Hi,
This line introduces a problem with NA values:
https://github.com/stineb/rpmodel/blob/master/R/rpmodel.R#L437
In the presence of NAs (but no other problems) the test fails:
> any(c(FALSE, NA)) # returns NA
[1] NA
> if(any(c(FALSE, NA))) { stop('a_j!=a_c')} # if() statement chokes,
Error in if (any(c(FALSE, NA))) { : missing value where TRUE/FALSE neededWith grid maps as inputs, NAs are pretty likely and are handled by all steps up to this one. One alternative here is all.equal, which handles NAs and allows a tolerance to be set:
> all.equal(c(1,2,3,4, NA), c(1,2,3,4.001, NA), tol=0.001)
[1] TRUE
> all.equal(c(1,2,3,4, NA), c(NA, 2,3,4.001, NA), tol=0.001)
[1] "'is.NA' value mismatch: 2 in current 1 in target"
It would probably be useful to handle $a_j != a_c$ and mismatched NA values separately though? So trap the output of that all.equal and report separately or perhaps use a separate xor(is.na(a_j), is.na(a_c))
Hi,
This line introduces a problem with NA values:
https://github.com/stineb/rpmodel/blob/master/R/rpmodel.R#L437
In the presence of NAs (but no other problems) the test fails:
With grid maps as inputs, NAs are pretty likely and are handled by all steps up to this one. One alternative here is
all.equal, which handles NAs and allows a tolerance to be set:It would probably be useful to handle$a_j != a_c$ and mismatched NA values separately though? So trap the output of that
all.equaland report separately or perhaps use a separatexor(is.na(a_j), is.na(a_c))