Hi there, As for Guice, I think that what JVZ said does stands: a very few people does understand how big and complex that work was (and is, since it is ongoing). Stuart did a real magic, with just a "drop in" replacement for Plexus-components backed by Guice. But don't stop there. With his foundation, it is very easy and straight-forward to create any "IoC lingo", just as Plexus "shim" is implemented. What I really loved about it, is the fact, that "reimplementing" Nexus Plugin API (and it's internals) using spice-inject boiled down to few very simple modules. Before, it was a pile of asm+deep-plexus-trickery+classworlds heavy lifting and complex, error prone code. Now it is a beautiful and simple. The previous was written by me, but I felt no sorry for ditching it, not a teardrop at all. A big +1 for it.
As for Aether, only few comments from Apache newbie. I did lurk a lot of time around Apache project, especially Maven. Was present on lists, was following the project even if I was not a "committer", had no merits (neither have now) and was not a member for a long time (today I am). But due to my private OSS project, I present with my Proximity Maven Proxy implementation a little bit, and was participating mostly on these lists. So, I am kinda aware of "The Apache Way", did seen it "in action". IMO, the decision here is more about "The Apache Way" and something else. I think everyone spent some time on Github, and I am always amazed how quickly things get picked up, changed, modded and thrown back at you there, The "speed" (in it's broadest sense) over there is awesome, like there is no mass, hence no inertia force is applicable there. Here, you have a limited (small) set of people with committer rights, with direct or indirect infra access, and even if someone is at full throttle, the project itself is doomed with this bottleneck, people cannot gain momentum unless they are "in" this circle. We saw a lot of "insiders" saying they can't do it now, they have other (like for example work related) obligations, family, etc. There, you have literally open set (kinda unlimited) of potential collaborators, and the "work" does not stop. In any moment, everybody being active are actually at the top of their momentum. But if someone does "retire" for a week or month, still leaves his work free and at disposal to others. But I have to agree with Brett, centralized issue tracking is a bliss. Over "here", I see a lot of bureaucracy, different dams that just makes things swim slower. And this is not the "git vs svn" story at all, this is more about flow. Code and idea flow. The juice. Just like Stephen Connolly told, here you hit "walls" and usually loose the momentum. When your JIRA with patch get's absorbed, you will usually have the "wtf?" moment at first when the JIRA mail lands in your mailbox. But there are lot of nice pros over here too not to be lost: things like backing organization, tied community, togetherness and all those other nice things Apache is known for. Maybe we need to merry these two? This "Apache Way" was maybe superior back then when it all started. But today, as everything is speeding up, I kinda feel this "way" just makes people... well, loose the momentum. And this applies to projects too. Aether is ASLv2, I see no problem here taking it under Apache umbrella if it's backing company shuts down. Aether as external project is really something as Plexus was (still is but I see positive opinionated people regarding ditching it). It was/is essential part of Maven, but even then it was external. I personally think, that trowing Aether into some whirling creek (like github, codehaus or even eclipse) is a very good option. Not just it will be more easily adopted by 3rd parties, but also it will be a general commissar for Maven out there in the wild. This is a problem of "merits" too. I do understand people dislike the idea to ditch some code with they were involved with for a long time, and was probably increasing/representing their "merits". We all were there at least once. But I cannot agree more with JVZ, that "merits" should melt and vaporize, should be considered somewhat temporal. The "single amount of work" made today should "count more" than "single amount of work" made a year ago and so on. At least wrt decisions made today. When it's about beer and fun, naturally the one having most "absolute" merits (without the time factor that makes it melt) should pay the beer and margaritas, beyond question, and we, without merits, should just enjoy the beer in great companionship ;) Any Maven emeritus anywhere around central Europe in near future? :D In general, and aside all this "philosophy" above, I agree very much with Arnaud. I think he nailed it. Thanks, ~t~ Disclaimer: I am involved with Sonatype. This mail does not represent any "internally approved" or Sonatype opinion, it is strictly my private opinion.
