Hi there,
I've followed the latest discussion a little (not in
every detail) and just have to add this, I think:
Whatever provider/host you will choose (or stay at SF), there is no excuse not to make backups!
Every single storage can fail. Accidentally, intentionally or for whatever other reason.
AND, as backing up the repository is very easy, there is one excuse less
svnrdump dump svn://svn.code.sf.net/p/shuttleultra/code/trunk > SSU.svndump
...that's all.
It takes a while depending on overall throughput, but after that you have a
full backup of the repository[1], that can be very helpful in case of a failure.
Needless to say, but backups only
really make sense if done regularly.
So, when there are 4 main developers and every one does a kind of backup (more sophisticated incremental backups might be better, though) the repository / history is pretty save.
Just my 2cents,
Kuddel
[1] in this example only /trunk/ is remotely dumped, so it's actually not a full full backup