Error when column name equals variable name about hospitalnetwork

[tjibbed]

I found out it's not exactly the same error that happens for the other variables.

for hospitalID, I get:

Error in get(hospitalID) : object '18091301' not found
where 18091301 is a hospital code

for admDate:

Error: invalid first argument

with traceBack:

10. stop(err$message, call. = FALSE)
11. .checkTypos(e, names(x))
12. value[3L]
13. tryCatchOne(expr, names, parentenv, handlers[[1L]])
14. tryCatchList(expr, classes, parentenv, handlers)
15. tryCatch(eval(.massagei(isub), x, parent.frame()), error = function(e) .checkTypos(e,
    names(x)))
16. [.data.table(base, get(admDate) > get(disDate), , which = T) at DataManagement.R#302
17. base[get(admDate) > get(disDate), , which = T] at DataManagement.R#302
18. checkDates(base = checkResults$base, patientID = patientID, hospitalID = hospitalID,
    admDate = admDate, disDate = disDate, convertDates = convertDates,
    dateFormat = dateFormat, deleteErrors = deleteErrors, verbose = verbose,
    ...) at DataManagement.R#62
19. checkBase(countryData, deleteMissing = "patient", deleteErrors = "patient",
    convertDates = TRUE, dateFormat = "%y%m%d", patientID = "pID",
    hospitalID = "hID", disDate = "dis", admDate = "admDate",
    verbose = TRUE, maxIteration = 250)

and for disDate the same with traceBack:

10. stop(err$message, call. = FALSE)
11. checkTypos(e, names(x))
12. value[3L]
13. tryCatchOne(expr, names, parentenv, handlers[[1L]])
14. tryCatchList(expr, classes, parentenv, handlers)
15. tryCatch(eval(.massagei(isub), x, parent.frame()), error = function(e) .checkTypos(e,
    names(x)))
16. [.data.table(base, get(admDate) > get(disDate), , which = T) at DataManagement.R#302
17. base[get(admDate) > get(disDate), , which = T] at DataManagement.R#302
18. checkDates(base = checkResults$base, patientID = patientID, hospitalID = hospitalID,
    admDate = admDate, disDate = disDate, convertDates = convertDates,
    dateFormat = dateFormat, deleteErrors = deleteErrors, verbose = verbose,
    ...) at DataManagement.R#62
19. checkBase(countryData, deleteMissing = "patient", deleteErrors = "patient",
    convertDates = TRUE, dateFormat = "%y%m%d", patientID = "pID",
    hospitalID = "hID", disDate = "disDate", admDate = "adm",
    verbose = TRUE, maxIteration = 250)

[PascalCrepey]

Hi Tjibbe,
now :

mydb = create_fake_patientDB(n_patients = 100, n_hospital = 5)
mat1 = hospinet_from_patient_database(base = mydb)

and

setnames(mydb, 1:2, c("patientID", "hospitalID"))
mat2 = hospinet_from_patient_database(base = mydb, patientID = "patientID", hospitalID = "hospitalID")

give the same thing.

The problem was that the resolution of the variables in R is done in multiple context, first the current, and if nothing is found, the parent, etc... Hence the issue can be simplified as:

dt = data.table(a = 1:5)
a = "a"
var = "a"

dt[, get(var)] #works as intended
dt[, get(a)] #fails 

One option to solve this would be to force resolution in the parent environment by deactivating the resolution within the data.table:

dt[, get(var), with = FALSE]
dt[, get(a), with = FALSE]

or use the shorter version ".." which disable nothing but indicate that the variable needs to be looked in the parent environment (hence not in the data.table)

dt[, ..var]
dt[, ..a]

I've implemented the second option in the last commit.