On the Incomparability of Gamma and Linda

We compare Gamma and Linda, two of the most prominent coordination languages based on generative communication via a shared data space. In Gamma computation is obtained by applying multiset rewriting rules, reminiscent of the way chemical reactions happen in a solution. On the other hand, Linda permits interprocess communication by means of the creation and consumption of shared data. Also a non-blocking input operator is allowed: it terminates indicating if the required datum has been consumed or it is not actually available. We first recall two simple calculi based on Gamma and Linda coordination models. Even if both the languages are Turing powerful, we show that their expressive power is not comparable; in particular, we prove that there exists no program distribution preserving encoding of one language in the other that respects at least the input-output behaviour.