convertInteger64ToNumeric question

[MPagel]

I received the following error in R

The data contains integers >= 2^53, and converting those to R's numeric type leads to precision loss. Consider using smaller integers, > converting the integers to doubles on the database side, or using options(databaseConnectorInteger64AsNumeric = FALSE)

I had been assuming the alternative to having the Int64 be numeric would be to have it converted to string. But, am I correct that setting the value to FALSE simply keeps it as bit64 throughout processing? As you've already introduced the library dependency on bit64 (given the maxInt64 <- bit64::as.integer64(2)^53 call), so the end user must also have this library installed. why is this conversion to R's type numeric the default behavior?

In any case, especially if it is to remain the default behavior, it would be helpful to better document what this variable means. I had thought that "numeric" was being used generically to describe a integer (or in some other cases floating point), and the alternative to that would have been to be a string type. So, I was essentially thinking the program was making a decision point: take a database-side BigInt/Int64 type value and bring it in as a number or a string.

[schuemie]

We at some point tried using int64 throughout our software, but found some serious issues as documented here. I agree this needs to be documented better. It is on my to-do list.

Numeric can contain much larger int values than Integer in R, hence it is the default behavior. I don't think many people would expect strings.