I'm trying to consolidate three smallish SVN repositories into a single large on, while preserving revision history for all files.
I've tried doing:
svnadmin dump repo_A > repo_A.txt
making a new subdirectory called REPO_A in the 'containing' repository, then saying:
svnadmin load --parent-dir "REPO_A" /path/to/containing/repo < repo_A.txt
This sort of works, but when I do a clean checkout I only get the contents of the original, containing repository.
I can explicitly check out the added repository by saying:
svn co /path/to/containing/repo/trunk/REPO_A mysandbox/REPO_A
but this is less than desirable.
I just want the result to be one big repository, and get the whole thing on checkout.
What am I doing wrong here?
Note that the enclosing repository and the repository I'm trying to include are disjoint; there isn't any overlap in terms of file names. There are several references on doing this scattered around the web; all suggest the same dump/mkdir/load procedure I've used, and nothing is mentioned about version conflicts. In fact, when I load the dump, the various versions it contains seem to be incremented from the youngest revision number found in the enclosing repository. For example, if the latest revision in the enclosing repository is 2000 and the number of revisions in REPO_A is 500, when finished the latest revision in the enclosed REPO_A is 2500.
That isn't the problem, however. The problem is that the newly loaded repository does not seem to be part of the enclosing repository. Although it can be checked out explicitly as noted, it isn't checked out as part of a top-level co, and issuing 'svn status' at the top level shows the subdirectory containing it as unknown to the repository, even though it came from there. Status messages within the REPO_A subdirectory work as expected.
