Name collision because of NSE with data.table

[SuperMayo]

Because the package makes extensive use of data.table, and because it uses extensively non-standard-evaluation(NSE). It may arrive that a column name is identical to some internal variable name, resulting in an error.

Example

library(didImputation)

data("did_simulated")
did_simulated$s <- rnorm(nrow(did_simulated))

didImputation(y ~ 0 | i + t,
              cohort = "g",
              data = did_simulated)

>> Error in s$time : $ operator is invalid for atomic vectors 

Explanation

The error comes from the following line

didImputation/R/prepare.R

Line 34 in 721876a

s$data[, k := get(s$time) - get(s$cohort)]

Indeed there is name collision between the internal object s of the package and the column name s.

Solution(s)

1. Get rid of data.table NSE.
2. Use quasiquotation.
3. Rename sensible variables with weird names to have low collision probability 🤷‍♂️
4. Use eval instead of get

In the meanwhile

Don't name a column s...