Folks, Really, after the first round of discussions on this subject, I didn't think that Mahout was a viable target. There just isn't a good alignment. However, when I over-summarized that discussion in an email to the members list, Grant called me on it. So I felt that I should trot it by one more time. It's really fine to reach the same conclusion.
--benson On Thu, Nov 19, 2009 at 7:55 PM, Ted Dunning <[email protected]> wrote: > For parking status, labs or even google code seem appropriate. > > I am not as unconvinced by the procedural arguments as Grant, but I do > worry > that there isn't either a consumer or a committed champion for the code. > That combined with the language gap would lead me to expect the code to be > pretty much orphaned. If that happens, I doubt that there is any advantage > to having the code in Mahout for either side. > > I DO think that the code is an important resource to preserve one way or > another. > > On Thu, Nov 19, 2009 at 4:49 PM, Benson Margulies <[email protected] > >wrote: > > > Grant, > > > > As someone who was given committer status on a project in a nonstandard > > way, > > I have, perhaps, more flexible morals about the principles than you do. > But > > I certainly have no intent to try to convince you that Mahout should > > deviate > > from normal community procedures if you are not so inclined. > > > > This is a high-speed implementation of > > http://acl.ldc.upenn.edu/J/J92/J92-4003.pdf. Both Brown and Mercer are > on > > the author list. > > > > Yes, the author is not I. He works with me. > > > > It may be that the best solution here is for me to park it at the labs > > using > > my status, and act as Jethran's agent. That would be a lot less work than > > teeing it up here, and I think that lazyness will win the day here. > > > > --benson > > > > > > On Thu, Nov 19, 2009 at 5:38 PM, Grant Ingersoll <[email protected] > > >wrote: > > > > > > > > On Nov 19, 2009, at 1:53 PM, Benson Margulies wrote: > > > > > > > Folks, > > > > > > > > We have the MI cluster code to contribute. It has been suggested that > > > > labs.apache.org might be a good target. However, topic-wise, Mahout > > > would > > > > make more sense. > > > > > > > > This poses two questions. 1) Does Mahout want the (small) distraction > > of > > > > this at the moment? 2) Is the PMC willing to do something a little > > > > out-of-the-ordinary with committer privs, as I'll explain below. > > > > > > > > to quote: "Apache Labs was established to provide a place where > trusted > > > > individuals can innovate without the burden of community building, > with > > > the > > > > comfort of the apache infrastructure and remaining close to their > > peers, > > > > making it easier to get feedback and seeding new potential incubator > > > > podlings." > > > > > > > > I think that I can make a strong case for the author as being > > trustworthy > > > at > > > > labs. I'm reasonably sure that I can make a case for *myself*, at > > least, > > > in > > > > a pinch. On the other hand, I don't think it's viable to expect him > to > > > > become a committer at Mahout via the usual process, and I'm not > > > comfortable > > > > with contributing the code and not ending up with commit rights. > > > > > > You know how the ASF works. Merit is earned. Many of us started out > > > making contributions w/o said right, trusting that if we keep it up, we > > will > > > be made committers. > > > > > > > > > > And I'm > > > > just not finding the time to lend enough assistance here to have much > > > chance > > > > of becoming a committer. > > > > > > > > So, I'm thinking that labs is the right path, but I wanted to give > the > > > > Mahout community a chance to think about it first. > > > > > > Labs is all about existing committers. There is no "convince labs to > > make > > > [the author] a committer.", AIU how labs works. Please confirm on the > > labs > > > mailing list. I also don't know if Labs handles "donations" of > complete > > > code. Of course, you probably could do it on labs, but that means you > > are > > > the committer. The ASF doesn't like proxy committers, either. > > > > > > While I very much think this code is in line with Mahout and would love > > to > > > see it here (I especially think it fits under the "Mahout as a TLP" > > model, > > > which hopefully will happen sometime in '10), I think we need to know > > more > > > about it. My suggestion would be for the author (from the sounds of > it, > > > it's not you) put up a patch that integrates it with Mahout (likely > just > > the > > > build system or tree structure) and demonstrates knowledge of how open > > > source and the ASF work by seeing the patch through to committal by one > > of > > > the existing committers. Doing so would also suggest that they can > speak > > > for themselves, which doesn't appear to be the case so far. Given > > > reasonable proof of that, which should be pretty easy to do if the > author > > is > > > really serious about maintaining the code and being a part of Mahout, > > > committership should follow shortly. In some cases, the PMC has voted > to > > > add people along with a donation (Sean and Taste for example, which was > a > > > major donation, including a Software Grant), but it is not something we > > > necessarily do outright, otherwise everyone would be petitioning for > it. > > In > > > the case of Taste, it was already out in open source (SourceForge), had > > an > > > established community which Sean had built and it was clear that Sean > > knew > > > what being a part of an open source community meant. > > > > > > Hope this helps. I do think it would be useful to have in Mahout, but > > I'd > > > like to hear more. > > > > > > -Grant > > > > > > -- > Ted Dunning, CTO > DeepDyve >
