Block-based programming abstractions for explicit parallel computing

With the majority of computing devices now featuring multiple computing cores, modern programmers need to be able to write programs that utilize these cores in parallel to extract the requisite levels of performance. Despite the need for such explicit parallel computing, few programmers are properly groomed in the mindset and the practices of parallel computing. Block-based programming languages, such as Scratch and Snap!, have proven to be a highly effective means of teaching fundamental programming concepts to a wide student audience. Nevertheless, the rich feature-set of mainstream block-based programming environments lack abstractions for explicit parallel programming, thus missing the opportunity to introduce this increasingly important programming concept at a time when the students' minds are most receptive. This paper reports on the results of an NSF-sponsored project for adding and integrating explicit programming abstractions, including producer-consumer, master-worker, and MapReduce, to block-based languages. We describe our reference implementation of adding the producer-consumer abstraction to Snap! and an educational project that utilizes this abstraction. This project clearly demonstrate the key features of parallel processing, without unduly burdening the programmer with the low-level details that this programming model typically entails. Our initial results show great potential in introducing the key concepts of parallel computing via block-based programming.