Recommended Builds

Discussion in 'Bukkit Help' started by TnT, Mar 28, 2011.


What should be done with the RB's

  1. Leave as is - I like the system as it exists today.

  2. New RB every week

  3. Scrap RB's - who need them?

Thread Status:
Not open for further replies.
  1. Offline

    TnT Retired Staff

    As there has been some discussion around RB's - here is a poll. The devs will be watching, so if you want change - make a good case for why it should be changed.
  2. Offline


    add option ban trolls pls :)
  3. Offline


    New RB every week, if there wasn't one in shorter period. (something like this)
  4. Offline


    Release an rb when they have an rb, it might happen every week twice a week or once every 2 weeks, an rb is about a somewhat stable and functional build that can be a target for everyone to use. There is no way to determine how often that will happen.
  5. Offline


    Yeah, pushing a release as an RB just because it's been a week is a terrible idea. Why? Because then you're stuck with a semi-stable build for a week before another RB is tagged.
    As it is, most of the commits between 556 and 592 were done in a 5 day period. People are just impatient. I would MUCH rather the devs only tag what THEY find to be stable as an RB instead of just whatever was up on Sunday. That makes having RBs just about as good as not having them.
  6. Offline


    New RB every week sounds good. Hope they will do it.
  7. Offline


    I'd like to see RBs posted more often, but don't rush if you don't have it completely done. RBs are great for those that want to run their servers without many problems, and is a good base for plugin authors to test their plugins as. Don't scrap the idea.
  8. Offline


    I totally agree. Why muck up the idea of a "recommended" build by making it "weekly"? Defeats the whole purpose of being "recommended".

    Besides, currently anyone can go grab whatever build they want, be it weekly, daily, whatever! No need to change anything.

    Furthermore, this would make maintaining plugins to support "recommended" builds even more cumbersome. :eek:

    And finally, this might make (Craft)Bukkit developers hesitant to make changes or add new features knowing they have a week (or less) before the "bukkit world" consumes it and starts complaining about whatever they may have broke or changed. ;)
  9. Offline


    So like they do now? The bulk of the commits were done during a 5 day period. They're working on large changes and probably don't want to mark an RB until it's done.
    Nobody has given a reason for wanting RBs more often.
  10. Offline


    I voted for weekly RBs if and only if the Bukkit Devs are willing/able to make sure the Sunday RBs are as stable as possible. Sunday could be the "deadline" for the devs to work towards. In fact, would it be possible for the Bukkit Team to let plugin devs know ahead of time what they think will be the next RB? Eh, I think I'll just make a weekly schedule for what the devs could work off of, as that would illustrate my point better:

    Friday-Saturday: Bukkit Devs let plugin devs know what they expect the newest RB to be, plugin devs prepare to have updated plugins by Sunday-Monday
    Sunday: Devs release RB, plugin developers continue work on updating
    Monday: By this time plugin devs should have updated to latest RB
    Tuesday-Thursday: Bukkit Team and plugin devs take into account ALL of the reported bugs, work on committing fixes to those bugs before the "deadline" of Sunday

    This seems like a reasonable timeline, in my opinion. What do you other plugin developers and Bukkit team members think of this plan? Is it reasonable? Is it possible for you to stick to on a weekly basis?
  11. Offline


    @aPunch Throw in 90k a year and you've got a deal. This isn't a job, it's an open source community project. There are no deadlines.
  12. Offline


    I too voted for a new RB every week. Perhaps a week is too short, but they need to come out more often than they do now. There are tons and tons of fixes in 592 that 556 simply does not have, but it's not marked as recommended. And I personally think at this stage of development, pushing fixes is a lot more important than having perfect stability.

    Also, at this rate, 1.4 is going to be out before we get a new RB.

    EDIT by Moderator: merged posts, please use the edit button instead of double posting.
    Last edited by a moderator: May 13, 2016
  13. Offline


    Third that.
  14. Offline


    keep it as it is because plugin devs would have to work on updating for the new build instead of making their plugin better
  15. Offline


    @Drakia How does 75k sound?

    Yes, I realize my deadline proposal seems unreasonable now. I've never had a job before as I am but a small child, so I had not fully grasped the fact that people have other things in their lives to do. I am just trying to propose some solution that would be good for the Bukkit Team and the community. As a community, "WE WANT MOAR UPDATES!!!1one". In the end, however, the Bukkit Devs make the call on what happens. I still stand by my wish to have weekly recommended builds, but really it is only a wish, to be honest. I have no legitimate evidence that weekly RBs would be better. In all honesty, I wouldn't mind if it stayed the way it is. I love Bukkit and enjoy tracking it's progress.
  16. Offline


    cant you make it as a header so that more people can see it?
  17. Offline

    EvilSeph Retired Staff

    Guess I should post this here, too:

    There appears to be some confusion as to how the Recommended Builds system works, so I hope this clears it up. With the Recommended Builds system, we try and aim for a new RB being promoted every week or so, but nothing says we absolutely have to stick to this schedule.

    The system is designed so that even if we were to never promote a new RB ever again, people would have a stable server to deal with and things will work fine. As such, plugin developers are NOT expected to update anything past the current RB. HOWEVER, we hope they are still actively following development to prepare their plugins for the next RB, if and when it comes.

    We try to give everyone as much of a heads up as possible so that when the next RB hits, people aren't scrambling to fix their code - it should already be fixed - they're just scrambling to release their newest builds. Ideally, plugin developers should have a development build system of their own (like so that people can help them test their plugins with the latest Bukkit development builds so that everyone is ready for the next RB, but we understand if some plugin developers don't have the time, want, care or resources to do this.

    A lot of people have asked us to slow down on developing new builds and that is what the Recommended Builds system allows us to do.

    When we release Bukkit, we'll begoing with a "release early, release often" attitude employing a Major.minor.bugfix versionng system where:
    • Major version increments will occur when breakages occur.
    • Minor version increments will occur when new features are added, but backwards compatbility is still retained.
    • Bugfix version increments will occur when small bugfixes are made.
    Development will be ongoing, but releases will be version based and staggered.
  18. Offline


    I'm just going to copy my other post here as well:

    I don't understand the people complaining about how long it's been between RBs, especially suggesting an RB be released every week. The brunt of the pushes between 556 and 592 were posted during a 5 day period, that wouldn't even be covered in the 1-week RB posting period.
    As for RBs slowing development, that's just BS.
    The devs push what they can, and when THEY find it stable they mark an RB. Do they need 5000+ people using the cutting edge complaining about a bug they know exists that they haven't gotten around to fixing yet? No.

    As a plugin dev, I only develop for the latest RB, I keep an eye on the git commits for Bukkit though, so I'm ready to update when it's marked as RB, but I don't bother changing my code until that point. Why? Because look at the three (Maybe more now?) builds that completely broke every plugin in existence. And that's another reason to use RBs, in a span of three days, they changed the plugin interface not once, but three different times. Think of how that would have gone over without RBs, you'd have plugins supporting ~12 different builds because nobody updates every time a commit is made, servers wouldn't be able to get everything working together, etc.

    Even WITH the RBs you get people who just don't listen, they'll run 492 and ask why your plugin doesn't work with it. The last thing I want as a plugin dev is people asking me why my plugin broke between arbitrary commit A and arbitrary commit B.

    RBs give plugin devs a target, before they implemented them we just picked a build at random, and hoped that things didn't change. Now we have something stable to aim for, and yes, it's been a week since the last RB (I don't count from the 15th to the 23rd as barely anything was commited during that time) but that's because the devs are currently working on implementing LARGE changes, not something they can get out in a day or two. Give them time, and when they're ready to say their work is done, they'll mark a new RB.

    Also, just because I find it funny, here's the commit times for the latest builds (Since a lot of people probably don't actually _look_ at this, they just see the number go up 40 and go into whine mode).
    Mar 17 - 21 : Build 557 - 560
    Mar 23 : 561 - 565
    Mar 24 : 566 - 571
    Mar 25 : 572 - 577
    Mar 26 : 578 - 587
    Mar 27 : 588 - 591
    So to everyone saying the number has gone up so much, it's only gone up that much in 5 days, give them time to finish the large projects they're working on.
  19. Offline


    Please keep releasing your RBs only when its safe to do so, not just to try to meet a weekly deadline. Has worked wonders for our server with near perfect uptime.
  20. Offline


    I vote for scrapping but I vote against weekly. Picking an arbitrary build isn't an improvement over no build, and I'm not for deadlines like that, too many variables.
Thread Status:
Not open for further replies.

Share This Page