Security and R

One of the more challenging aspects of OpenCPU is security in R (or lack thereof). This is actually one of the reasons OpenCPU runs on Linux only at this point; other operating systems simply lack superpowers to implement open computing. (Maybe one exception is BSD, for which I lack superpowers).

Security is actually a general concern in R, or at least it should be, and we thought that this part of the framework deserves its own package. The result is the RAppArmor package, now available on CRAN and Github. Because many R users might not be very familiar with security in Linux, below some introduction video tutorials to get you started! The first couple of tutorials are now on youtube, if they are a success, some more might be added.

Of course you should watch the videos in high quality / full screen so that you can actually read the code.

The videos

Part one [link] explains what AppArmor profiles R, and how they can be applied to a process, in this case R.

Part two [link] shows how the RAppArmor package can be used to apply security profiles, resource restrictions and other security measures to a running R process, from within R.

In part 3 [link] , the very powerful eval.secure function is introduced, that ties everything we learned so far together in one simple function.

Hope you like it! Feedback, questions, tips, bugs, complaints and tokens of appreciation are welcome of course! Make sure to have a look at the github page.