Christophe Mckeon wrote:
> http://drp.rubyforge.org
>
> i'll be doing a presentation tonight in secondlife, read the recent post
> titled Rubyists of Second Life Meeting, 09/07/2006.

When I first saw this, I was thinking that it would be fun to play with
whenever I get time to look at evolutionary programming again, but it
just hit me that there's another place where this project would be very
useful:

Testing. I was just writing some code to generate sequences of test
operations, and I remembered this project - using it to generate test
scripts from a grammar seems to be  a good fit. The question would be
what type of search algorithm to use, and if using a GA etc. what would
make sense for a fitness function.

In the latter case, I guess it depends on the purpose of the testing:
In the case of performance, using a fitness function that assigns
higher fitness the lower the system performance would encourage test
scripts that converge on patologically bad performing cases. In the
case of logic failures, it may be tricky to assign a gradual scale of
failures, and I'm not really sure if a GA would be a good search
alogrithm for that case. 

Vidar