[tahoe-dev] Experimenting with SHA-3 candidates in Tahoe-LAFS

[Pål Ruud]

Hi Zooko (and other Tahoe-devs),

with reference to the email with similar subject dated 11. Mar 2010 from Danilo,
we have started on our project to try out the 14 remaining candidates of the
SHA-3 competition with Tahoe-LAFS. In this matter, we have several questions we
hope to get an answer to.

All of the candidates comes with (optimized) C implementations, so we have
started making Python bindings for these.

Our knowledge of Tahoe-LAFS is at the moment rather limited, as we have just
read what we can find in the docs/ [0] and skimmed through some of the source
code.

Do you think it would be sufficient to do the following?

# util/hashutil.py
# from pycryptopp.hash.sha256 import SHA256
from sha3 import BlueMidnightWish as SHA256

This seems to work for immutable files, but not for mutable files (e.g.
directories), because one of the tests fails (with an
«UncoordinatedWriteError \x0a»). We are looking in to this.

Our main goal with this project, however, is to get a reproducible, automated
benchmark going. Do you have any tips for us, i.e. on how to attack this
problem?

We are also idling at #tahoe-lafs, as Zerqent and ruudud.

[0] http://tahoe-lafs.org/source/tahoe/trunk/docs/

Best regards,
-- 
Pål Ruud and Eirik Haver,
Department of Telematics (ITEM), 
Norwegian University of Science and Technology