Hi Titus, Istvan, this sounds like a good plan to me, for a first foray into code review, and seqdb seems like a good choice to begin with. We should distinguish two different aspects of code review brought up in this discussion thread:
- reviewing the existing code base: that's what Titus was originally proposing a solution for. - a permanent system for reviewing new code: as Titus said, I am reviewing other people's code changes when they send me patches, so the main question is who will review *my* code changes. Marek can probably do this; alternatively we may want to split it up among different people who take responsibility for different modules, like we're planning for the review of the existing code. It would be very nice to have people who know at least one module well... One warning: I did try synchronizing git with the Google Code subversion service, as a number of people have advertised. But it turned out to be quite flakey and is a great way of irreversibly corrupting both your git repository and the subversion repository. If you want the blow-by-blow story of pain, see the comments below this article (starting from my comments as "foobaron", or see the email pasted below): http://quirkygba.blogspot.com/2007/10/using-git-with-google-code-hosting.html I'm hoping that Google Code will eventually support git directly; there have been a lot of noises about that (including from within the Googleplex) so maybe that will happen soon. In the meantime let's stay away from SCM bridges that *almost* work... -- Chris From Richard Quirk: On Wed, Apr 9, 2008 at 10:06 PM, Christopher Lee <[email protected]> wrote: > Hi Richard, > thanks for all your help! I cleaned out the SVN repository and > followed > your instructions. Things looked really good right up to the final > git svn > dcommit step, which unfortunately failed with a cryptic error > message after > sending only a few commits to SVN: > Activity not found: The specified activity does not exist. at > /Users/leec/bin/git-svn line 3269 http://subversion.tigris.org/faq.html#write-over-dav Oh great: "yeah, that might happen, we don't know why". It looks like Google Code is flaky and git-svn didn't handle the error nicely. It's crapped out on me at least once with something similar (when svn returns a 502 bad gateway), but I think I fixed it somehow. Have you tried rebooting your PC? ;-) One solution that occurred to me was to do all this on a local svn repo, then svnsync that to Google Code.. i.e. instead of doing all this to https://pygr.googlecode.com/ use a pristine svn repo on a local URL like file:///tmp/repo for the "import shuffle": svnadmin create /tmp/repo ... check this out, create branches, tags, etc with svn... git svn clone file:///tmp/repo -s . ... do the all the funky steps, rebase --onto local and all that again... git svn dcommit ... commits your history to your local svn repo, which is just a file:/// .... then: svnsync init http://pygr.googlecode.com More info here http://svn.collab.net/repos/svn/trunk/notes/svnsync.txt You'd probably have to ask the Google Code guys to reset your repo to 0 so you could do this.. Then git-svn clone your "new" Google Code subversion repo, and fetch the lot again. This can't possibly fail! Except: the time stamps will be wrong (now instead of 4 years ago), the authors are going to be messed up (you, instead of the real author), etc... :-( > Holy crap. If this picture is accurate, git has lost the entire > history of > commits to master from the point at which the git svn dcommit failed > (commit > #5 on the entire master branch, back in 2004). As expected, my > Google Code > SVN repository looks like that very, very early version of the tree... It looks bleak. However, I get the feeling that there must be some way to get the history back and carry on - it's just a matter of figuring out what. I'm tempted to suggest the git mailing list - someone there will surely know how to fix it. Ah, you are using a new version of git, right? Like 1.5.4 or 1.5.5 even. I'm pretty sure git-svn "hides" the history in a place it knows where to get to, so there's probably some mystic command that sets your master back to how it should be. And then you can try the svn dcommit again later. Hopefully... > My options now: > - I have a backup of my git repository from before I started this > attempt > to use git svn. I guess I'll just revert to working with that, as > git seems > to have hosed this repository. Thank goodness for this! I was reading the first part like this "Oh no! I've made this guy lose his whole history... he's going to hate me..." > - Starting again from that base, presumably I could clean out the SVN > repository yet again, and try to follow your instructions correctly > from the > top. However, will the fact that my SVN repository is not truly > empty, but > instead has successive commit + removal passes screw this up? Hmm, interesting point - but I don't think it'd matter that much, after all you're just bunging more changes into SVN and git doesn't care what happened in the past, just how it got from one point to the next. > - I could give up on git svn / hosting on Google Code, and just > distribute > the git repository directly (e.g. on a web server). That means anyone > wanting the very latest source code will have to use git. I'm > inclined > towards this, as git-svn is starting to seem like a much less robust > proposition than I had hoped. I still think git is the way to go, > since I'm > starting to get involvement of lots of other developers in my > project... > > Sigh. Do you have any advice? It's looking like vanilla git is going to be your best bet, isn't it? Something like http://repo.or.cz/ or http://github.com for source code and google code's site for the issue tracking, downloads and wiki. The Google Code setup now allows you to set the source tab to point to alternate instructions, so it would be fairly well integrated. Git has more advantages if people send you patches too - it keeps author info and makes sending patches to mailing lists easier - but Windows users may not be able to get as involved. Good luck! --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "pygr-dev" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/pygr-dev?hl=en -~----------~----~----~----~------~----~------~--~---
