Confining imperative languages for parallel processing

Imperative languages have built-in notions of global updatable variables. This has been claimed to be one of the major reasons why they are inferior to functional languages for parallel processing. The paper examines the `bilateral´ effects-side effects and aliasing-of updatable variables. While aliasing hinders code parallelization, side effects do not necessarily cause problems in parallel processing if a certain discipline, e.g. a high-level dataflow scheme, known as the LGDG computation, in their use is obeyed. Furthermore, a point is made for the exclusion of nested procedure declarations from parallel imperative languages, since they may cause serious aliasing and degrade locality of reference