This repository is for testing cvs2svn's choices of which directories
to copy when filling symbolic names.  The layout is:

     /one.txt
      two.txt
      three.txt
      /sub1/four.txt
            five.txt
            six.txt
            sub2/seven.txt
                 eight.txt
                 nine.txt
                 ten.txt

Every file was imported in a standard way, then revisions 1.2 and 1.3
were committed on every file.  Then a branch was made:

    BRANCH_1: sprouts from every file's 1.3 (so branch number 1.3.0.2)

Then a revision was committed on that branch, creating revision
1.3.2.1 on every file.  Next a branch was made from that revision, on
every file:

    BRANCH_2: sprouts from every file's 1.3.2.1 (so branch number 1.3.2.1.0.2)

BRANCH_3 to BRANCH_8 all sprout from either trunk, or from
the first revision on BRANCH_1 (that is, from 1.3.2.1), in various
combinations.  Every branch below exists on every file, the only
question is where the branch is rooted for each file.

    BRANCH_3: Sprouts from trunk everywhere except sub1/sub2/*, where
              it sprouts from BRANCH_1 for all four files.

    BRANCH_4: Sprouts from trunk everywhere except for
              sub1/sub2/ten.txt, where it sprouts from BRANCH_1.  Note
              that this is a clear minority in sub1/sub2/, since it
              still sprouts from trunk on the other three files there
              ('seven.txt', 'eight.txt', and 'nine.txt').

    BRANCH_5: Sprouts from trunk everywhere except for
              sub1/sub2/nine.txt and sub1/sub2/ten.txt, where it
              sprouts from BRANCH_1.  This is an even division in
              sub1/sub2/, since it sprouts from trunk on two files
              ('seven.txt' and 'eight.txt') and from BRANCH_1 on the
              other two ('nine.txt' and 'ten.txt').

    BRANCH_6: Sprouts from trunk everywhere except for
              sub1/sub2/eight.txt, sub1/sub2/nine.txt, and
              sub1/sub2/ten.txt, where it sprouts from BRANCH_1.  This
              is a clear majority in favor of BRANCH_1, since BRANCH_6
              sprouts from trunk on only one file ('seven.txt') and
              from BRANCH_1 on the other three ('eight.txt',
              'nine.txt' and 'ten.txt').

    BRANCH_7: Sprouts from trunk everywhere except sub1/five.txt and
              sub1/six.txt, where it sprouts from BRANCH_1.  This is a
              majority in favor of BRANCH_1 there, as the only other
              file in that directory is 'four.txt', but note that both
              the parent directory and the sole subdirectory are
              majority from trunk.

    BRANCH_8: The reverse of BRANCH_7.  Sprouts from BRANCH_1
              everywhere except sub1/five.txt and sub1/six.txt, where
              it sprouts from trunk.  This is a majority in favor of
              trunk there, as the only other file in that directory is
              'four.txt', but note that both the parent directory and
              the sole subdirectory are majority from BRANCH_1.

To test the filling of a tag set on a branch, a new branch is created.

   BRANCH_B1: sprouts from every file's 1.3 (so branch number 1.3.0.12)

A single change to one.txt in BRANCH_B1 is committed so that 1.3.12.1 is
created on that file, and TAG_B1 is set on the tip of that branch.

      TAG_B1: set on 1.3.12.1 on one.txt, and 1.3 on the rest

TAG_B1 should be created as a single copy from BRANCH_B1.