* cvsroot
    * crossfire module:
      * size: 193356010 bytes (184 MB)

  * cvs20hg:
    * crossfire module:
      * conversion time:<code>
real    268m31.821s
user    242m3.478s
sys     1m45.126s</code>

  * Mercurial:
    * crossfire module:
      * size: 111895365 bytes (106 MB) (I suspect binary diffs made the history for the collected images much smaller)
      * local copy time ('hg init' + 'hg pull' + 'hg update'):<code>
real    1m40.124s
user    1m01.218s
sys     0m9.595s</code>
      * local copy time ('hg clone'):<code>
real    0m5.619s
user    0m2.991s
sys     0m0.974s</code>
      * copy creation bandwidth:
        * 'hg clone' over http: 77.0MB
        * 'hg pull' over http: 77.0MB
        * 'hg push' to copy over ssh (uncompressed): 181 MB
        * 'hg push' to copy over ssh (compressed): 78.4 MB
      * Local commit of one file with small edits:<code>
real    0m0.440s
user    0m0.332s
sys     0m0.070s</code>

  * Tailor (cvs to bzr):
    * crossfire module:
      * conversion time:<code>
real    378m40.796s
user    211m26.975s
sys     8m31.729s</code>

  * Bzr:
    * crossfire module:
      * size: 123277815 bytes (120 MB)
      * local copy time ('bzr branch'):<code>
real    1m3.882s
user    0m43.203s
sys     0m4.979s</code>
      * copy creation bandwidth:
        * 'bzr branch' over http: 83.9 MB
        * 'bzr get' over http: 97.5 MB
        * **Note**: I've heard that doing this from remote hosts can be very slow due to the fact that currently it establishes separate connections for each file it wants. Apparently it takes about 2 hours to transfer the data on the internet in a fairly typical case.
        * (Installing pycurl prevents this issue, as connections to the server are reused.  A version of Keep-alive with urllib is in the works for 0.11.)
      * Local commit of one file with small edits:<code>
real    0m2.410s
user    0m2.051s
sys     0m0.183s</code>