On 30 March 2015 at 17:58, Keith W <[email protected]> wrote: > > > Its possibly worth [someone other than me, since I'm meant to be on >> vacation just now :P] discussing with infra before we proceed with svn >> changes, to find out what they think in case it should impact any >> subsequent migrations to Git that were to be made. >> > > I have pinged just such a mail onto the infra @ list. > >
You may have seen that I got an initial positive answer from Infra (Jake) to my question "after the trunk re-org, if we were later to move some components to git, would history be retained?". However, I later experimented with asfgit-clone-svn.py and a mini svn repo (created to look like ours will), and found that history was retain only as far as the move and on earlier. You can see the transcript of the commands on the infra thread. I found a couple of threads on stack overflow [1] where people had reported similar findings using the (raw git tooling). I posted my findings to Jake and I am awaiting a response. I don't have a very deep understanding of git so it is very possible I have done something wrong. However, from a java tree perspective, I see no point further in delaying the svn reorganisation. We already know svn move is the correct technique to retain history _within_ svn. Even if it turns out migrating to git can't retain a history where there has been an svn mv involving a trunk organisation, we still would not want to migrate the java tree to git in its current form. I agree with Rob's earlier comments that we will want to make further restructuring within java [separation of broker from what will be legacy client) before we consider a move to Git. If history is not migrated it will be a pain, but not insurmountable. I have raised a Jira to cover the Java tree re-org [2] which I intend to action later this week. [1] http://stackoverflow.com/questions/12938189/how-do-i-keep-svn-history-in-git-when-trunk-has-moved [2] https://issues.apache.org/jira/browse/QPID-6481
