I am trying to use furrr::future_pmap in R to replace purrr::pmap in a function call within another function.
Presently I have it set up so pmap is passing other arguments using the ellipsis ... however when I try and do this using future_pmap I get unused argument errors (see example below). I know from comments in here passing ellipsis arguments to map function purrr package, R and other previous research that for the ellipsis to work with pmap you need to use function(x,y,z) blah(x,y,z,...) instead of ~blah(..1,..2,..3) but the same approach doesn't seem to work for future_map. Is there some other secret to making this work?
I've created a very simple reprex, obviously my real functions make a lot more sense to run in future_pmap
library(purrr)
library(furrr)
#> Loading required package: future
plan(multiprocess)
xd <- list(1, 10, 100)
yd <- list(1, 2, 3)
zd <- list(5, 50, 500)
sumfun <- function(indata, otherdata){
out <- sum(c(indata, otherdata))
return(out)
}
test_fun_pmap_tilde <- function(ind, ...){
return( pmap(ind, ~sumfun(c(..1,..2,..3), ...)))
}
test_fun_pmap <- function(ind, ...){
return( pmap(ind, function(x,y,z) sumfun(c(x,y,z), ...)))
}
test_fun_future_pmap <- function(ind, ...){
return( future_pmap(ind, function(x,y,z) sumfun(c(x,y,z), ...)))
}
#doesn't work as need to use function(x,y,z) instead of tildes
test_fun_pmap_tilde(list(xd, yd, zd), otherdata = c(100,1000))
#> Error in sumfun(c(..1, ..2, ..3), ...): unused arguments (.l[[2]][[i]], .l[[3]][[i]])
#this one works
test_fun_pmap(list(xd, yd, zd), otherdata = c(100,1000))
#> [[1]]
#> [1] 1107
#>
#> [[2]]
#> [1] 1162
#>
#> [[3]]
#> [1] 1703
#but using future_pmap it doesn't work
test_fun_future_pmap(list(xd, yd, zd), otherdata = c(100,1000))
#> Error in (function (x, y, z) : unused argument (otherdata = c(100, 1000))
Created on 2020-08-31 by the reprex package (v0.3.0)
