Hi ivo,
Thanks for the help. Our Subversion scripts are brilliantly stupid:
- Run
svn info on the main URL to see whether or not we are up-to-date.
- Run
svn log --stop-on-copy to get a list of all the revisions we need to download.
- For each revision we don't already have, run
svn co --ignore-externals.
Once we have a local checkout of the code, we store the revision into a local Git repository. Essentially, we have a very brute-force Subversion-to-Git converter. All of our code analysis will be run on our local Git version of the repository.
So if you can survive a call to info and log, and a checkout of each revision, we can swallow your repository.
This brute-force mentality was what we settled on over other more elegant, off-the-shelf methods precisely because its simplicity means that it works with the greatest fraction of existing repositories.
We do have another Subversion importer that uses svnsync to create a local mirror. It's much more efficient, but currently we only use it with a few select forges that we know support svnsync.
Making a long story short, I'll reschedule the download and we'll see how far we get. I'm optimistic. :-)
Thanks,
Robin