On Friday 19 November 2010 22:19:56 Boyd Stephen Smith Jr. wrote:
> In <[email protected]>, Arno Rehn wrote:
> >Maybe the initial svn2git rules are not correct so that git somehow thinks
> >the commits would be related... but some of those rules would then be
> >REALLY wrong, which I don't quite believe.
>
> I tend to doubt that. I think filter-branch preserved empty commits that
> weren't at the tip for some reason. The documentation seems to imply that
> keeping empty commits is the default behavior. --subdirectory-filter does
> imply --remap-to-ancestor, but that will only drop empty commits near the
> tip.
Pushing the branches one at a time to a clean repository, I've found out what
goes wrong. Here's what happened:
Prior to 4.0 the kdebindings layout looked like:
kdebindings/
...
qtruby/
korundum/
....
Shortly before 4.0 this was changed to:
kdebindings/
...
ruby/
qtruby/
korundum/
...
So if I --subdirectory-filter for the ruby/ subdir, the filter will only match
after this change has happened. In everything < 4.0 it obviously can't match
anything and thus all those branches simply stay the same, with all the
history from every other subdirectory.
I don't know yet how to solve this best, but at least I know now what's going
wrong.
--
Arno Rehn
[email protected]
_______________________________________________
Kde-scm-interest mailing list
[email protected]
https://mail.kde.org/mailman/listinfo/kde-scm-interest
