I don't think unmaintained code should stay in our codebase. This will only create frustration amongst our users, as they will not get questions answered and bugs fixed. It would also be an obstacle for a 1.0 release, where we have to guarantee more stability and compatibility.
If someone wants to have a look at removed implementations, they can always check them out from older SVN revisions. Maybe we should provide a wiki page that describes how to do this. Best, Sebastian On 02.06.2013 09:36, Stevo Slavić wrote: > Hello Mahout community, > > Multiple algos got axed for various reasons, e.g. more scalable solution > exists and got implemented, or algo is not really used much in real world > for whatever reason. I'd like Mahout developers to consider not removing > those implementations but to create a separate submodule for them, or a > project, and minimize maintenance effort for that part. Even those > algorithms do not meet project main goals and are not best for running in > production, I believe it would be useful to have them in the project for > educational purposes, machine learning in general but also to learn Mahout > (APIs, command line) on presumably simpler, more widely known algorithm. > > Kind regards, > Stevo Slavic. > > > On Sun, Jun 2, 2013 at 4:07 AM, DaiQinghao <[email protected]> wrote: > >> hello sir,may I ask what dev support do you expect? >> >> --Best Regards,Qinghao >> >>> From: [email protected] >>> Subject: FP Growth >>> Date: Sat, 1 Jun 2013 23:15:56 +0200 >>> To: [email protected] >>> >>> FP Growth seems to not have a lot of dev support. Are there users out >> there using it? Should it live on or get the axe prior to 1.0? >>> >>> -Grant >> >> >
