testing followup

[James Howison]

We had a conversation a while back about the difficulty of unit
testing in the graphical domain, since often errors are only apparent
when people look at the graphs.

I read about this today, it's a mech turk type proposal for pushing
human in the loop tests out to a user community:

http://www.infoq.com/news/2009/03/test-swarm

Might be a possibility or an inspiration for regression testing on
ggplot2 etc.

--J

[hadley wickham]

Sounds like a neat idea. Thanks for the pointer!

Hadley

--
http://had.co.nz/