I'm not sure if this has come up before, but having been actively involved in two projects that entered the incubator and eventually graduated (Mesos and now Spark), I would like to start a discussion about the the pain that comes with a 2 phase path for email list migration for projects that go through the incubator.
When a mature open source project enters the incubator, it has to move mailing lists from its existing solution (which for Spark was Google Groups) to the apache incubator lists. This was a painful move for Spark since many devs found the Google groups mailing list interface easier to use, but I understand that as Apache, we value having all of our projects' bits live on our own Apache infra. Also, once we discovered and set up nabble, we had back a lot of the features we found most useful, especially being able to see all emails in long threads (based on email subjects) all on one page. However, as Apache, we don't ask the project to move mailing lists only once, we ask them to move twice: once when entering the incubator and once at graduation. Though the second move is easier, it is still painful in our case since (unless I'm missing something) you can't migrate a nabble mailing list from the @incubator.apache.org to @apache.org and so now the history of our projects email archives will be spread out over three different locations. Maybe other incubator projects haven't found the 2nd move to be very painful and maybe search engines make this not a very big deal. I'm just wondering if there might be any alternatives to this 2 phase migration process that we as Apache could consider in the long term? Andy
