A node's repos_id is generated based on root URL and UUID.
If a file external is configured as coming from URL1, and a working copy is
checked out from the same repository via a different URL2, the file external
gets a different repos_id than other nodes in the working copy.
This leads to problems because all files are currently assumed to come from the
same repository.
We might need to revisit this and generated unique repos_id's based
on the UUID only. Not sure if that can be made to work without a working
copy format bump though.
See http://svn.haxx.se/users/archive-2011-12/0305.shtml for a users@ report of
this problem.
Major