Having testing as a temporary branch of rolling is not a strict requirement, although I’d argue that it’s important to not waste resources towards maintaining two separate yet very similar distributions.

Should Debian embrace those goals?

Ignoring all possible problems that will surface while trying to implement those goals, can we agree that Debian should embrace those goals because it means providing a better service to a class of users that is not satisfied by the current stable release?

I will come back to the expected problems in a subsequent post and we will have the opportunity to discuss them. Here I just want to see whether we can have broad buy-in on the principles behind Debian rolling and CUT.

Can Debian offer a Constantly Usable Testing distribution?

October 4, 2010 by Raphaël Hertzog

Debian’s “testing” distribution is where Debian developers prepare the next stable distribution. While this is still its main purpose, many users have adopted this version of Debian because it offers them a good trade-off between stability and freshness. But there are downsides to using this distribution and the “Constantly Usable Testing” (CUT) project aims to resolve those. This article will present the project and the challenges involved to make it happen.

About Debian unstable & testing

Debian unstable is the distribution where developers upload new versions of their packages. It happens frequently that some packages are not installable due to changes in other packages or due to transitions not yet completed.

Debian testing, on the contrary, is managed by a tool that ensures the consistency of the whole distribution: it picks updates from unstable only if the package has been enough tested (10 days usually), if it’s free of new release-critical bugs, if it’s available on all supported architectures, and if it doesn’t break any other package already present in testing. The Release Team (RT) controls this tool and provide “hints” to help it find a set of packages that can flow from unstable to testing.

Those rules also ensure that the packages that flow into testing are reasonably free of show-stopper bugs (like a system that doesn’t boot, or X that doesn’t work at all). This makes it very attractive to users who like to regularly get new upstream versions of their software without dealing with the biggest problems associated to them. This is all very attractive, yet several Debian developers advise people to not use testing. Why is that?

Known problems with testing

Disappearing software

The release team use this distribution to prepare the next stable release and from time to time they remove packages from it. Either because it’s needed to ensure that other packages can migrate from unstable to testing, or because they have long-standing release-critical bugs without progress towards a resolution. It also happens that they remove packages on request of the maintainers because they believe that the current version of the software cannot be supported (security-wise) for 2 years or more. The security team also regularly issues such requests.

Long delays for security and important fixes

Despite the 10-day delay in unstable, there are always some annoying bugs (and security bugs are no exceptions) that are only discovered when the package already has migrated to testing. The maintainer might be quick to upload a fixed package in unstable, and might even raise the urgency to allow the package to migrate sooner, but if the packages gets entangled in a large ongoing transition, it will not migrate before the transition is completed. Sometimes it can take weeks for that to happen.

This delay can be avoided by doing direct uploads to testing (through testing-proposed-updates) but this is almost never used, except during a freeze, where targeted bugfixes are the norm.

Not always installable

With testing evolving daily, updates sometimes break the last installation images available (in particular netboot images that get everything from the network). The debian-installer (d-i) packages are usually quickly fixed but they don’t move to testing automatically because the new combination of d-i packages has not necessarily been validated yet. Colin Watson sums up the problem:

Getting new installer code into testing takes too long, and problems remain unfixed in testing for too long. […] The problem with d-i development at the moment is more that we’re very slow at producing new d-i *releases*. […] Your choices right now are to work with stable (too old), testing (would be nice except for the way sometimes it breaks and then it tends to take a week to fix anything), unstable (breaks all the time).

CUT’s history

CUT finds its root in an old proposal by Joey Hess: it introduces the idea that the stable release is not Debian’s sole product and that testing could become — with some work — a suitable choice for end-users. Nobody took on that work and there was no visible progress in the last 3 years.

But recently Joey brought up CUT again on the debian-devel mailing list and Stefano Zacchiroli (the Debian project leader) challenged him to setup a BoF on CUT for Debconf10. It turned out to be one of the most heavily attended BoF (video recording is here), there is clearly a lot of interest in the topic.

There’s now a dedicated wiki and an Alioth project with a mailing list. The rest of this article tries to summarize the various options discussed and how they’re supposed to address the problems identified.

The ideas behind CUT

Among all the ideas, there are two main approaches that have been discussed. The first is to regularly snapshot testing at points where it is known to work reasonably well (those snapshots would be named “cut”). The second is to build an improved testing distribution tailored to the needs of users who want a working distribution with daily updates, its name would be “rolling”.

Regular snapshots of testing

There’s general agreement that regular snapshots of testing are required: it’s the only way to ensure that the generated installation media will continue to work until the next snapshot. If tests of the snapshot do not reveal any major problem, then it becomes the latest “cut”. For clarity, the official codename would be date based: e.g. “cut-2010-09” would be the cut taken during September 2010.

While the frequency has not been fixed yet, the goal is clearly to be on the aggressive side: at the very least every 6 months, but every month has been suggested as well. In order to reach a decision, many aspects have to be balanced.

One of them (and possibly the most important) is the security support. Given that the security team is already overworked, it’s difficult to put more work on their shoulders by declaring that cuts will be supported like any stable release. No official security support sounds bad but it’s not necessarily so problematic as one might imagine. Testing’s security record is generally better than stable’s one (see the security tracker) because fixes flow in naturally with new upstream versions. Stable still get fixes for very important security issues sooner than testing, but on the whole there are less known security-related problems in testing than in stable.

Since it’s only a question of time until the fixed version comes naturally from upstream, more frequent cut releases means that users get security fixes sooner. But Stefan Fritsch, who used to be involved in the Debian testing security team, has also experienced the downside for anyone who tries to contribute security updates:

The updates to testing-security usually stay useful only for a few weeks, until a fixed version migrates from unstable. In stable, the updates stay around for a few years, which gives a higher motivation to spend time on preparing them.

So if it’s difficult to form a dedicated security team, the work of providing security updates comes back to the package maintainer. They are usually quite quick to upload fixed packages in unstable but tend to not monitor whether the packages migrate to testing. They can’t be blamed because testing was created to prepare the next stable release and there is thus no urgency to get the fix in as long as it makes it before the release.

CUT can help in this regard precisely because it changes this assumption: there will be users of the testing packages and they deserve to get security fixes much like the stable users.

Another aspect to consider when picking a release frequency is the amount of associated work that comes with any official release: testing upgrades from the previous version, writing release notes and preparing installation images. It seems difficult to do this every month. With this frequency it’s also impossible to have a new major kernel release for each cut (since they tend to come out only every 2 to 3 months) and the new hardware support that it brings is something worthwhile to many users.

In summary, regular snapshots address the “not always installable” problem and changes the perception of maintainers towards testing, so that hopefully they care more of security updates in that distribution (and in cuts). But they do not solve the problem of disappearing packages. Something else is needed to fix that problem.

A new “rolling” distribution?

Lucas Nussbaum pointed out that regular snapshots of Debian is not really a new concept:

How would this differentiate from other distributions doing 6-month release cycles, and in particular Ubuntu, which can already be seen as Debian snapshots (+ added value)?

In Lucas’s eyes, CUT becomes interesting if it can provide a rolling distribution (like testing) with a “constant flux of new upstream releases”. For him, that would be “something quite unique in the Free Software world”. The snapshots would be used as starting point for the initial installation, but the installed system would point to the rolling distribution and users would then upgrade as often as they want. In this scenario, security support for the snapshots is not so important, what matters is the state of the rolling distribution.

If testing were used as the rolling distribution, the problem of “disappearing packages” would not be fixed. That’s why there have been discussions of introducing a new distribution named “rolling” that would work like testing but with adapted rules, and the cuts would then be snapshots of rolling instead of testing.

The basic proposal is to make a copy of testing and to re-add the packages which have been removed because they are not suited for a long term release while they are perfectly acceptable for a constantly updated release (the most recent example being Chromium).

Then it’s possible to go one step further: during freeze, testing is no longer automatically updated which makes it inappropriate to feed the rolling distribution. That’s why rolling would be reconfigured to grab updates from unstable (but using the same rules than testing).

Given the frequent releases, it’s likely that only a subset of architectures would be officially supported. This is not a real problem because the users who want bleeding edge software tends to be desktop users on mainly i386/amd64 (and maybe armel for tablets and similar mobile products). This choice — if made — opens up the door to even more possibilities: if rolling is configured exactly like testing but with only a subset of the architectures, it’s likely that some packages migrate to rolling before testing when non-mainstream architectures are lagging in terms of auto-building (or have toolchain problems).

While being ahead of testing can be positive for the users, it’s also problematic on several levels. First, managing rolling becomes much more complicated because the transition management work done by the release team can’t be reused as-is. Then it introduces competition between both distributions which can make it more difficult to get a stable release out, for example if maintainers stop caring of the migration to testing once the migration to rolling has been completed.

The rolling distribution is certainly a good idea but the rules governing it must be designed to avoid any conflict with the process of releasing a stable distribution. Lastly, the mere existence of rolling would finally fix the marketing problem plaguing testing: the name “rolling” does not suggest that the software is not yet ready for prime time.

Conclusion

Whether CUT will be implemented remains to be seen, but it’s off for a good start: ftpmaster Joerg Jaspert said that the new archive server can cope with a new distribution, and there’s now a proposal shaping up. The project might start quickly: there is already an implementation plan for the snapshot side of the project. The rolling distribution can always be introduced later, once it is ready. Both approaches can complement each other and provide something useful to different kind of users.

The global proposal is certainly appealing: it would address the concerns of obsolescence of Debian’s stable release by making intermediary releases. Anyone needing something more recent for hardware support can start by installing a cut and follow the subsequent releases until the next stable version. And users who always want the latest version of every software could use rolling after having installed a cut.

From a user point of view, there are similarities with the mix of usual and long term releases of Ubuntu. But from the development side, the process followed would be quite different, and the constraints imposed by having a constantly usable distribution are stronger: any wide-scale change must be designed in a way that it can happen progressively in a transparent manner for the user.

This article was first published in Linux Weekly News. If you want to see more articles like this, join Flattr and click on the flattr button below every article that you like.