Team Foundation Server 2010 and Crucible

[Jonathan Reckner]

I am adding some new information to an old article: http://forums.atlassian.com/thread.jspa?threadID=40957

I have spend the last few days attempting to get Crucible wired to TFS 2010 using the SVNBridge. The last comment posted on this issue shed some light on a source code issue with SVNBridge itself.

svn log -vr 0:HEAD file:///path/to/repo/@HEAD --limit 1

The author then explains the following:

The results of the above command should yield the first changeset of that part of the repo but its returns the HEAD instead. This is because of a known issue logged against SVNBridge here: http://svnbridge.codeplex.com/workitem/12861?ProjectName=svnbridge which shows SVNBridge does not understand the difference between 0:HEAD and HEAD:0 because it does not respect sort order. I have added my findings to that team as well.

But I have fixed the sort order issue for SVNBridge. However this seems to be the first layer to the onion. Once a list of changesets was correctly received from TFS those change sets failed PROPFIND requests for the files in those changesets with "premature end of file" exceptions for each and every file within all of those changesets.

Is there anyone willing to work with a fixed SVNBridge to help narrow down the communication issues between SVNBridge and Crucible? I would likely need a list of svn commands being issued to the svn server during the indexing process and the expected returns before I can narrow down the issue at hand. I can provide a wireshark capture of the returned data during the indexing process using TFS 2010 and SVNBridge. Maybe the returned results can shed some light on why Crucible does not like the returned information from SVNBridge.

[Conor]

I cant help you with SVNBridge to any significant degree but I suggest you connect FishEye to a svn repo with debug logging turned on. You will see in the Debug log, the equivalent svn commands that FishEye is executing.