Eden is a concurrent functional programming language intended to develop a new perspective on parallel programming: give programmers enough fine-grained control to implement parallel algorithms efficiently while at the same time avoid requiring them to deal with the low level details of process management. An Eden program defines a system of processes which exchange data on communication channels. Eden requires explicit specification of processes and their incoming and outgoing data, but frees the programmer from having to deal with the actual transfer of data between processes and the necessary synchronisation.
Process communication channels are modeled by head-strict lazy lists , in a manner similar to the way stream-based I/O is usually handled. Eden extends the lazy functional language Haskell but enforces strit evaluation semantics when necessary to support parallelism.
Eden has been jointly developed by groups at Philipps Universität Marburg, Germany and Universidad Complutense de Madrid, Spain.
Code sample
The following function is at the heart of a simple ray-tracer program. It computes an image with y lines and x columns of a scene consisting of spheres. The sequential function body of the ray function is simply the expression map (traceLine x world) [0..y-1]. The parallel version produces the image by several processes each computing a chunk of lines:
ray :: Int -> Int -> Int -> [Sphere] -> [[RGB]]
ray chunk x y world
= concat ([process (map (traceLine x world)) # linenumbers
| linenumbers <- splitAtN chunk [0..y-1]]
`using` spine)
The function concat flattens a list of lists into a list, thus removing one level of nested lists -- the one introduced by the list of processes. The addendum `using` spine is needed to produce early demand for the evaluation of the process instantiations.
External links
References
- Silvia Breitinger, Rita Loogen, Yolanda Ortega-Mallén, Ricardo Peña: Eden - Language Definition and Operational Semantics, Technical Report 96-10, Reihe Informatik, Fachbereich Mathematik und Informatik, Philipps Universität Marburg 1998.