Incorrect bzrlib usage or a memory leak?

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 10/5/2011 2:13 AM, Martin Pool wrote:
> Just in case other people on this list are curious, John did an 
> awesome analysis in 
> <https://answers.launchpad.net/bzr/+question/173096> 
> <http://imagebin.org/177384> showing that the memory does grow, but
> it plateaus, and the growth can reasonably be explained by the
> increasing size of the tree as you go forward through history, and
> caches that grow to a bounded size.
> 
> Martin
> 

Unfortunately, imagebin will expire my paste soon, but for those
following along, I have an updated version after running for ~16 hours:

http://people.canonical.com/~jameinel/peak_working.png

The surprising bit for me is that the time to extract each revision
hasn't gone up significantly, even though the number of bytes has:
http://people.canonical.com/~jameinel/extract_time.png

Total bytes extracted is 734GiB over 16 hours, or about 12.4MiB/s. My
original estimate was off, and we actually started at about 6MiB/s, so
we've been speeding up significantly.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk6MB9kACgkQJdeBCYSNAAPqGACcC6yrs1gcOXI3sFch0SLY20WK
Cp8AnRUSKheSkNrtngFyHHblSWum4ghv
=WpWj
-----END PGP SIGNATURE-----