This seems weird and stems from the fact that c is never initialized, only declared. I don't think that we should support this. Decay should actually throw an error for empty compositions, I think.
We can change it to use non-empty compositions. It doesn't really matter because the functionality being tested in that test doesn't have anything to do with a decay calculation. It is just testing that a compositions decay dt history is tracked properly so as to avoid repeating decays that have already been calculated.
Or we could even just throw an exception if anybody tries to create an empty composition - since it is sort of undefined.
Well people might want to create an empty composition and then build it up. Should they just be using CompMap in this case, though?
Leaving this be for 1.0
moving to 1.2
Moved to v1.3
@rwcarlsen / @scopatz, can you give an update on this issue?
the short is that nothing will happen for 1.3. Also, the level of bugness of this is arguable - there are many other higher priority things IMO.