Knoware status, July 27th

[Brian Beck]

Progress report for the past two weeks (sorry about lack of update last week):

* Studied R [1], the statistical package for which RandomForests is
written.  RandomForests [2] was recommended to me by a couple
statician friends as the most appropriate method for Knoware's goals.

* Practiced with RPy [3], the Python package used to interface with R.

* Database design is almost complete.  I'm going with SQLite [4] as a
backend for now because I've used it for a few projects and have never
had any problems.

* Dirk Mueller asked why I chose lshw [5] instead of Debian's hwinfo
[6].  Well, I simply found lshw first (it was recommended to me) and
it has multiple output formats including XML.  lshw must be run as
root for *some* information, but right now it looks like the non-root
output is also acceptable.  If anyone knows differently about the
quality & quantity output from various packages, I'm all ears.

This week:

* Wrap my head around RandomForests with some simple test cases, hopefully.
* Still looking for a widespread (i.e. cross-distribution) library
detection package.  For the software part I can always fall back on
KAppFinder, but library versions are more important.
* Plug hardware detection into the interface and maybe even the
client-server sending part.
* Plug problem reporting/searching/browsing into the interface.

[1] R: http://www.r-project.org
[2] RandomForests: http://www.stat.berkeley.edu/users/breiman/RandomForests/
[3] RPy: http://rpy.sourceforge.net
[4] SQLite: http://www.sqlite.org
[5] lshw: http://ezix.sourceforge.net/software/lshw.html
[6] hwinfo: http://packages.debian.org/unstable/admin/hwinfo

-- 
Brian Beck
Adventurer of the First Order