<br /> Debian Desktop – Debian Planet

Welcome to Debian Planet

News for Debian. Stuff that *really* matters


Debian Planet is hosted by Bluelinux Internet Services Ltd. Offering a special discounted rate for Free and Open Source software community members.

Buy your Debian merchandise at DebianShop.com.

These are important Debian sites one should not be without!

  • Official Debian site
  • Package search
  • Mailing list archives
  • Bug reports
  • Debian on CD
  • Debian Weekly News — excellent news source!
  • Unofficial APT sources

  • Developers’ Corner
  • Community
    Need help? You’re not alone on this planet.

  • Planet Debian
  • debianHELP
    (User support site)

  • Debian Administration
    (SysAdmin resources)

  • Debian International
  • DebianForum.de

  • DebianForum.dk

  • EsDebian

  • DebianWorld

  • Debian-Fr

  • MaximumDebian

  • DebianItalia
  • DebianUsers

  • Debian-BR

  • DebianHOWTO

  • Russian Debian (Русский)
  • Debian-JP
  • Debian Suisse
  • Contribute
    Got that latest or greatest scoop? Perhaps you have some important news for the Debian community? Submit a news item!

    Or perhaps you’ve written a rather ground breaking insight into some aspect of Debian and you feel compelled to share it with others? Knock up a longer editorial article and send it to the editors.

    General feedback should be sent to staff@debianplanet.org

    The place to get help on a Debian problem (after reading docs) or to just chat and chill is #debian on irc.oftc.net.

    Many of the Debian Planet staff live there so pop by and say hello.

    Debian Planet also has its own channel on the same network called #debianplanet.


    Debian Desktop
    Submitted by junglewiz on Monday, May 03, 2004 – 14:22
    DebianThere are two kinds of users of Debian (this is true for other distributions too):

    1. Desktop users – who would like to run the latest software. This means stable upstream releases and also shorter release cycles. Most Debian users would be willing to put up with one release per year.
    2. Server users – who would like longer release cycles and long term support, with the added stability benefits.

    The interests of the above mentioned two user groups are fundermentally different and conflicting. Therefore, many other major distributions (namely Redhat and SuSE) have different solutions for the separate user groups. They produce server edition with a long release cycle and a desktop edition with shorter release cycle.

    Debian provides a one size fits all solution which aligns more with the server users than with the desktop users.

    Should Debian consider something similar? If not we will be asking when will the next stable will be released all the time!

    Category: Opinion

    Control panel

    Comment viewing options:

    Select your prefered way to display the comments and click ‘Update settings’ to activate your changes.

    Subject: release cycle
    Author: yasuo.hiroshi
    Date: Friday, 2005/02/11 – 17:30
    I really think that releases shouldn’t be hurried. I’m a SuSE Pro 9.1 user currently; however I intend to migratre to Debiain Sarge when it is stable. I’m really in no hurry; Although it would be nice to see shorter release cycles. I think an average of a year to a year in a half would be appropriate and a two year legacy support for older versions to keep the server end happy. This whole Server Vs Desktop thing is stiffling innovation and progress on both ends. Perhaps the Debian development team should focus more on legacy support than building support for old systems like ALPHA which consume less than 1% of the total computer industry. More effort could then be reallocated towards architectures that really matter and making Debian more accessible to both the average Joe-Schmo computer users and the IT tech genius.
    It’s all about compromise. A one size fits all solution is great, but needs to be released in a timely manner so that desktop users can reap the benefits, but also Debain should continue to support previous legacy versions to keep the server side satisfied.
    I really like the Debian philosophy, however I am not willing to run an outdated OS or an OS that is in “testing” no matter how stable. So I guess until Debian Sarge releases I won’t be switching.
    [ Please login, or register ]

    Subject: Unstable>testing>stable: That’s Debian all about
    Author: mustaffah
    Date: Sunday, 2004/11/21 – 15:00
    I dont use linux, yet! My work is mainly around AUTOCAD… The company needs Windows…:( But i’m starting a project involving a Cyber CAFE and want to have at least 25% of the workstatinos on LINUX. Im studying all distros looking for the one (for the project i mean). There’s something i learned: among all package systems, DEB with apt-get is the one. If you want to be on the edge, you’ll know that stability is compromised. I’ve used 3 distros. Mandrake, Red Hat and Debian. The only one i could shoose between stable, unstable freely, and i mean freedom and almost no cost (the download and burn processes…) is DEBIAN.
    The DEBIAN system is perfect. We just can’t press them to turn it in a system like Fedora where you’re at the newest among the new software avaiable with no warranty, because they already have it. It’s called UNSTABLE.
    In Portugal we say “Para quem não está satisfeito, a porta da rua é a serventia da casa” or For the ones complaining we have the door out of this house. If you want Debian to turn into Red Hat, SUSE or MANDRAKE, I DON’t. And i sugest you should change to one of those distros.
    Thank you.
    [ Please login, or register ]

    Subject: freeze
    Author: Seegras
    Date: Wednesday, 2004/06/30 – 16:24
    There obviously is a Problem. Initially I wanted to have the servers on stable, and the workstations (which are not that important, and set up easily, since they contain no special configs, and the homedirs dwell on NFS) on unstable. It was nice until about 18 month ago, where we realized that we _needed_ to have more modern software on our servers. And the work of backporting seemed to be too much work, so we run unstable now. We really would like to have these on stable, but with the current release-practics, this is just a no-go.

    I now read most of the comments, and I concur, that the only solution is to

    freeze after 6 months of development

    This won’t make the differences between stable versions as big as now, but they will be much more up to date. And the actual release-cycle might still be somewhere between 12 and 18 months. Oh, and why not make them point-releases? At least you could name the stable releases decent. 3.0r2 ist just awkward, make it 3.2, why not.

    [ Please login, or register ]

    Subject: What about debian core?
    Author: jsebrech
    Date: Sunday, 2004/06/06 – 00:04
    My apologies for the length of this post, but I think it’s necessary to get my point across.

    I don’t think the problem is a desktop/server dichotomy. Like other people have pointed out, stable is as useless on servers as it is on desktops. Running stable, even with backports (which circumvent the system, and are not an appropriate solution) is pretty much out of the question, and not running stable means you have no security upgrades, so that’s not a good situation either.

    In my mind, this problem is caused by the inherent impossibility of stabilizing an entire collection of 10.000+ packages in time for release. Either you trim down debian, or you accept that it is almost always out of date. But debian’s strength IS the large number of packages in the system, so you can’t trim down the entireness of debian.

    So, what to do about it?

    Split up debian into a “core” distro and a surrounding “extended” distro.

    Core would be developed and released on a fixed (reasonably fast) schedule. It would include the kernel, the basic libraries and tools (like glibc, perl, …), and would probably best include X too. It would not have to grow beyond one cd in size, so it could be managed more effectively.

    Then for every core you would have an extended release managed through stable/testing/unstable.

    Extended packages would not be guaranteed to work on anything but the stable core release (to discourage developers from running an unstable core).

    Benefits of this system would be:

    – a faster release cycle for the base platform so there has to be less backporting (because the libraries in the core are up to date enough to run the newest stuff out of unstable).

    – no requirement to stabilize debian entirely for each release (who here honestly believes you can stabilize 10.000+ packages every year?)

    – Since there would be an extended for every core, you could provide updates for old debian releases if there are developers willing to do the work, and not have to circumvent the debian system to do it. A new release wouldn’t instantly obsolete the old release entirely and freeze all development on it. Corps would love this long-term debian-supported package stability.

    – because the package maintainers working on core stuff tend to be closer on the ball, you wouldn’t have to wait months and months for all release-blocker bugs to get fixed.

    – It would incentivize developers to make sure they have package releases out which work with the stable core platform. Currently developers run unstable, and so have little incentive to get their packages in stable, because they see little personal pay-off. If package maintainers could only use their own unstable packages if they worked with the stable core, you would see releases that can be installed on top of stable more often. By releasing the core first, you would create a flurry of activity to get stuff running on it as developers move to the new core.

    – you could provide security updates for the core platform, and not have them be on obsolete versions of the software

    – end-user software package releases would no longer be bound by the debian release schedule. If end-user package releases happen faster, not a problem, you’ll be able to use them without upgrading your glibc. If they happen slower, you would get public pressure on package maintainers to put a release out for the new core (and if it keeps going too long you’d see NMU’s or a package changing maintainer hands).

    – This would put a spotlight on unmaintained packages. A new core would create very public visibility of the lack of maintainer, since there wouldn’t be a release for the new core.

    – This would make release on cd easier. Core would be smaller, easier to package up. End-users could install core from cd, and then during the install core could ask whether you want to install extended from cd or from the internet. This would make sure users could be given recent kernels (with the faster release schedule), and still install unstable end-user apps right from the initial install.


    – tools would have to be rewritten.

    – when new cores are released, there would be no packages in the new extended release. So upgrading cores immediately would be problematic. But developers would have to upgrade to get new versions of the core libraries, and once they upgrade, you would see new extended releases. So I don’t think this would be a real problem, especially given that users could wait a few months before they upgrade core releases (with low-intensity development still going on in the old release).

    – It might be more complicated to understand for new users. But on the other hand, new users currently have to move to unstable if they want to do anything useful, and that’s not exactly an optimal situation with respect to ease of use.

    I’m not sure whether the stable/testing/unstable system should be kept for managing the extended system, but gradual change is always preferable, and splitting the distro up into a core and an extended release but still managing both of those with the existing system would be les disruptive. Then afterwards you could always still change the management system for either core or extended as you want.

    I’m certain there are plenty more upsides and downsides to this idea, and I’m certain I’m not the first to offer it, but I’m curious what kinds of reasons people would have to consider this a bad idea.


    I’ve given the idea more thought, and the big flaw there is to this scheme is that core would not see any end-user testing. The way it seems to me this would best be solved is by creating a new extended when a new core testing/unstable branch is started (instead of when that branch is released as stable), but to only allow packages to be checked into this extended if they are also checked in to the extended for the core stable release. That way the adventurous and the developers could run an unstable core with full functionality, but we wouldn’t fall into the “no incentive to get packages into stable” trap of the current scheme, and extended would be prepopulated with packages the moment core is released as stable, which would be a nice side benefit.

    Anyway, the big problem seems to me to be the lack of personal incentive to get packages into stable, so unless that is somehow fixed, I doubt any change will improve things.

    [ Please login, or register ]


    Subject: Lack of quality in systems using this approach
    Author: gwolf
    Date: Tuesday, 2004/06/22 – 17:10
    Before turning to Debian, I was an OpenBSD fan for some time. And it was mainly this point (together with the attitude problem, i.e., the lack of importance of the users – ‘we make this for ourselves’ vs. ‘our priorities are our users Free Software’) that drove me to Debian. In the BSD world, you have a core OS, which has guaranteed quality, performs perfectly as a whole, (in the case of OpenBSD) is constantly subject to thorough code audits. And then you have the ports. People maintaining the ports are almost like second-class citizens. You get no warranties of any kind on the ports (i.e., security updates. Even more, some of them do not even build correctly, although this is not too common, but still, it is quite annoying).
    Debian, I agree, does not build each component of the OS – We take Linux (or your kernel of choice, but let me assume it is Linux), the basic utilities that must surround it, things that everybody expects to find, and just about anything else. And we provide (or, at very least, we try to provide) equal support for each of our packages, be it core or not.
    In the truest sense, we _do_ have a core system – it is called ‘Section: base’. Inside that, we have the minimal Debian system: ‘Priority: required’. But we do not discriminate against a package that happens to be ‘Priority: extra’ and ‘Section: web’.
    [ Please login, or register ]

    Subject: No need change current system, just add “latest” on the side
    Author: arborint
    Date: Saturday, 2004/05/29 – 23:44
    I have read many of the comments here and elsewhere about the stable/testing/unstable+experimental system. I think the problem is that two different groups are talking about two different things, but those groups are not desktop and server: they developer and user.

    The first group is the Debian release machine that is forever moving forward toward the next release. To them the: experimental > unstable > testing > stable process is the only way to get their huge task accomplished. It works. They are continuiously improving it. They are working on improving the release cycle. Don’t bother them, help them.

    The second users, often desktop users, who find themselves in the strange position of using a release called unstable on their machine. We found unstable after searching for the packages we wanted and discovered that they were mostly in testing or unstable. So we gave unstable a try and have been using it for years because it’s not unstable from a use point-of-view. It’s only unstable from a release point-of-view. Then we learn that there is this thing called experimental, and some of the stuff we want is in there and runs fine too. Hence this discussion with the releasers and the users speaking different languages.

    My proposal is to have an autobuilt, apt-getable thing called “latest” that has all the latest packages from unstable and experimental. The package maintainers could mark their packages and they would be automatically added. Anyone using “latest” would understand that they need to send bug reports (which we do already). Teams like the GNOME group can put their stuff in latest and get feedback while still following the rules to get into unstable. This will fix situations where the latest is better/more stable than the current stable version.

    And the Debian Release Machine can roll on, no longer bothered by people complaining to them.


    Developers: experimental > unstable > testing > stable

    Users: latest or stable

    [ Please login, or register ]


    Subject: No need to change the current system
    Author: tilolevante
    Date: Thursday, 2004/06/03 – 19:10
    I have several servers with different customers since years running the unstable distribution. Without any major problem.

    I have more problems with SUSE.
    Debian, I can update over years usually in 15-30 Minutes.
    Even Updates from relative old versions (unstable, 2 years old), worked
    without major problems (perl and apt is difficult).
    SUSE does not support a decent upgrade to newer versions, so a update
    means reinstallation, and this is hard for a server.

    My impression is, that unstable gets more unstable in the last time.

    I would suggest to rename the levels:

    unstable -> latest
    testing -> current
    stable -> release


    [ Please login, or register ]

    Subject: Pre-release syndrome
    Author: stoft
    Date: Saturday, 2004/05/22 – 05:03
    Judging from previous posts there are a lot of different suggestions that have merit (Official backports, split into Server/Desktop, split into x86-arch/other-arch, shorter release cycles, re-evaluate what the Debian target userbase is etc.). But I can’t help wondering, isn’t this all just pre-release syndrome? Once Sarge goes stable won’t this discussion just go into hiatus for 16 months and reawaken just a bit too early for the DDs nerves? Debian is stable. In the long run that is the important thing (especially if you, like me, come from a multi-user environment such as a campus). If you’re a desktop junkie wanting the latest eye-candy it’s not Debian’s fault, learn (i.e. rtfm) to adapt (backport, unstable etc.) or switch distro. Same thing for sysadmins, if you’re good enough you adapt your system to your requirements (that’s why it’s opensource) or you switch distro.

    The problem isn’t Debian not releasing quick enough, the problem is the rest of the world (Gnome, KDE etc.) not providing Debian with stable enough releases as quickly as they should. Or something. Bah… I guess this turned into a rant. Let the flames cleansen me.

    So what was my point? Kudos to the DDs of course. The rest of you… go develop something. 🙂

    /I want to be a developer when I grow up

    [ Please login, or register ]


    Subject: RE: Pre-release syndrome
    Author: calvinpriest
    Date: Thursday, 2004/05/27 – 20:28
    Certainly part of this is Pre-release syndrome, but it is also at the heart of the reason why Debian does not yet rule the world. Debian is the planet’s rightful ruler, but first she needs to cut her release cycle in half.

    Debian has been trying to move toward a 1-year release cycle, but of course currently she is on something closer to a 2-year cycle. If key backports were officially tested and blessed this *might* bring more more people to Debian. But when the competition is on a 6-month cycle Debian is doomed to look very much behind the times. And until devolopment of most key applications reaches a certain plateau Debian will in fact be behind the times.

    This is not about stability. It is about changing the “Debian releases when it’s time” philosophy just a little bit. That philosophy needs to be amended to “Debian freezes on schedule and then releases when it’s time”. IMHO, the freeze date for a release should be more or less cast in stone, and then release should occur once stability is at the required level. Packages that threaten the freeze date should be held back until the next release, which would be only 12 months away, after all.

    As for separate Server and Desktop distributions, I think that issue would die if the release philosophy changed. After all, there are several excellent Desktop derivatives of Debian built on Unstable and typical desktop users will choose those anyway.

    [ Please login, or register ]


    Subject: Why six months?
    Author: hazelsct
    Date: Friday, 2004/05/28 – 23:29
    Why six months? Why not two years? Why not three weeks?

    The kernel release cycle is about every two years. The GNOME major release cycle is about every two years, so these RedHat and other users have just been beta-testing GNOME 2 on its way to maturity for Debian sarge (while woody has been rock-solid with GNOME 1.4 in the meantnime). Others have been beta-testing the 2.6 kernel so it can be ready for sarge.

    Furthermore, Windows releases are about every two to three years. Why does Windows matter? Well, it doesn’t, but it sets expectations regarding frequency of retraining, particularly with large organizations (many of which are still running Windows 98 — so much for six months!). Heck, I don’t want to have to retrain my group of technical university graduate students every six months (they’re still running woody desktops), two years is just fine with me, thanks.

    Your assertion that Debian will be irrelevant without shorter cycles lacks any coherent justification, nor does it measure up to the major release cycles of its foundation software packages, nor with established practice in the rest of the industry. Try again.

    -Adam P.

    [ Please login, or register ]


    Subject: RE: Why six months?
    Author: calvinpriest
    Date: Thursday, 2004/06/03 – 22:23
    You make some good points, however…

    I am not recommending a six-month cyle, I am recommending a one-year cycle. Genuinely sorry if that wasn’t clear. The six-months figure represents what I see as a relatively common release cycle for major gnu/linux distribution vendors. Some of those distributions are six-months, some are one-year, but I cannot think of another major linux distribution with a two-year cycle like Debian, please correct me if I’m wrong.

    Also I’m including point releases (6.1, 6.2) etc. in my logic if they include updated libraries and applications. So even though Gnome may only put out a “major” release every two years, their point releases are much more frequent and often include significant feature changes (witness 2.6) as well as updated libraries and applications.

    Your comparison with Windows is a reasonable one. However largely because of relative uniformity of available libraries, new and updated Windows applications are easily installable between OS releases. Not so in the free software realm. In the RPM world you generally need a package built specifically for the version of the distribution you are using. In the Debian Stable world we can use backports. If backports were tested and blessed by Debian this would greatly reduce the need for a shorter release cycle. But in the end, wouldn’t that be more work than simply commiting to freeze and release about once a year? Also, maintaining systems (servers or desktops) with multiple backports can be quite cumbersome as things currently stand.

    So this is my (hopefully) coherent assertion: without official backports a one-year release cycle is all but necessary to keep from making servers and desktops behind the times and difficult to maintain. One further point: the free software world is moving very fast. Some free applications go from being alpha to production quality in less than two years. It sucks to wait for them or add them to our list of backports to personally test and maintain.

    These problems are widely recognized by major commercial linux distributions and I believe they are fueling their shorter release cycles.

    Remember, I’m not saying quality and stability should be sacrificed. Debian’s commitment to these values is a large part of the reason for it’s success. Certainly it’s a large part of the reason I choose Debian.

    In short: Freeze on schedule, release when ready.


    [ Please login, or register ]

    Subject: Another Way
    Author: DaGoodBoy
    Date: Thursday, 2004/05/20 – 22:29
    I’m certain that some will consider me a heretic with regards to Free Software and Debian policy, but I am just pragmatic and the scheme I describe here works very well for my company.

    We maintain a custom package repository for our local servers and desktops as well as local mirrors of the debian repository for stable, security and unstable. We have duplicated a small part of the build system to allow us to easily generate packages from dsc, diff and source with artificially increased release numbers to prevent collisions with the real Debian packages. The non-Debian custom repository allows us to distribute them easily inside our network.

    Our Servers:
    Debian Stable with certain key software backports for specific applictions like Cups (including Ghostscript, foomatic, and the lastes printer drivers like hpijs), Samba 3.0 for our two Windows systems, Apache/PHP4, MySQL, PostgreSQL, OpenLDAP, etc.

    Our Desktops:
    Debian Unstable against a forked local repository we update against the public unstable tree using rsync about twice a year. We generally have to tweak and build custom versions of a few packages to work out any real bugs in our snapshot of the unstable tree. Once the kinks are worked out, we update the 30 or so desktops usually in a single afternoon.

    We have been doing this for over two years now with great success. Our servers are stable, still grab security updates for the standard packages they run and we only have to monitor the security fixes for our special backported packages related to our development and special needs.

    The desktops run very current software and are very stable once we complete our short evalution and tweaking process. We have custom versions of rdesktop, Mozilla and some of the patent encumbered libraries for video, audio and image formats that supercede the default Debian ones.

    Our development systems have custom built versions of all the Debian Java related packages using either Sun’s or IBM’s JVM so we can develop apps for our customers that will be compatible with their choice of app server and JVM. Kaffe developed Java apps still have odd errors and we can’t risk having to debug on the customer’s dime.

    This works out as the best solution because commercial distros like Red Hat and SUSE don’t provide the critical build and distribution environments that allows us to do this. They also don’t have the strict packaging discipline that provides good dependency and package granularity for individual packages. For example, we don’t allow peer-to-peer software or instant messaging clients for our workstations. If you use Red Hat or SUSE you don’t get to pick individual KDE packages to include, you are forced to take KDE as whole cloth and delete specific executables or libraries post-install. Using Debian we can more easily enforce our client software policy than with any other commercial distro.

    The idea of being able to simply reach out and grab the current set of packages when it is convenient for us based on our project load and staff availability is very compelling. While I might appreciate a slightly more regular stable release cycle, the existing Debian architecture is still the best one out there for managing a business I.T. environment in my opinion.


    [ Please login, or register ]

    Subject: make two !
    Author: Dalton
    Date: Sunday, 2004/05/16 – 11:00
    make two one for desktop , one for severs. /Its a big world / and more then enough room for both.
    [ Please login, or register ]

    Subject: Debian Desktop
    Author: bilarson
    Date: Friday, 2004/05/14 – 05:04
    On a personal basis, I completely disagree with this.

    I consider myself a non-technical desktop user who has been using Debian as my only desktop since 3.0 came out and Gnu/Linux as my only desktop since 1999. The only important think for me is whether the software works or not. I don’t need beta software. If a particular program does not work, I don’t want it only my computer and I found certain other distributions of Gnu/Linux always seemed to have broken software. Just give me a good consistent operating system that works. I don’t even need an xserver if the software that works is available for the console. As long as I can use the internet, publish attractive looking documents, listen to music and do basic financial things on my computer, I’m happy and don’t need anything else so I guess I don’t fit into any of the two categories that everyone is supposed to fit into.

    [ Please login, or register ]

    Subject: Desktop and Server / stay the way it was?
    Author: anselmoso
    Date: Thursday, 2004/05/13 – 09:11
    First, let me start by addressing a big “Thank You” to all of these developers out there who spend their free time to make such a great project like Debian!

    In the past 4 years, I’ve tried nearly *every* taste of linux one could taste and I kept trying the new releases of each distro, so I would be up to date. I know Mandrake 10, Suse 9.1, Fedora Core 2 (test3), TurboLinux (in the good ol’ days), Knoppix, Lindows (Developer Edition for free), etc.. etc…

    And I’m always glad of the fact that one day I tried Debian even though the installer was a shock to me that time.
    Last week I tried the new Debian-Installer and I must say it’s a big step forward. Although the partitioning seemed kind of easier to me with CFDisk.

    Anyway – the Server / Desktop discussion:

    In the past I also remember being unsatisfied with the release cycle of the stable release. At home I am a pure desktop user, seeking for the “perfect desktop system” but I also set up some servers for little companies or friends sometimes. And i always chose “testing” these days. Why? Because it’s fuc… stable:-)
    “Testing” in Debian is as stable as Suse 9.1, Mandrake 10 or Fedora in my opinion, even more stable!
    My Debian system never freezes out of nowhere – Mandrake for example did several times and had some other weaknesses for me.

    Nethertheless I think a split into Desktop and Server wouldn’t hurt.
    Although the arguments from previous writers about the “need for latest packages on server systems” are also justifiable. Some may want to run latest database software because of new functions, or latest servers (Apache2) or whatever. Also the 2.6 Kernel indeed brought some nice new features / performance updates for servers.

    So in the end I come back to my first argument: It’s OK the way it is:-)

    Personally I think you can run “Testing” on a webserver today with the condition that you will update your system regularly.
    Otherwise stick with “Stable” and with “old” packages.

    [ Please login, or register ]

    Subject: leveraging resources
    Author: Henry Hollenberg
    Date: Wednesday, 2004/05/12 – 12:31
    How about a summer camp…at one of the Universities
    once a year. This could serve as the focal point
    for churning out a new release once a year. University
    students and those of us with “real” jobs tend to have more
    time set aside in the summer for “outside” activities so
    it should be a good fit for both.

    It could also serve as a great opportunity to initiate new
    developers into debian and discuss difficult issues.

    Those with $$$ corporations/individuals/Universities could
    be called upon to help support the event thru tuitition,
    food and housing allowances, travel expenses and setting up
    an electronic lab.

    Henry Hollenberg

    [ Please login, or register ]

    Subject: Its the number of arch’s
    Author: gomadtroll
    Date: Wednesday, 2004/05/12 – 07:01
    I don’t think that the solution is in tweaking , stable. testing & unstable, that topic has been discussed enough. Note that instead of speeding up the release cycle there is another version..experimental. The reason debian has a slow release cycle is the 11/12? arch’s it supports. I like a slow release cycle but a slow release cycle only works when you have achieved a minimum functionality in a core set of apps for whatever task is at hand, and I don’t mean by using backports..etc.

    The meta-distro’s like Xandros, Libranet, Progeny, UserLinux ..are more nimble because they support x86 arch only, but they all still depend on Debian developers for most packages.

    Instead of talking about server vs desktop,, etc.. split Debian by arch’s. The ones that don’t build timely do not have to hold up the others. X86 in one release, go by the number of users for release, not aimed at kicking out any arch, just put it in perspective.

    Greg Madden

    [ Please login, or register ]


    Subject: Bzzt, wrong again
    Author: hazelsct
    Date: Wednesday, 2004/05/12 – 19:29
    Debian-installer has not taken 2-3 years because it’s trying to work on all architectures. If that were the case, it should have been ready 18 months ago, and we would be catching up the other arches in the meantime.

    In fact, D-I is not yet ready for i386, because of the scarcity of resources focused on it. Debian, like Open Source in general, is a volunteer organization of developers driven by the needs/wants/itches of those developers. As all of us can use boot-floppies just fine, there’s not a strong motivation to sink significant time into D-I, we just keep packaging, upgrading and using our favorite software on machines already installed.

    Corporations can say, “We need a release out soon,” and direct their developers to work on it. Debian can’t do that. That is why it’s taking so long.

    -Adam P.

    [ Please login, or register ]

    Subject: stable is too old for modern hardware
    Author: Miikka
    Date: Tuesday, 2004/05/11 – 08:43
    I like running stable but with recent hardware that is not possible anymore,

    here is my current machines:

    office workstation:

    compaq hp d330

    Broadcom network adapter: requires 2.4.25 kernel
    Intel display adapter: requires 4.3 XFree86

    also my laptop dell lattitude 640

    Ati display adapter: requires 4.2.1 XFree86
    intel network adapter: requires 2.4.21 kernel

    my home pc (for playing and stuff)

    clone pc with intel P4 82845G mainboard

    Ati display adapter: requires 4.2.1 XFree86 (4.3 if you want 3D support)
    intel network adapter: requires 2.4.21 kernel

    my server:

    using stable but php and mysql had to be hacked to support recent features

    im forced to use testing in all machines and thats not good…

    and servers _do_ use modern harware too.

    i dont really need newest desktop software but if i really like to get atleast XFree86
    and network working on stable release …

    [ Please login, or register ]

    Subject: binary gentoo
    Author: joh
    Date: Monday, 2004/05/10 – 17:46
    I am running gentoo, and its great. I am allways using the newsest software, and I have very little problems.

    But gentoo is source based, and a real pain sometimes. If you have a hd crash or something, it usually takes weeks before everything is back to normal. And its terrible when you need software fast, for excample when someone invites you to a netmeeting session, and you have to compile gnomemeeting and 4 deps.

    Why couldnt Debian use the same package policy? I understand that it takes some more testing before releasing a binary package, but do I really have to wait for over a year?

    Flame: To me stable is useless, testing is useless with a twist, and unstable is like the developers think: “This package breakes my system, lets put it in testing so it breaks everyones system. That will probably generate some bugzilla traffic, and eventuialy a sollution!”

    Why not make a desktop branch that follows Fedora and Gnomes 6 month releace cycle? It looks like Redhat is trying to make a desktop-debian with Fedora, and to me it looks like they can steal some marked shares from Debian.

    [ Please login, or register ]

    Subject: Too slow
    Author: BrendaEM
    Date: Saturday, 2004/05/08 – 18:26
    I am a Gnome affectionado who is interested in Debian because of its openness. I have used RedHat 8 and 9. I have also tested Core 2. I have used apt and synaptic before. I like them so much, I want to try the whole distro : )

    As an outsider perspective Debian user, I does appear to me that the Debian release cycle is not anywhere near quick enough for the needs of the desktop. Debian’s long development cycle is the only thing keeping me from trying it.

    Compare Debian to some other distributions. Why are you nice people testing obsolete software?!

    Are there are some fourteen potentially obsolete packages listed in Debian testing? Why waste time testing old packages? I am not suggesting to throw alpha and beta stuff in there. But, perhaps there is something wrong when a released upgrade is not even worthy of testing.

    The Linux desktop is in a process of growth, usually making the newer releases better and more stable than the old. In fairness, it’s good to see OpenOffice 1.1.1 in there.

    What about security? Is the old software as secure as the new?

    “Most Debian users would be willing to put up with one release per year.” Please, do a poll, here, and one slashdot or Linux today. See how many users won’t try it because they think it’s not keeping up with the times. The fact that that sentence includes the words “put up,” seems to suggest to me, that they are not happy, and whatever fact-finding that was done might be slanted or flawed. This whole controversy wouldn’t be happening if there wasn’t anything to it.

    Perhaps some kind of feedback software would help, so testers can quickly rate a program for stability?

    Things like the underpinnings and things like apt, and installer GUI’s for the desktop should be really stable. But, can’t buggy versions programs be at the end of the dependency chain be removed and older ones installed if need be?

    [ Please login, or register ]

    Subject: Stable bit slow, unstable good for desktop software
    Author: wouter@jabber.org
    Date: Friday, 2004/05/07 – 16:05
    Stable is a bit slow for servers, too. You don’t want to run last years’ software, in one year some packages can evolve quite a lot, and serious problems (either general or requirements/aids for your own specific implementation) could have been solved. Ofcourse, it doesn’t have to be the latest CVS checkout either, but I still think stable (say, as of 2004) is getting quite out of date. Granted, apt-get updating minor corrections and security fixes is major cool for colocated servers.

    For desktop use, I prefer unstable (especially for people with fast and cheap internet connections). If connectivity is expensive (dialup users) or frequent updates are less relevant, I tend to advise slackware.

    [ Please login, or register ]

    Subject: Even server update too slow
    Author: ttuncmd
    Date: Thursday, 2004/05/06 – 23:59
    Well I run debian on servers… and even there the update frequency is slightly too slow. Stability and security are the prime concerns in operations but if your customers go elsewhere because your software is WAY out of date then eventually you end up not running any servers.

    So my two cents: much more than 18 months between updates hurts “server” as well.

    Stewart’s suggestion fo two streams would, I belive, increase the update frequency of the server side as well. As the server side would indeed be able to live without KDE, Gnome, the newest snaziest X11, etc. It would be easier for the release manager to update the core of debian occasionally.

    I would expect the server-core would be the basis for the desktop stream.

    Perhaps two release managers working together would also ease the load?


    [ Please login, or register ]


    Subject: Same problem
    Author: hopfgartner
    Date: Friday, 2004/05/07 – 06:38
    Personally, I agree that release cycles of more then 18 months cause a lot of problems on servers.

    My servers have currently a lot of backports from different places and self compiled backports of testing packages. I’m not very happy about this, but I can hardly do it in another way.



    [ Please login, or register ]

    Subject: main non-free contrib backports
    Author: vilemaxim
    Date: Thursday, 2004/05/06 – 15:43
    Arguments for non-free

    I think the argument made for non-free also works for backports. That is, non-free is a compromise between the Debian ideal of free software and our users needs.

    Semi-official backports
    Currently I admin a small shop with a few Debian boxes. All stable boxes are run with a few backports. Most people that I know that use Debian ether use stable+backports, testing or unstable. I don’t know of anyone that just uses stable, though I’m sure they are around somewhere. If you look around (apt-get.org) there are all kinds of backports, often the same ones.

    I think making a semi-official, similar to non-free, backports repository would solve a lot of this, i.e. serve the users, end duplicate efforts, help backport maintainers and users by providing a familiar bugreport system, etc.

    Backports.org basic system seems like a great idea, i.e. where in your apt repository you can put in the programs you want upgraded.

    deb http://www.backports.org/debian stable mozilla-firefox xfree86 memtest86

    Maybe the backports can have incremental releases so the user has to modify the source.list in order to get the next release of backports so they don’t get a surprise upgrade.

    non-x86 archs
    If you look http://bjorn.haxx.se/Debian/ you will usually see a non-x86 arch that is holding up some package from getting into testing, for whatever reason. Attempting to support so many arch seems to weight down the release process, and maybe scare off potential maintainers (not sure on that one). I’m not saying that we should drop non x86 archs, but just noting the process. I think that Debian should encourage, but not force backport maintainers to package for all archs. so if we get a backport maintainer that wishes to backport for the x86 arch only, let them have at it. This, like non-free, is not Debian’s goal, but a compromise for the users, and, like non-free, doesn’t have to be considered part of Debian.

    That just my 2 cents, maybe the componentized Linux would work.


    [ Please login, or register ]

    Subject: I may be an oddball but….
    Author: mikeage
    Date: Thursday, 2004/05/06 – 13:25
    I think Debian is great as-is.
    Though I’m not a programmer, I consider myself
    a ‘technical user’. Have a STABLE desktop system
    with Fluxbox for the GUI. Seems pretty ‘shiny’
    to me 🙂
    I am concerned though, about the server people
    here, who seem to feel under-served. People who
    want the blingbling desktop can run Knoppix,
    Xandros, Libranet, Mepis, yaddayadda. All ‘stable’
    within themselves, and with active support.

    My greatest fear is that this (I can’t say it) ‘whole thing’
    will dry up like a beautiful mirage because of people
    (server admins, etc?) drifting away over lack of edge-ness.
    I think a perspective check is in order…
    For me Debian has always been about doing things in an
    academic way. No loosey-goosey follow the fad crap. The DD’s
    are *building* something. They are taking the best technology
    of the times, and nailing it down into a glowing record of meaning that the whole world could be proud of. If that takes two or _ten_ years, may the Maker bless them and keep them sane!

    [ Please login, or register ]

    Subject: 2 kinds of users…
    Author: nomar
    Date: Thursday, 2004/05/06 – 12:11
    There has been much discussion lately regarding the differences between desktop and server users, yet there seems to be little regarding the differences between home and business users. However, this is probably the most important difference (I actually believe that many people have confused server/desktop for home/business).

    At home I don’t have a problem using debian unstable on my desktop. I quite like the idea of doing a dist-upgrade every couple of days to see what new stuff there is, and it’s all near or at the cutting edge. If something doesn’t quite work I can fix it or put up with it until the update comes. I also use ‘unstable’ as a server, providing me with a test web-server and network services for my home network – again if things break it’s not the end of the world, I can fix or wait.

    If I were installing a system for a small business for example, it would be difficult for me to choose debian. For one, I’d want to give them an up-to-date desktop, yet I can’t have it break or change all the time. And as for server, most people are telling me that stable is just too old (until sarge anyway), I know that I’d probably have to upgrade PHP, Mysql and Samba just off the top of my head. Under these circumstances I’d probably look at ‘White Box Linux’ (based on Red Hat Enterprise Linux 3) for server and ‘Fedora core’ on the desktops.

    I know I could set up a local ‘unstable’ repository and be selective about what I let the clients install, and I may look into this, but it’s a lot of work when I can get what I need elsewhere. Debain, however much I’d like to use it, seems to be limited in this area.

    What do you people think? Am I missing something? Is there a simple way for me to implement debian in these scenarios?

    [ Please login, or register ]

    Subject: The ‘Kinds’
    Author: devians
    Date: Thursday, 2004/05/06 – 11:57
    I agree, there are (basically) 2 kinds of Debian users and they have different upgrade needs. However, the root of this difference is not illuminated.
    Server users do not require frequent changes (upgrades) because they happy with their present state. Their needs are almost perfectly satisfied. That is, Debian is a very good system for this purpose.
    Desktop users are not happy with Debian (either Woody or Sarge) thus they are eagerly waiting for the newer and newer versions and hope if the newest ones will be more usable. That is Debian is a weak system for desktop purposes.
    For citing from my life: I waited for the newest DOS version (and later Windowses) until Windows98. And stopped there because this mostly fulfilled my needs (excepting its instability). And I did not want to upgrade anymore for the followers (as WindowsMe, etc).
    If once, Debian will be useful and convenient for Desktop usage, the majority of Desktop users will not want it upgrade very often. Only a small fraction of them (say ‘power users’) hunts the latest features in operation systems.
    [ Please login, or register ]


    Subject: Kinds?
    Author: WanderingGhost
    Date: Thursday, 2004/05/06 – 12:15
    Server users do not require frequent changes (upgrades) because they happy with their present state. Their needs are almost perfectly satisfied. That is, Debian is a very good system for this purpose.

    Please see my post (the one right before yours). I had several problems setting up a server because the packages in Woody are too old… Maybe Woody is good for a basic server setup — but it didn’t work for the virtual domains+web-admin setup I needed.

    [ Please login, or register ]


    Subject: Yes, to WanderingGhost
    Author: devians
    Date: Thursday, 2004/05/06 – 13:23
    Yes, I agree that Woody is too old. Despite, there are lots, who are satisfied with that even now.
    What I discussed above is a general trend about our needs and usage.
    That does not oppose your case.
    Obviously, even in the server field, there are news regarding either hardware drivers or software solutions. And there is some demand to follow that. But this demand is less expressed then in desktop usage and, thus, a less frequent upgrade is enough.
    Specially, this is the case now with Woody, IMHO.
    [ Please login, or register ]

    Subject: I don’t think it’s a “desktop versus server” problem
    Author: WanderingGhost
    Date: Tuesday, 2004/05/04 – 23:25
    1. Desktop users – who would like to run the latest software. This means stable upstream releases and also shorter release cycles. Most Debian users would be willing to put up with one release per year.
    2. Server users – who would like longer release cycles and long term support, with the added stability benefits.

    I don’t agree with that. I admin a server that hosts several sites and virtual email domains. It is important that things be configurable via web, and intuitively. So… I tried to do that with woody. No scalemail, so I had to get pieces of software from all over the place (some backports, some from source).

    • I had to get postfixadmin from high5.net
    • Then I had to install a postfix backport so postfixadmin would work well — and I also ended up needing some features of postfix 2.x…
    • Spamassassin changed a lot recently – the old version doesn’t perform well anymore
    • MySQL is also old, and one of the guys who will develop websites wants a new version. He knows PHP and MySQL, and I can’t use “stability” as an argument to not let him use his skills and PHP libraries that use new MySQL features (which I don’t remember anyway)
    • I also got a backport of clamav
    • Got a backport of amavisd-new
    • chkrootkit, 2 years old (no need to explain)
    • Had to use a backport of squirrelmail in order to get new plugins working (the ones I needed require squirrelmail >= 1.4).
    • rsync doesn’t work. (The old rsync in woody won’t talk to the new rsync in our backup server, which is not under my control) — several times our backup script stopped and didn’t finish because of that.
    • Even grep is outdated! I tried using the WPBL scripts, and they use “grep -o”, which the version in woody doesn’t support

    And there are things that we can’t do yet. Woody lacks functionality, as a server. I can’t tell people “we’re going to use this software, 2 years old, because it’s more stable”…
    The “world of servers”, so to speak, can also evolve rapidly.

    Anyway – my point is, the “woody is old” argument doesn’t apply to “cute desktops” only. Some people really need new functionality for their servers – either because they need to work with other people who want that (and while I control the server, I don’t control other people), or because the software in woody is getting so outdated that it’s not compatible with other software out there, and this causes problems!

    Sorry for the rant… But please, don’t make it sound like “only desktop installs need to be up-to-date” — that is not true.

    [ Please login, or register ]


    Subject: I agree
    Author: KjetilK
    Date: Monday, 2004/05/17 – 21:29
    Yep. It is definately not a desktop vs server problem. Like you, there are many backports on my server, spamassassin, razor, chkrootkit and snort to mention a few, but also critical stuff like Exim 4, which I found I needed to combat all the MS viruses.

    My own workstation runs testing now, and I find that testing is great for me now. My parent’s Desktop, however, has actually less backports than my server: It’s just KDE 3.0.5 and OpenOffice 1.0.2 there. They are experienced computer users, but they don’t want to learn new things too often, so actually, the Debian release cycle is not bad at all for them. They’ll get Sarge when it is released, but they’ll probably follow the stable release cycle.

    I think that in some environments, such as for example my old university department, where most people really need a stable system, not neat eye-candy, and the sysadmin has enough to do with supporting heavy number-crunching, the Debian release cycle is great too.

    So, it is not a Desktop vs Server problem.

    The thing is that you have to keep certain things uptodate, things that shoot against moving targets, such as spam and exploits, and those you need to release more often.

    Additionally, we need something for those of us that want the latest and greatest (like I admittedly do on my desktop).

    I feel that the componitized distro of Progeny is the way to go.
    The problem is that you would probably need to have several generations interoperating. For example, you want a base system that’s rock solid, but occasionally, you need updates too, so you might need to support three generations of the base system.

    Fast-moving stuff like SA and chkrootkit would need just a single generation, but it needs to work on many generations backward.

    Things like KDE should also be supported with a few generations for those who want a stable desktop, in the case where you want the latest and greatest, it might require you do use a newer base system.

    [ Please login, or register ]

    Subject: Yes there is a need for change
    Author: stootles
    Date: Tuesday, 2004/05/04 – 07:51
    But not in one area.

    Let me gaze over the MS camp for a second.

    As someone pointed out there are people (many I would think) still running Win 98 and happy (using their machines) but because windows is a base, they then get App X and App Y perhaps only released yesterday, but are supported.

    IMHO Debian does require to the 2 streams – client and server (and no I do not mean a rebadged testing). For example in the server path openoffice and gnome 2.6 might be dropped entirely, but under desktop a large number of servers could hit the cutting table.

    This could then lead to platform support being also based on server desktop. For example, someone cares a great deal about x86, in the current way debian runs everything must compile on x86. With a desktop/server break down, now it needs to be sponsored on x86 server AND x86 desktop. Now if nobody gives 2 hoots about x86/desktop Debian could remove it from the desktop branch. This has to have some benefits to Debian in minimizing wasted compiles and porting (potentially).

    This is NOT the only split that would be beneficial for Debian.

    Debian could also (and/or to above stuff) split it’s package area’s (let me loosley define 3 for this post). BASE, G(RAPHICAL)DESKTOP, SERVER – BASE is BASE. BASE is going to stay in the game for an extended period. It has some really cool stuff in there like the kernel, libc, the BASE! It might see stable releases every 2-3 years DESKTOP has things like kde gnome etc. These branches get updated more often…e.g. 12months – but still with the same base, as for server. This could see releases of every 12months as well, but as the server arena goes I am thinking it would prefer to a release along the lines of base. But…

    There are exeptions to some of these applications that come to mind are spamassassing clamav and nessus. (I am syre there are some that would fall into the desktop side) these apps really require frequent refreshes. The current method debian uses sort of stops short of updating these packages as it goes against the “stable” idea. Well there needs to be a method to ensure these packages are refreshed often enough to remain effective throughout the lifecycle of the release.

    Finally, Keep this in mind as well. Large environment change(upgrade) their desktop OS maybe once every 3 to 5 years. They like this, try to manage upgrading 30,000 desktops every 12 months…just not feasable.

    I am sure if the Debian people sat down and thought how can Debian improve to benefit enterprise users? how can Debian improve to benefit Home Users? How can Debian improve to benefit admins? How can Debian improve to benefit Debian? I am sure some very smart answers will come up. Someone needs to remind Debian though. Home users want to know nothing about their computer, many admins want to know nothing about their computer (do not get me started on these morons), but if the shit hits the fan, if there was something Debian could have done to stop it, Debian would be blamed. Right or Wrong is not valid in this argument.

    I hope something I said here helps prompt an idea in someones head that becomes a benefit to the community.

    For my final comment.

    I like Debian, I like Debian enough that I have gone against the flow in my area and run Debian on all my Linux systems. I run it on my Desktop and Laptop. If no change ever happens and Debian stays as it is, I will keep running it, it’s fine by me, I only posted the above because DP asked us to.


    p.s. if anyone wants to flame could you do it in around an hour, I think I am having steak for dinner tonight….mmmmmm grilled steak.

    [ Please login, or register ]


    Subject: agreed
    Author: undefined
    Date: Wednesday, 2004/05/05 – 19:44
    read my recent comments and you’ll see that i agree with you (though, actually, as you’ll see by the date of my comments, it’s YOU who agrees with ME ;-).

    a stable base, and individual development of desktop (gnome, kde, etc) and server (courier, apache) applications. this is just a starting point. then maybe break desktop into gnome, kde, etc. we are kinda seeing this development (not in the way of releases, but) in the way packages are being maintained, with maintainer groups/teams.

    debian is too monolithic. i envision a day when the release maintainer will set a date for a release, and a representative from each maintainer group/team will be responsible for getting their set of packages ready by that date. granted there will be some interdependencies (like gnome (epiphany) depends on mozilla (gecko)), so the packages groups won’t be completely independent, but then the head of the gnome team can ask the head of the mozilla team what version they expect to have ready for the jan 1st debian stable release, and they can work towards that (making accomodations if necessary).

    i think debian needs to divide (into teams, groups, etc) and conquer this problem.

    and i have been paying attention to progeny’s “componentized”/modular linux, but i don’t know what to expect. progeny’s previous work was largely ignored by the linux community at large, but specifically the debian community. i see progeny’s work in individual packages (ie discover), but anything higher-level, like at the level of a whole distro, i haven’t seen progress. not that i’ve given up hope, but based on previous history, i don’t have any great expectations.

    [ Please login, or register ]


    Subject: Componentized Linux
    Author: hazelsct
    Date: Tuesday, 2004/05/04 – 14:58
    Gee, you’ve just reinvented Componentized Linux from Progeny. (A Debian modification which I’d guess we’ll see in Debian at some point.)

    -Adam P.

    [ Please login, or register ]


    Subject: Fantastic
    Author: stootles
    Date: Wednesday, 2004/05/05 – 00:43
    I had a quick read at the URL you pointed me to and the about section.

    You are right that sounds very much what I have in mind.

    I hope Debian stands up and pays attention to it, it very well good benefit.

    Thanks for the pointer. I’ll be watching that alot more closely now.



    [ Please login, or register ]

    Subject: no need to change
    Author: lima1
    Date: Tuesday, 2004/05/04 – 01:52
    in my opinion, current strategy is fine. it is not debians fault, that you need to upgrade desktop software so often. i know a lot of people who are still running w98 oder w2k, 3-5year old software with a few additional progs installed. version junkies can use sid or libranet. normal user wants a stable system and they don’t want to upgrade more than one times a year…highquality backports are great, but not if skilled people are then missing for the next stable release. put a stable version out every 16-20 months would be a great thing though (because then, even server software is out of date)…
    [ Please login, or register ]


    Subject: release cycle
    Author: dale
    Date: Wednesday, 2004/05/05 – 00:40
    unstable and testing are just that and I think they should
    remain as is. As for release cycles they are two slow and
    two unstructured, and backports work to a certain point but
    as it gets older it gets more and more difficult because of
    the dependencies. Also both desktop and server become way out
    of date.
    I think a release cycle of once a year on a “Debian release day”
    maybe the only way to get Debian moving again, if you can’t get
    a release out in a year then when can you?
    You may say people like Micro$oft release less often 95,98,2000 but
    remember that they charge for upgrades, if they release too often
    people get upset by costs.
    A pack of CDs (or DVD) and a donation to Debian once a year is
    affordable. I myself only donate to Debian at this time.
    Also backporting becomes much easier as the base is never more than
    a year old.
    Can I suggest that Debian Planet put up a poll on release cycles with
    options such as 6 months, 9 months, 1 year, 1.5 year, 2 year, 3 year.
    And see what people want as a release cycle?

    — Dale;

    [ Please login, or register ]


    Subject: Re:
    Author: lima1
    Date: Wednesday, 2004/05/05 – 12:33
    maybe I am a little bit naive, but I think already the release after sarge will get desktop software, that has a really polished and “finished” look and feel. I remeber times, when I compiled VERY often mozilla releases (even beta and rc-version) by myself (in my redhat times), because it was just buggy oand slow…now I don’t really care, when they release a new milestone, because I can’t find any bugs and some new features are maybe nice, but they are not really necessary. If KDE and Gnome is in such a mature state, why make the hard work of a release every 12 months? and 3 years of security updates are a nice thing (never change a running system), but I don’t think the security team would support 3 different versions…But 12 months as a goal and max. 16 in reality would be very nice…

    I know I can’t really compare windows to linux-desktop, but if gnome and kde is more mature, then apis are more stable and so backports should be very easy.

    [ Please login, or register ]


    Subject: Annual Release?
    Author: dale
    Date: Thursday, 2004/05/06 – 12:51
    Annual release was a suggestion, with testing often being in a
    usable state getting things together shouldn’t take more than
    3 months + 9 months of updates and improvements.
    Maybe another group ‘pre-release/pending’ which has proven packages
    from testing moved into it over time, so its always near release.
    Regardless of the period and mechanism I think, and I suspect many
    others, that Debian needs a faster more controlled release cycle
    without putting to much on the great volunteers that make Debian
    a reality.

    — Dale;

    [ Please login, or register ]


    Subject: Road map???
    Author: junglewiz
    Date: Thursday, 2004/05/06 – 14:07
    Does having a road map (release schedule) would help DD’s to plan their work better and achieve an aceptable release cycle. Currently not having a release schedule makes unstable / testing to be in constant flux, net effect of this is DD’s are working against a moving target and they never catch up (RC bugs etc).

    The question is whether having a shedule would help DDs plan their work better???

    [ Please login, or register ]


    Subject: From “The Tao of Programming”
    Author: BobRobertson
    Date: Thursday, 2004/05/06 – 19:32
    One day, a Manager called the Master Programmer into his office.

    “I have a project for you.” The Manager then discussed in general terms what the project entailed.

    The Master Programmer replied, “I think I can have a working prototype in 4 months.”

    “Oh, no need for a rush job…”

    The Master thought about it a moment and said, “Alright, 6 months and it should be more polished.”

    “Even that may mean rushing it. Please, just let me know when it is finished.” the Manager said, benevolently.

    “I’ll get to it, then.” said the Master Programmer, and left right away for his own office to begin.

    Several years later, when the Manager was retiring, he chanced to walk past the office of the Master Programmer. He saw the Master still hard at work. With that, the Manager was enlightened.

    [ Please login, or register ]


    Subject: Releases
    Author: dale
    Date: Friday, 2004/05/07 – 01:01
    Debian is different from a project, Debian has no end date
    just constant expansion and improvement, so comparing it too
    closely with a commercial project is wrong.
    I’m not suggesting pressure for Debian Developers in fact I
    suspect quite the opposite. Why?
    1) DDs know when releases are coming up and can plan.
    2) All I’m asking is for the last stable package set:
    if the old release had version 1.4, the last stable package
    set was 1.5 and the latest version is 1.6 but still has problems
    then the release goes ahead with 1.5.
    1.6 is backported or waits until the next release, if you really
    want it download it from sid or testing.
    3) Because of #2 users don’t get the very latest but Debian is about
    stability and if you release every year the average age of a
    package will be 6 months, or something like it.

    — Dale;

    [ Please login, or register ]

    Subject: Current structure is fine
    Author: guilherme_eco_unb
    Date: Monday, 2004/05/03 – 18:04
    I agree with people who say that the ordinary desktop user and server admins would prefer stable. Just the fact of knowing that it *will* work, no matter how many updates/upgrades you do is fabulous! I am a desktop user myself and I run Debian Woody with less than 10 backports or unofficial packages (I can recall gaim, amsn, Mplayer, dvdrip Opera, and Open Office). I think that if I have the official debian apt sources plus Christian Marillat’s, my system is complete!

    Most of my computer usage is Emacs/AucTeX, Opera, Mplayer and XMMS. I always keep a Kurumin (a brazilian Knoppix-based distro, so it is Debian Unstable) and a Mandrake installed and updated to know what is coming and to suggest to my friends the distro that best fit their needs.

    What I think that could be improved (and it seems to me it’s being done) is the installer and the config tools. Nowadays, when a friend of mine asks me to install a Linux distro on their desktops, I suggest either Kurumin or Mandrake. If Debian Woody were easier to install and configure, I would surely recommend it to my linux-newbie friends instead. Why? Because once it’s installed and everything is configured, you can keep it for a *very long* time and mantain it painlessly. A newbie can use a installed and configured Debian system… You just have to write down some basic routines and show him where his apps are. And if you really want more shiny stuff, you can always find some backports or unofficial packages that will do the job.

    Also, Stable/Testing/Unstable are precisely what their names suggest. And this seems very apropriate to me. And there has been already many talking about that so I won’t contribute to the repetition…

    Just had an idea: what about some desktop-oriented configurations scripts? Something like autoconfiguring audio permission, sudoers, firewalling and other security stuff, cd-rom burning permission, things like that.

    [ Please login, or register ]


    Subject: Autoconfig permissions
    Author: hazelsct
    Date: Tuesday, 2004/05/04 – 14:53
    For permissions, see (and help fix) adduser bugs 36019/223932 (hooks) and 147518 (automatically add new users to a configurable set of groups).

    -Adam P.

    [ Please login, or register ]

    Subject: Red Herring
    Author: theantix
    Date: Monday, 2004/05/03 – 17:28
    The names aren’t important, just a red herring. What I suggest is that the debian project use their stable releases at platforms. Each release would target stable versions of libraries and core applications, for example Debian 3.0 has a base of gcc 2.95, Python 2.2, Apache 1.3.26, and so forth. Of course each release would come with a suite of applications as they do today.

    For each release you would have two trees, “security” and “updates”. Security would contain security updates exactly as the situation is today, while updates would contain backports of oft-updated software (gaim, openoffice, mozilla, the kernel, etc) built against the foundational libraries and application versions.

    Every 1.5 years the project leaders would need to pick a new set of foundational versions, and new releases could be built around those new foundational library and application versions. This wouldn’t be substationally different in an organizational sense compared with what the Debian project is today, but for the end user it would have make an incredible difference.

    [ Please login, or register ]

    Subject: the current names and structure are fine
    Author: esekla
    Date: Monday, 2004/05/03 – 16:46
    While the actual content and/or processing can always be improved, I think the current structure offers the variety the orginal poster is asking for, and the current names describe that variety very well.

    Sometimes I wish that Testing would get more support/attention as I like the idea, but dependency holdups can be pain. I’m sure people running Unstable or Stable would have similar wish lists, but that’s natural, and one can always jump in and try to help. I don’t see any problems at all that are directly related to the Unstable, Testing, Stable structure.

    [ Please login, or register ]

    Subject: Already a fact of life.
    Author: BobRobertson
    Date: Monday, 2004/05/03 – 15:31
    Junglewiz, look into the “unstable”, “testing”, “stable” structure. I think you’ll find what you are looking for.

    Stable is server style. Long release cycles, well tested, “stable” by any definition of the word outside of horses.

    Right now I’m using Sid (Unstable by another name) for my laptops. It works quite well and is as up to date as I would hope with the obvious caveat that I am waiting for someone else to build the .debs for me.

    After long experience and keeping my eyes open, I wouldn’t run Testing except as an attempt to give something back to the community with bug reports. But again, that’s just me.

    Debian has something for everyone, if they look for it.


    [ Please login, or register ]


    Subject: release cycle hopelessly out-of-date
    Author: sam
    Date: Wednesday, 2004/05/05 – 17:41
    Funny, I found this site searching for information about how people
    can possibly be dealing with no choices other than stable and testing… and what do I find? I’m not alone in my pain.

    I moved from “stable” to testing because I couldn’t take the KDE
    bugs anymore. I resolved to help KDE out by sending in my konqueror crashes, then noticed that they wanted nothing to do with KDE2, they
    were working on KDE3… So, I moved to testing to get KDE3.

    But, I have been having more and more problems with testing. I can’t
    even update now (md5 cksum problem with the e2fsprogs, libc won’t
    install, says my kernel version is too high – and I’ve never, ever, built a custom kernel, …). I’m starting to feel I should use
    debian to get the base system, and build all my other packages
    myself, and manage them with stow.

    *Maybe* server software doesn’t change much in 2 years, but we’re
    coming up on the 2 year anniversary of debian 3.0, and its hopelessly out of date – people have been hacking like mad, there
    is tons of exciting code out there, bug fixes, rewrites, enhancements, completely new stuff (hell, ruby has gone through two minor revisions, from 1.6 to 1.8), and debian is living in the past.

    If everything you wanted in an OS existed two years ago, you’ll be fine with debian/stable, if not, good luck, you’ll need it.


    [ Please login, or register ]


    Subject: unstable / testing
    Author: junglewiz
    Date: Tuesday, 2004/05/04 – 13:36
    Unstable and testing cannot be recommended for new users. As Linux becomes more and more popular, more and more people will “convert” to Linux. Debian does not have a ready made solution for those, where as Redhat (Fedora now) and SuSE have tailor made desktop editions with shorter release cycles already, to capture that market.

    Perhaps Debian should define who are they aiming at (either server or desktop), then it would become more clearer for the potential user whether to use Debian or not.

    [ Please login, or register ]


    Subject: Debian has something for everyone, if they look for it.
    Author: reshey
    Date: Tuesday, 2004/05/04 – 12:03
    If you want a stable system with up to date software… What does Debian offer me:
    No it is way from stable
    Haha : )
    Ut of date..

    You see? We need a system for desktop and home users. Because there are no thing like that now…
    And the people who are fighting this also fought the new installer or any other step forward..

    [ Please login, or register ]


    Subject: You want what doesn’t exist.
    Author: BobRobertson
    Date: Thursday, 2004/05/06 – 18:57
    “a stable system with up to date software”. Sorry, no can do.

    Up to date software by definition is unstable.

    Debian *has* a system for desktop and home users. It is for them to decide if they want their system to be “stable” or “unstable”.

    It is perfectly reasonable for someone to install a system from “unstable” and then leave it alone, thus giving them a stable as in unchanging system.


    [ Please login, or register ]


    Subject: Yes we can
    Author: reshey
    Date: Friday, 2004/05/07 – 12:39
    Have you tried Suse Linux?
    It is stable and UP TO DATE.
    With up to date I dont mean the last point release of every singel packet, but I think all can agree on the a desktop release every year would make Debian more competetiv on the Desktop.

    If Debian does not manage to take it pice of the growing Linux desktop marked, if fear Debian will become less and less mainstream in the Linux world… Debian need to take some steps to become the desktops users choise..

    And for Gods sake, dont push the release of sarge into 2005 with al this new regulations…

    [ Please login, or register ]


    Subject: Reinstall every 3-4 months?
    Author: hazelsct
    Date: Monday, 2004/05/10 – 20:04
    So you think that having a new version for users to upgrade every 3-4 months, like SUSE, will make Debian more attractive to the average user?

    Think again pal. Only power users want to upgrade that frequently. For 95% of the world, Debian’s upgrade frequency is just fine — with timely security patches of course, which Debian provides.

    [Then there’s the issue of ease of upgrading, including config files, in Debian vs. SUSE… for another time.]

    -Adam P.

    [ Please login, or register ]


    Subject: Hear, hear!
    Author: hazelsct
    Date: Monday, 2004/05/03 – 16:16
    You’re absolutely right, stable for people who want things “stable”, testing/unstable for those interested in “bleeding edge”.

    Power desktop users want frequent updates. Ordinary desktop users do not want frequent updates. Organizations with lots of users and lots of systems to admin do not want frequent updates either.

    Upgrading lots of machines is a pain, particularly with subtle breakages from version to version. Retraining is a pain, if you have to do it every six months for each GNOME release. Speaking of GNOME, it only truly stabilizes about every two years or so: GNOME 1.4 in woody was a really mature and stable version, GNOME 2.6 is finally shaking out the GNOME 2 bugs and getting complete app support now that GIMP 2.0 is out (though GnuCash is still quite ugly, gnome-pilot and galeon are quite broken for different reasons; hopefully those will be fixed in time for the sarge release). Maybe we’ll even have Evolution 1.6 in time for stable, finally with spam filtering.

    Let’s get things truly stable, then release a great distro which people can use comfortably on the desktop or server for years. And let folk who want to bleed on the edge keep chasing after it with testing/unstable.

    -Adam P.

    [ Please login, or register ]


    Subject: Hear, Hear! Too
    Author: Peter Firmstone
    Date: Wednesday, 2004/05/05 – 06:23
    Lets give the people who put this leading distribution together a chance to get their work done.

    We can’t have too many versions being pumped out or quality will suffer, let them do it in their own time. Two years between releases is a short time considering the amount of work, just think of all the different architectures that are supported.

    I like the fact you can have one product consistent across many computers / platforms.

    By the way I’m a desktop & server user and greatly appreciate what the Debian community have produced. I’m sure the next release will be even better and will be worth the wait.

    [ Please login, or register ]


    Subject: What about a name change so t
    Author: TekMate
    Date: Monday, 2004/05/03 – 15:33
    What about a name change so testing becomes desktop and unstable becomes testing. This would make it much less confusing for users new to Debian.

    CafeComputer home of Yet another Linux FAQ

    [ Please login, or register ]


    Subject: a difficult but good idea
    Author: joelfranco
    Date: Wednesday, 2004/05/05 – 04:23
    I think that the actual scheme name is really confusing for the new users: a think called “unstable” is like insecure, or like unproved. It sounds really bad. But in fact, it is a nice name in contrast to the “stable” version of debian. The core of the question is the newbie people. To the old users, this is a good name scheme.

    Then the question is: do you want to attract new users to use debian? I think that could be just a marketing question, nothing more.

    Backing to the initial question: there exists a server and desktop releases, and i agree that it is called stable and unstable respectively (testing could be just both)

    [ Please login, or register ]


    Subject: Name change again…
    Author: joib
    Date: Monday, 2004/05/03 – 16:32
    As explained e.g. in this message a few weeks ago that wouldn’t necessarily be a good idea.

    As an example of why testing shouldn’t be recommended for Joe User, consider that of the relatively small subsection of packages I use, during the last few weeks I have had to upgrade gnucash and octave-forge to the unstable version because the versions in testing are currently broken.

    That being said, the original article does have a point. However, I wouldn’t necessarily draw some line between server and desktop use. There are certainly many server admins whose fingers are itching to upgrade to sarge (say, samba3, subversion, postgresql 7.4, LDAP/tls, etc.). At the same time, people administering large numbers of desktops (say, at some university) perhaps appreciate the stability of stable more than having the latest whizz-bang features. But the point remains; there are people who would like to run a more recent version of debian, with security support and all.

    Now, the question is, should debian have a proper, supported release for those users that want more recent software? The argument in favour of this is, I guess, basically that lots of users clearly want it. Some valid arguments (?) against it are:

    • The “Oooh, shiny!” effect, i.e. the users don’t really need feature X but the still want it cause /the saw a cool screenshot/their buddy running gentoo says it’s oh so cool/our current capitalistic society has brainwashed them into thinking newer is automatically better/insert your own favourite excuse/.
    • The distro diversity argument. If the user wants the latest-and-greatest at 6 month intervals there are plenty of other (dare I say good?) distributions to choose from. Why should debian strive to be a clone of everybody else? The free software cause is better served by diversity, when the user can choose whatever distribution fits best.
    • It’s impossible to achieve a considerably shorter release cycle. Compared to all other distributions, Debian has a huge number of packages (maintained at a very high level of quality, in general), a huge number of platforms, and a huge developer base, ( =communication problems according to F. Brooks, although one should keep in mind that packaging is not the same as development).

    Now, what to do? Sorry, I don’t know. And considering that I’m not a DD, I don’t have any say in the matter either… 😉

    [ Please login, or register ]


    Subject: Shorter release cycles….
    Author: junglewiz
    Date: Tuesday, 2004/05/04 – 13:47
    At most once a year release will do for the folks who wants to run Debian on desktop instead of other Linux distributions. Two years+ release cycle is not ideal for a desktop.
    [ Please login, or register ]


    Subject: From experience….
    Author: BobRobertson
    Date: Monday, 2004/05/03 – 15:58
    Debian was my first Linux, in 1995, and I’ve installed … ah … certainly more often and on more different hardware than I needed to since then. :^) BTW, Debian is great on SPARC hardware.

    In all seriousness, I think the names are ok. “Unstable” really is unstable. Last summer for example, during what I thought was a routine update, the wireless tools changed and my “wlan0” didn’t exist any more. Had I been Joe User, would I have known to look in iwconfig for what the new interface name was, then “vi /etc/network/interfaces” to change “wlan0” to “eth3”?

    That kind of thing can turn a user off of Debian, or even Linux, if they don’t know already how to deal with it.

    Testing has some serious problems once in a while. Entire subsystems can be broken for weeks while the developers hammer out what is and isn’t working.

    And stable is seriously stable. Run “apt-get update && apt-get upgrade” once every couple of weeks or when you see a security notification, but otherwise it will just run, and run, and run.

    Desktop systems work just fine running stable. If I were in an enterprise right now trying to support multiple desktops, I would not want the chance of something going “unstable” on me without warning. I’m also a great fan of thin clients and a central server, which Xwin does quite well indeed even several versions behind, in which case I don’t want changes to my “desktop” systems at all. It works, don’t mess with it.


    [ Please login, or register ]


    Subject: The name change again…
    Author: nomar
    Date: Monday, 2004/05/03 – 15:47
    There has been quite a few threads regarding this, and the conclusion is always that ‘testing’ and ‘unstable’ are called such because that is what they are, they are not intended to be more than development tools to produce ‘stable’.

    I use testing/unstable on my home desktops, it’s fine for home use but in a production environment there would be quite a bit of admin to review each and every change, not to mention possible re-training for updated pieces of software.

    Testing is quite stable at the moment as it’s close to becoming the stable release, however this isn’t always the case. Also, testing is often the last to recieve security updates.

    It wouold be nice to have a dedicated desktop release which only recieved security updates (without big changes weekly, daily…). This way it would be easier and more convenient to deploy Debian as a desktop in the commercial world.

    [ Please login, or register ]

    Search articles

    ·News (407)
    ·Features (5)
    ·Site News (16)
    ·HOWTOs (80)
    ·Tips (21)
    ·Opinion (29)
    ·Q & A (35)
    ·Sponsorship (1)
    ·Press Releases (5)

    Log in


    Remember me

    » Register
    » New password

    Debian Security Announcements
    DSA-943 perl
    DSA-942 albatross
    DSA-903 unzip
    DSA-941 tuxpaint
    DSA-940 gpdf
    DSA-939 fetchmail
    DSA-938 koffice
    DSA-937 tetex-bin
    DSA-936 libextractor
    DSA-935 libapache2-mod-auth-pgsql

    Planet Debian
    Wouter Verhelst: On flames.
    Joachim Breitner: Fixing my planet.debian.org subscription
    Steve Kemp: She has the blood of reptile just underneath her skin
    Pierre Habouzit: Married …
    Pierre Habouzit: whitelister 0.4 (SPF) and aaege ….
    Pierre Habouzit: kde 3.4.1 upload
    Holger Levsen: In case you are running OpenWRT
    Michael Janssen: Shiny roofs are good for the environment!
    Matthew Palmer: Work it out yourself, dammit!
    Axel Beckert: Tell me which music you like and I tell who you are

    Debian Administration
    How do I prevent rebuilt packages from being upgraded?
    Disabling the print-screen key inside X?
    Monitoring your bandwidth usage with vnstat
    Ruby on Rails on Debian
    Choice for Virtual Private Servers?
    Monitoring your hardware’s temperature
    Sending mail with Exim from ‘dialup’ IP
    How to recover GRUB Debian Sarge after reinstalling Windows
    Getting a GUI
    Spam filtering with Pyzor and SpamBayes

    Latest poll: Which release scheme should Debian follow?
    Continue this way (release when ready)
    Give up on releasing
    Split the release up
    Speed the release up
    Crank the workload up (see DebianWiki ReleaseProposals for details on these)

    Total votes: 372
    0 comments · older polls

    home · archives · news feeds · about · polls · search · sections · user account

    Powered by the amazing Drupal

    Debian Planet is not officially related to the Debian Project.
    Debian and the Debian logo are trademarks of Software in the Public Interest, Inc.