protected wrong buffer-length write by initializing to 0, but there is more debugging...

protected wrong buffer-length write by initializing to 0, but there is more debugging needed: there is still something fishy with the last out-commented test of test/api/t16.c.