d'Arc is a Neutrino

["H. Diedrich" <hd2010@...>]

Preparing the jigsaw curves I promised at the conference, I am finding 
that d'Arc can be 30x faster than LuaJIT 2.0 API for very big tables.

Am I missing a preferred way of how to use the API for loops with LuaJIT?

My understanding is that using the API with LuaJIT in a loop is usually 
slower than LuaJIT ipairs and pairs, because pairs can be optimized and 
API calls not.

Is there a way that while using the API I don't get that badly left behind?

Thanks,
Henning