Bryce:
Would someone mind doing a quick test of cvs2svn on the Inkscape codebase and see if it works cleanly, and if not, what sorts of issues we may need to sort out?
I did a run on two other codebases: djvu and cimg
First summary: cvs2svn did not have problems given the right options but is somewhat slow (python). Interpolating from the larger run projects a runtime of one maybe two hours for inkscape on a 1.5GHz machine. The version used was 1.3.0 which appears to be 30% faster than 1.2.1. The repository size appears to increase up to one third (Berkeley DB). The process is not automatic as there might be conflicts from identical tag/branch names, where a restart with additional options is necessary. Also, downloading from SF stalled frequently, even with wget -c, and had to be interrupted/continued.
Now who will give me access to his broadband-connected 3 GHz machine? 2 GB are likely needed. I will install subversion (user space), cvs2svn, download the 87MB repo, convert, and report.
cimg djvu inkscape ---------------------- CVS Repo size tar.bz2 (blocks) 6556 13916 87000 CVS Repo size unpacked (blocks) 13772 71776 ? Number of CVS revisions 1039 18531 ? Time to create SVN repo dir (s) 34 675 ? Number of SVN commits 332 3828 ? SVN Repo size (blocks) 14684 96252 ?
ralf