New Post: R.NET failing on library(RODBC) call (and other external libraries)

I could reproduce your twitteR issue, and fix it by following the advice already in this message thread.

The short version:

• By default do not set a R_HOME in the machine environment variable.
• Add c:\Program Files\r\R-3.2.0\bin\i386 to the machine or user's PATH env variable (or \x64 if your IIS server runs 64 bits, see details)
• I advise you look at the solution in rdotnet-onboarding, under folder Solutions/WebApp. Credits: this material was contributed by @skyguy94
• Install the NuGet packages (there are quite a few - couple of minutes)
  

The long version:

Environment:

• Win7 64 bits
• R 3.2.0
• VS2013
• IIS Express, don't know the version
  

Putting breakpoints into CodeController.Execute and its constructor, in the app of the onboarding guide

library(twitteR)

and using the "immediate" debug window on breakpoints

Environment.Is64BitProcess
false

Curious, as I am running a 64 bits operating system and the build is AnyCPU. Whatever, 32 bits it is.

Now if I query for the environment variables:

Environment.GetEnvironmentVariable("PATH")
"c:\\Yadi\\Yada\\Lots\\of\paths;F:\\bin\\doxygen\\bin"
Environment.GetEnvironmentVariable("R_HOME")
"C:/PROGRA~1/R/R-32~1.0"

The call to Evaluate fails indeed.

PATH should have included "c:\Program Files\r\R-3.2.0\bin\i386", but it is missing. R.NET would have appended it. Don't know/understand why IIS does that, whatever. R_HOME has been set correctly by R.NET (and the path format for R_HOME is deliberate)

To fix it it seems I then need to:

• kill IIS Express
• Add c:\Program Files\r\R-3.2.0\bin\i386 to the Machine or user's PATH env ver
• AND restart VS2013 (Huh?)
  to get a fully clean slate.
  

Restart in debug mode, same breakpoint, immediate window:

Environment.GetEnvironmentVariable("PATH")
"c:\\Yadi\\Yada\\Lots\\of\paths;F:\\bin\\doxygen\\bin;c:\\Program Files\\r\\R-3.2.0\\bin\\i386"
Environment.GetEnvironmentVariable("R_HOME")
"C:/PROGRA~1/R/R-32~1.0"

Then things work:

library(twitteR)
tuser <- getUser('geoffjentry')

I have not set up OAuth so I get an error of course, but the package seems to work fine.