On Tue, Oct 5, 2010 at 10:38, Thomas Zander <[email protected]> wrote: > On Monday 4. October 2010 21.34.30 Niko Sams wrote: >> Another thing I found was abandoned applications that where moved into >> playground or >> unmaintained. With a monolithic repository this would not be possible. > > In svn there is a usability bug that if you delete an application its terribly > hard to find out which revision the deletion happened in. To get around this > problem in KDE we first had a wiki page with the name and then the revision > number of the deletion. > Unfortunately the task of writing the full url for that checkout was still to > obscure and even the experts needed to think a while coming up with the magick > checkout line. > So in the end we got the current solution; which is to move code to an > 'unmaintained' repo. > > In git this problem totally goes away; an application inside a module like > kdeedu can easily be tracked down in the history and revived (it even doesn't > cause the repo to get bigger). I don't see the difference to svn here. Why is it easier to track the application down in the git history than it is in the svn history? And even if it's possible - it's really good hidden in the history - making it hard to find for developers who want to continue working on it.
>> The remaining problem for split repositories are cmake modules (they >> could be copied >> into the applications that need them), libkdeedu and keedu/data. > > There are some more problems, see > http://lists.kde.org/?l=kde-scm-interest&m=128394136707263&w=2 > I was more talking about problems with writing the rules. Niko _______________________________________________ Kde-scm-interest mailing list [email protected] https://mail.kde.org/mailman/listinfo/kde-scm-interest
