Robustly check if lower bound function is equivalent to fixed gs_b()

[jdblischak]

Closes #411
Follow-up to #408

Here's my attempt at robustly comparing the input lower bound function to gs_b(). Unfortunately there is no test case where fixed_futility_bound evaluates to TRUE. Here is a quick demonstration of how it works:

devtools::load_all(".")

# Function that is equivalent to current gsDesign2::gs_b()
gs_b_yihui <- function(par = NULL, k = NULL, ...) {
  if (is.null(k)) par else par[k]
}

x <- list()
x$input$lower <- gs_b_yihui

gs_b_observed <- try(x$input$lower(par = 4:2, k = 2), silent = TRUE)
gs_b_expected <- gs_b(par = 4:2, k = 2)
fixed_futility_bound <- identical(gs_b_observed, gs_b_expected)
fixed_futility_bound
## [1] TRUE

# Function that is not equivalent to gsDesign2::gs_b()
x$input$lower <- gs_spending_bound

gs_b_observed <- try(x$input$lower(par = 4:2, k = 2), silent = TRUE)
gs_b_observed
## [1] "Error in par$timing : $ operator is invalid for atomic vectors\n"
## attr(,"class")
## [1] "try-error"
## attr(,"condition")
## <simpleError in par$timing: $ operator is invalid for atomic vectors>
gs_b_expected <- gs_b(par = 4:2, k = 2)
fixed_futility_bound <- identical(gs_b_observed, gs_b_expected)
fixed_futility_bound
## [1] FALSE

cc @yihui

[yihui]

That sounds like a better approach to me.

[LittleBeannie]

Thanks, @jdblischak, I like this solution!

I also tried lower = gs_spending_bound, then I have gs_b_observed is an error, leading to a FALSE value of fixed_futility_bound.