The survey has turned out to be very useful. I should have done this last time around. We have 19 responses so far. I get a nice summary page with charts and graphs that I have not yet figured out how to share publicly. However, I can share the raw survey data as a spreadsheet:
We should wait a few days to let more people respond, but the initial count for question 1 (would you like to see VXL migrate from subversion to git?) is
13 - Yes
2 - Maybe
1 - No
3 - Don't care
I'll address some initial concerns from the survey.
Ian is concerned about maintaining existing VXL history and approaches to navigate history to investigate regressions. If we convert to git we will maintain all existing VXL history including tags. I believe old SVN revision numbers will be encoded in the commit messages. Brad King will do the conversion. He is a git guru and has done this conversion for several other big projects. Brad also did our CVS to SVN conversion and our recent SVN server migration. As for working with history to find regressions, git has powerful tools to help with this (e.g. git bisect). However, you are correct that it will take some time to get used to this. Git does require a different mindset than SVN or CVS, but once you get used to it, it's very hard to go back.
Peter Carr is a VXL user who is looking to become a developer. Peter is concerned about SourceForge support for features like forking and pull requests used by GitHub which make it easier for non-maintainers to contribute back to projects. We have not yet worked out the details of our git workflow. That is another round of discussion after we have decided to adopt git. However, since git is distributed, one option is to mirror VXL on both SourceForge and GitHub to use features from both services. This will require further investigation and discussion.