We have a fairly large SVN repository which we are looking to migrate to perforce. We very much want to keep the ~20k revisions, the branches, etc, but in some initial tests the svn2p4 script that perforce provides wasn't able to replicate the full structure.
Have people had success with this tool, or were there others that my google searching didn't turn up? Best practices and tips are welcome.
Vitalii Pokrovskii and Mark Fridrich created svn2p4, a perl script that will 'sync and replay' to import every change set into perforce.
You can find it on the perforce wiki. They also gave a presentation about it at the 2007 Perforce User's Conference.
My biggest tip is don't do it. From a usability perspective, perforce is much worse than svn. I'm forced to use it at work, and its much less intuitive than the file based/windows explorer svn interface. Setting up the workspaces is non-intuitive and they're difficult to delete. It occasionally gets confused and won't commit something even though its been changed. Everything is read-only by default. It commits files that are checked out, but unaltered. I could go on...
I'd second the notion that it just shouldn't be done. IMHO.
If anything I'd move to GIT if anything, but wouldn't touch another source control system that is doing the same thing. It doesn't solve any problems really, and my experience leads me to the simple fact that it isn't going to save you any real time or money in the long run.
As fuzzymonk mentioned, it seems the only real option is to use the perl script svn2p4. I've used this a few times and it has worked well although slowly especially with many branches.
One thing that has been very useful with this script is the ability to minimize downtime to virtually none, regardless of geographic distance between servers. This is possible because svn2p4 is fully resumable.
This means you only have to take down the server for the few revisions that happened since your last backup. This is especially helpful if your migration is over large geographic distances (svn and perforce servers are far apart) because the large part of your import is done locally, possibly on the same machine, instead of over the internet.
We're in the middle of a large import at the moment (20K revisions, 18GB svn root) and I'm curious as to what kinds of problems you experienced in your initial tests.