Nullable doubles

[vanhumbeecka]

There is an option to pass nullable booleans over the connection:

Sexp.Make(bool? x)

But there aren't options to pass on nullable strings, ints, doubles, ...

Why is that?
And more importandly, what would be the best solution to pass on nullable doubles?

Regards

[SurajGupta]

Hello. I never been happy with the NULL/NA implementation in this project (strings are reference types, so they are already nullable). It's important to distinguish NA from NULL in R. NULL is simply ignored. Try this: test = c(1,NULL,2); test NA, on the other hand, is treated as a value. Try this: test =c(1,NA,2); test

On the .net side, for nullable booleans, NULL is converted to NA in R. For double, double.NaN is converted to NA in R. For ints, SexpArrayInt.Na is converted to NA in R (-2147483648). So for your immediate need, just use double.NaN whenever you want a NA in R. Long-term, I should find the time for a proper fix, which off the top of my head means having additional construcutors in SexpArrayInt and SexpArrayDouble that accept nullable ints/doubles. Of course, we have to also solve for the R -> .Net scenario. So maybe when reading from R, everything gets converted to nullable double/ints by default so it's super clear whether we have read an NA from R. Not sure if I'll have time in the next few months to fix this up - you free to try (please write unit tests!) or just use the work-around. I'll keep this issue open so we can track the long-term fix.

[vanhumbeecka]

Thanks for the reply.
You're right about pointing out the difference between NA and NULL in R. My question was not clear enough. I was indeed looking to import NA's into R from NULLs in .NET. Right now I only need a way to import NAs in R and get a set of non-NA's back from R. Double.NaN is exactly what i needed.