OK, the notebook showing the translation from the old to new API is chugging along is chugging along.

I'd still like to

1. document how do you specify tests using a given hull, either the types supported by default or a pre-sepcified hull polygon. This can be done by the end of the notebook, in the virginia example
2. check p-values for validity. I need to double check the logic
3. code a hand-computable example for the validity tests.

I've completed tests for all the helper functions, but still need to write the correctness tests.

I also need to document the hulls.

Tests are now in place and they're pretty comprehensive (if the backlog of commits is anything to show for it :)

I just need to document the hull logic in a notebook & then we're done.

Locally, this works for all the tests, and fails for some of the Knox statistics I didn't touch... not sure what's up there? We'll have to see in travis, but I think this is rtg. Documentation can be added later to beef up the info on the simulator alternatives.

While we should deprecate those (I think) I'm much less interested in doing so because they're performant.

OK, well, I'm not sure how to get the specification correct for travis, but tests pass locally aside from the Knox stuff (which I haven't touched). This is ready for review.

OK, well, I'm not sure how to get the specification correct for travis, but tests pass locally aside from the Knox stuff (which I haven't touched). This is ready for review.

See if this solution addresses the travis issues.

@ljwolf do you think this is the way we should go on the testing? Yesterday at the dev meeting we briefly mentioned this but I wanted to circle back in case it got lost in the flurry of activity.

If it is the way forward, please merge that pr into your branch, which will update this pr and i think have the tests pass.

Yes! Sorry, took the recommendation and merged it! I forgot about it when tinkering around today.