I’m working with a dataset that uses long strings of numerics as identifiers. I always make this id column into class character because the numerics do not actually represent numbers. I saved and then loaded a dataset using write_csv and read_csv and noted that those functions seem to be actually changing the values in the dataframe, which is really troubling. Here’s a reprex:

library(readr)

# example from my actual dataset 
dat <- data.frame(id = c("196307010100010157", "196307010100020158", "196307010100030163", 
                         "196307010100040161", "196307010100050162", "196307010200010159", 
                         "196307010200020160", "196307010200030164", "196307010200040165", 
                         "196307010200050173"))

class(dat$id) # definitely characters

write_csv(dat, file="dat.csv")

dat <- read_csv("dat.csv")

class(dat$id) # now it's numeric, which is issue #1 -- why change the class?

sprintf("%.0f",dat$id) # issue #2, and much more serious, is -- the values are different! 

I’m pretty alarmed by this because I use these functions all the time. Does anyone know why this behavior arises in this case?