bzr svn-import with a deep branch layout

[Russ Brown]

Hi,

I'm trying to import our old svn repository into bzr for easier historical 
digging. Partway through its life merging became managed by svk, so I'm really 
hoping for merges to be properly represented in the resultant bzr repository.

However, I'm having a little trouble. The repository is over 43k revisions and 
13G in size. Trying to do the import in one go has proven to be impossible due 
to memory being exhausted, so I'm now doing it using the --until option and 
incrementing by 1000 revisions at a time.

However, it is getting slower and slower, and I'm starting to wonder if svn-
import is having trouble properly detecting our branches properly.

We structured our repository like this:

trunk
branches/
branches/categoryA
branches/categoryA/branch1
branches/categoryA/branch2
branches/categoryB
branches/categoryB/branch1
branches/categoryB/branch2

etc.

In other words, the directories beneath the branches directory are not 
themselves branches, but other plain directories which contain the actual 
branches.

Looking in my bzr import directory currently, I see the category directories, 
but no branch directories within them. Furthermore, the category directories 
contain .bzr directories, which makes me think that svn-import is treating 
them as directories. If that is the case, svn-import could see actual branch 
creation operations as being massive copies into what it sees as being 
branches, which could explain the slowdown.

Are my suspicions correct, and if so is there any way to get svn-import to 
detect branches by noticing copies of trunk, rather than simply looking for 
subdirectories of branches?

Thanks.

-- 

Russ