Use %g for all floating point output.
Accept INF, 1.#INF0, NAN, 1.#QNAN etc. for floating point input
Authored by: johnhartman 2010-11-30
Parent: [r867]
Child: [r869]