At the risk of starting an endless rant, I'm wondering if anyone is using Subversion (SVN) for source control?
I have been a Visual SourceSafe user for years, and it constantly worries me. In the early years, before version 6.0, it would crash and die, destroying your database. For the last 3 or more years it has worked very well. But I hear stories...
Now we are considering conversion to SVN and our first trials with it are extremely trying. We have the latest released version installed on a server and are using TortoiseSVN on winXP as a client and the speed is unacceptably slow. How slow? So slow I wonder if the install is bad.
On Saturday I used vss2svn to convert one of our small VSS databases to SVN. On a local machine the project is 2400 files totaling 38 MB. Archived the VSS database is 88 MB. It took 45 minutes to convert the db to a 310 MB dumpfile and then well over an hour to load the dumpfile into SVN. I gave up and went home while it finished.
Today I created a local folder and used TortoiseSVN to export a project tree from the server onto my desktop. About 30 minutes later I killed it because it had only loaded 63 of the 2438 files onto my machine. It really consumed the machine and the network in the process. It had loaded 1.3 Mbytes of the
38 MB project. Visual sourcesafe loaded the entire project from the same server in under 3 minutes.In looking at some Task Manager performance indicators, VSS was only using about 10% of the network bandwidth and CPU usage as subversion.
As for details, we have svnserve running on the server (not apache) and we are making file accesses to the repository using the file://x:/SrcRepos scheme.
Is this unusual or typical for what others are getting?
Scott