Cantor, Scott <canto...@osu.edu> writes: > I don't know enough to say what the cherry-pick warnings mean, [...]
I did a bit of googling on this one and it appears to be harmless. > For us it took weeks of time over months to get it right, but we don't > have that kind of time. Right. I did spend a couple of days on this interspersed with 24h+ re-fetches of the whole history. So some pain and frustration has been experienced, if that's what you are looking for ;-) > The one issue I did see right away is that the old tags all get > turned into a stubby branches, which we worked around in our > conversion, but I asked and it's apparently not trivial to fix, > so probably not worth it here unless others care strongly. Yes, apparently SVN tags are full branches (whatever that means). In particular, the tag commits are not part of the trunk history. So I've created a corresponding Git branch for each SVN tag and then Git-tagged the tip of each branch. The nice thing about this approach is that the complete commit history is preserved so if later anyone wants to do some more advanced surgery on this, they have all the parts. > Eyeballing it looks sound, but I think we need to do some selective > "make dist"s on the tags and do some file comparisons, aside from > the autotools files that will be different by definition. Wouldn't checking out corresponding tags from Git and SVN and then running diff on the directories (ignoring .git/ and .svn/) be sufficient? I think the only case where this might not be equivalent is if we run svn in our build scripts, which I don't think we do (but if we do do that, I don't think it's worth spending time upfront fixing that -- I find it unlikely we will ever need to do anything like this). --------------------------------------------------------------------- To unsubscribe, e-mail: c-dev-unsubscr...@xerces.apache.org For additional commands, e-mail: c-dev-h...@xerces.apache.org
