Great Job Brett! Way to go on persevering with the svn on codeplex. Let us know how we can help with the transfer.
Thanks, -- Joe Ocaba ----- Original Message ----- From: "Brett Porter" <[email protected]> To: [email protected] Sent: Tuesday, August 31, 2010 10:17:23 AM Subject: Re: Code import I'm making progress here slowly. It will be ready this week, hopefully by the time accounts are created to be able to commit. After a few different attempts I found rsvndump to be the most reliable alternative for getting all of the SVN data out of Codeplex. I made some modifications to it to work a bit faster and avoid some timeouts retrieving logs, etc. However, it's still choking on operations on the /tags/ directory. You may be aware that we had problems copying into the tags directory due to these issues and had to move to /releases/. I did some auditing of this and have removed several failed release attempts and branches that were ultimately deleted and not used for anything, then recreated release tags properly instead of dumping it (using the original author and timestamp). So it's not all the history, but it is all of the relevant history and will actually make it easier to trace the changes back. In addition to this, I've extracted the history for the application of a number of patches that was done in the lead up to starting the codeplex project, so there'll be some additional history available linking back to the fork from the NMaven podling. Let me know if you have any questions or comments! Cheers, Brett On 19/08/2010, at 11:44 AM, Liit Padilla wrote: > + 1 on option 3 > > thanks, > > liit > > On 8/19/2010 9:39 AM, Josimpson Ocaba wrote: >> I'm +1 on starting from the NMaven branchpoint and applying the patches. We >> could use really use those histories. >> >> >> ----- Original Message ----- >> From: "Lars Corneliussen"<[email protected]> >> To: [email protected] >> Sent: Wednesday, August 18, 2010 2:18:56 AM >> Subject: Re: Code import >> >> Sometimes I missed the history from NMaven. So option three would be great, >> if you ask me. >> >> And yes I have a working git clone. I First fetched the oldest revision >> explicitely and then did the svn rebase. >> >> -- >> Message sent from mobile device >> >> Am 17.08.2010 um 14:02 schrieb Brett Porter<[email protected]>: >> >> >>> Hi folks, >>> >>> I've made a few attempts at how we might import the code from Codeplex a >>> while back. It's quite difficult due to the svnbridge that doesn't support >>> most of the mirroring without difficulties (svnsync, git-svn, hg). The most >>> workable solution to keep history intact seems to be recreating the >>> individual commits (with svn:author, svn:date and svn:log values), which is >>> something "tfs2svn" tries to do. That's been working ok, but I didn't get >>> the whole way through. I think Lars managed a git clone somehow that might >>> be a helpful basis too? >>> >>> There's 3 alternatives: >>> - direct import of latest code, leaving history behind >>> - take the codeplex revs across as a new import into the svn repository >>> (maintain history as it was at codeplex) >>> - start from the previous nmaven branchpoint in the incubator and apply the >>> changes on top of it up until today (full history since where it forked >>> from). I have the revision details to achieve this if wanted. >>> >>> Thoughts? How much history do we want to keep? >>> >>> Cheers, >>> Brett >>> >>> -- >>> Brett Porter >>> [email protected] >>> http://brettporter.wordpress.com/ >>> >>> > -- Brett Porter [email protected] http://brettporter.wordpress.com/
