# remotes::install_github("WinVector/wrapr") library(wrapr) a <- 5 b <- 7 do_not_want_1 <- 13 do_not_want_2 <- 15 # save the elements of our workspace we want saveRDS(as_named_list(a, b, do_not_want_1), 'example_data.RDS') # clear values out of our workspace for the example rm(list = ls()) ls() #> character(0) # notice workspace environemnt now empty # read back while documenting what we expect to # read in unpack[a, b] <- readRDS('example_data.RDS') # confirm what we have, the extra unpack is a side # effect of the <- notation. To avoid this instead # use one of: # unpack(readRDS('example_data.RDS'), a, b) # readRDS('example_data.RDS') %.>% unpack(., a, b) # notice dot # readRDS('example_data.RDS') %.>% unpack[a, b] # readRDS('example_data.RDS') %.>% to(a, b) # no dot # readRDS('example_data.RDS') %.>% to[a, b] ls() #>  "a" "b" "unpack" # notice do_not_want_* are not present # do_not_want_2 was dropped when we wrote the RDS # do_not_want_1 was not specified during the unpack, so dropped print(a) #>  5 print(b) #>  7
We have new documentation on the new as_named_list helper function here.
The idea is: this is a case where non-standard evaluation is working for us (and clarity/safety) as it forces the user to document each object they want written out (by explicitly naming them), and exactly what objects they expect to come back (again by explicitly naming them right at the assignment location).
Data Scientist and trainer at Win Vector LLC. One of the authors of Practical Data Science with R.