The OpenCPU System is an important step towards a universal interface for scientific computing. Even though bridges to embed R in general purpose software have been available for several years, they have not been able to facilitate the big breakthrough of R as a ubiquitous statistical engine. The primary cause for the limited success is that low-level tools are difficult to implement, do not scale very well, and leave the most challenging problems unsolved. Substantial plumbing and expertise of R internals is required for building actual applications on these tools. What is needed to scale-up embedded scientific computing is a system that separates the application layer from the computational back-end, similar to how e.g. SQL separates application from database. The OpenCPU API defines an interface that captures the domain logic of scientific computing and abstracts implementation details in a way that allows for independent development of client and server components by different people that do not speak each others language.
The following video is the recording of a talk about OpenCPU by Jeroen Ooms at useR! 2014 in Los Angeles. The slides can be accessed HERE.
Sign up for the free insideBIGDATA newsletter.
Speak Your Mind