on Fri Mar 02 2012, John Wiegley <jwiegley-AT-gmail.com> wrote:
Dave Abrahams <dave@boostpro.com> writes:
If you want to know the precise differences between the two, John will have to explain them to you... but if you look at the former repository you can find lots of spurious-looking branches, e.g. "trunk@38326," "trunk@38328," etc. So I wouldn't be concerned by the smaller number of branches in the latter mirror.
One difference between git-svn and subconvert is that git-svn doesn't preserve ancient branches which had the same name as current branches.
For example, you create a branch foo, then later delete it, then create a new branch foo. git-svn will throw away any history related to the first branch foo.
There are a few other cases where git-svn tosses history to make its job simpler. Case in point:
git-svn version of Boost history: 660,068 Git objects subconvert version of Boost history: 1,204,586 Git objects
Clearly a lot of something is getting lost on the cutting floor with git-svn. subconvert's goal is preserve every Subversion revision that every existed somewhere behind a branch or a tag in the resulting Git repository.
Thanks, John --- It'd be awesome if you could update the github repos' comment areas with that information. Also, I'd be happy to throw out the git-svn version and stick with the subconvert version from here on out, provided you could get the subconvert repo to update itself more frequently. What do you think? -- Dave Abrahams BoostPro Computing http://www.boostpro.com