Knowledge base

This article is a knowledge base for the{rSharp} package. It is a collection of useful information that may help troubleshooting and understanding the package.

Loading the workspace

Saving the RStudio workspace (which is the default option on RStudio exit) will save the R6 objects, but the pointers to .NET objects will be lost. Trying to access the objects after loading a workspace will result in an error. There is currently no workaround for saving .NET objects.

Data type conversion

  • NA is a concept of marking “missing values” in R which is not available in .NET. What might appear conterintuitive is that NA has a type “logical” and is interpreted as TRUE when sent to .NET. Attention should be payed when sending NA to .NET methods.
  • If the signature of a .NET method require an integer argument, it must be passed as an integer using as.integer() from R. R assumes all numerics as doubles by default, and will cause an error if passed to a method that expects an integer.
  • Objects in R are vectors, and single values are just a particular case of vectors of length one. In the .NET, scalar values and arrays are different types. In current implementation, an R vector with length >1 will be translated to an array in the .NET, a length of one becomes a scalar value. This might lead to problems when calling .NET methods expecting a an array when passing a vector of 1.
  • When trying to pass a vector containing a NULL from R to .NET, the NULL will be removed from the vector. This is how R treats NULL in vectors. E.g., trying to call a .NET method with an argument that expects an array of doubles with a vector c(0, NULL, 3) will result in a call with c(0, 3). In order to correctly pass NULL to .NET, it should be passed as a list: list(0, NULL, 3).