On Fri, Aug 15, 2008 at 9:21 PM, Robert Greig <[EMAIL PROTECTED]> wrote:
>> and I think review-then-commit >> actually requires less discipline, and the majority of it is focussed >> on the person with the biggest stake (the person writing the patch). > > The issue I have is that with the current toolset it is a pain and I > know I would hate dealing with patch files on jiras. People will want > the code committed so that things can progress (like dependent changes > etc) and if there is any delay we may get into an annoying merge > cycle. Provided that reviews happen promptly, this shouldn't be a big problem. Yes, there'll be a bit of a delay, but it shouldn't be huge in the normal course of things. If people aren't rigirous in following through, then there's immediate impetus for somebody to sort it out, rather than just letting it rot. The pain caused by laziness is still there, but it's up front and obvious and doesn't get a chance to build up to the "oh good grief, we need to review a month of commits" level. >> This should make it more likely that process is followed. > > I think we could determine how effective it is by monitoring the > number of issues in the "review pending" state. We have a good number I suspect that we would keep up with it for a while, but once the initial surge of enthusiasm wears off it would build up and start to become problematic. I could be wrong, but that's been my experience in the past, both on this project and on others. It's basically what's happened with the weekly calls. > of people getting paid to work on this project so those organisations > should be able to insist their staff follow the process. All committers should follow the processes we agree here, regardless of their employer. One Team, One Project, One Dream. ;) >> svnmerge in 1.5 is better, but it's still nowhere up to the standards >> of a proper DVCS (it still loses history, for one thing) > > My point was that at last it's making progress in this area. I haven't > tested it out myself yet although I shall be doing so in the next > couple of weeks. If svn gets better merging, that's great, and will make review-then-commit easier, since people can just take feature branches if they really, really need to get patches out of their tree and don't want to use git-svn. Hurrah! - Aidan -- Apache Qpid - World Domination through Advanced Message Queueing http://cwiki.apache.org/qpid "Nine-tenths of wisdom consists in being wise in time." - Theodore Roosevelt
