Should RItest return an upper p value in a hard coded way?

[jwbowers]

Perhaps it is late in the day, so that it why I'm not sure whether or not having RItest return an upper p value achieves the goals of that function or not. On one hand, an bunch of upper p-values, I can calculated two-sided p-values by: p2sided <- 2*min(c(upper, 1-upper)). On the other hand, for some reason, I've tended to do p2sided <- 2 * min(c(upper, lower)) where I don't require upper = 1-lower. Right now, we do not require all test statistics to pass positive values, for what it is worth. Quick thoughts?