Home » Uncategorized » E-mail I sent to Mark Shuttleworth

E-mail I sent to Mark Shuttleworth

Hi;

I like Ubuntu very much, but I find it annoying how behind the curve you guys are with your releases. Jaunty is the first release of Ubuntu that has Mono debugging support out of the box. Jaunty will ship with Mono 2.0, which was released October 6, 2008, yet Mono 2.4 has just been released! I’m going to have to wait till October to get the Mono bits that were released in March.

Here is a list showing more examples of how Fedora is more up to date:
http://subbisays.blogspot.com/2009/03/ubuntu-904-vs-fedora-11-lot-can-change.html

In general, new software is better and more reliable than old software. You guys spend lots of time backporting fixes from newer builds that would be solved more efficiently by just taking newer builds! You guys also aren’t helping advance the state of the art by working on old software. Novell could care less about bugs in Mono 2.0.

If you guys ship every 6 months, using software that is 6 months old, did you really ship on day X, or 6 months ago and just sit on it?

I realize you guys have a tradeoff between stability and freshness, but I think your team is not making the right tradeoff, and I see this as a problem that crosses many teams. If there are any problems (that likely affect just a few customers), you can fix them right after release. What is the whole point of having this infrastructure of repos and backports?

-Keith


5 Comments

  1. Disclaimer: I’ve had a few to drink and felt like commenting …

    This strikes me as a a symptom of too much of the Ubuntu system being encumbered by the central repositry paradigm imho …

    If ‘linux’ software distribution became more author-centric rather than distro-centric then authors could get their latest bits out to users more easily. Right now we have to trust the so-called benevolence of the distro decision makers before we are permitted to easily use new software.

    I guess its a double edged sword though…

    I have grown fond of the increased stability of recent Ubuntu releases, to the point where i am growing less excited by the release… beacuse, well, things more or less “just work (for me)” sweet, such that i ain’t tempted to anything else at this point.

    TO BE FAIR, though, maybe you should have sent two emails — the one you did to Mark, and another to the Mono crew. Ubuntu apparently has the Lions share of the Linux desktop usership… so why is it that Mono makes it so difficult to get their latest and greatest on Ubuntu? After all Banshee (a Novell afair also) are pretty proactive with regard to making their stable releases easily available to Ubuntu users via Launchpad’s PPA’s.

    my 0.02. peace!

  2. Oops, I never get any comments on my blog, so I didn’t notice this for a while!

    It isn’t just the benevolence of the distros that we must trust, but their competence.

    Your idea is interesting, and I’ve thought about something like that, but for the cases of inter-component dependencies. You can’t swap in a new mono, without also potentially needing to plug in a new Tomboy, Banshee, etc.

    But another way to go about it is to have more people with the right to add stuff to the repositories. Mark should not really be deciding what version of OO or Mono, etc. to put in the repos — it should be “us”!

    Yes, it is true that the reasons to upgrade from one Ubuntu to the next are less interesting. I do it partially because I consider it my job to stay up to date. That is a good sign.

    But I don’t just consider this a Mono problem.

  3. @Matt
    I can conceive of one practical reason why the “distro-centric” approach is dominant, and not the “author-centric” one.

    Normally, the 3rd-party developer writes the code, and the distro maintainer takes that code and integrates it with the distro. Now, imagine if now the author himself is responsible for distributing the software to the end users of each distro – he would have to ensure it runs satisfactorily on a variety of distributions. Needless to say, that will most likely never succeed; there are simply too many distros with many differences for which to account. Unless you somehow unify the entire GNU/Linux community behind a single distro, letting each distro maintainer apply the specific changes/patches is more realistic, IMHO.

    Some developers certainly do provide their own binaries for major distributions, but the majority (especially those with small-scale projects) don’t.

    (Hopefully, I didn’t misunderstand what you meant by “author-centric”…)

    @KeithCu
    “But another way to go about it is to have more people with the right to add stuff to the repositories. Mark should not really be deciding what version of OO or Mono, etc. to put in the repos — it should be “us”!”

    I suspect that these limitations are partly a man-power problem. Obviously, maintaining the repositories isn’t simply a matter of “dumping” new releases; we don’t really need more people who can add stuff, but people who can test and to fix the candidates. If we can get through the testing phrase quicker, then it may be possible to have more recent versions without more breakage.

  4. I think the problem is that the release managers or others serve as a bottleneck / impediment. They tend to say “no” to be safe, even though the package maintainers know better what is in it, whether it fixes other problems, etc. They could even slow development down by suggesting that the package maintainer backport fixes from the new version rather than just taking the new version.

    I’m not suggesting people “dump” stuff into the repos. If there is a problem, the person who added it needs to fix it quickly. That is all part of the responsibility.

    I don’t see that there is a testing phase with Ubuntu.

Leave a comment

Your email address will not be published. Required fields are marked *