This is a draft resolution to#701 and some related edge cases. The issue seems to be more general than originally described with handling any of the non-finite values (including NAs).

The currently implemented behavior is the following:

• Numeric non-finite values (NA_real_,NA_integer_,NaN,Inf and-Inf) will be treated as aqty > 1 for pluralization purposes.

• Non-numeric NA values (NA_character_ andNA) are treated as a non-numeric input soqty is based on the vector's length.

I've opted not to throw an error for numericNAs orNaN cases since that seemed a bit severe but it also seems like a plausible outcome and happy to switch to that behavior if that is a better option.