The Linda model features a rather ignored primitive called "eval", meant to delegate simple context-aware expression evaluation to the tuple space in charge of mediating interactions.
Its unclear semantics and expressiveness reach pushed researchers toward novel definitions and/or interpretations of such a primitive.
TuCSoN, for instance, provides its own interpretation of the eval primitive, suitably defined and extended, called "spawn".
TuCSoN spawn primitive is meant to delegate coordination-related computations to the coordination medium, provided certain restrictions on expressiveness, for the sake of consistency and safety, are satisfied-e.g., a spawn can only perform coordination operations locally.
The thesis aims at surveying other existing implementation of the eval primitive, comparing them with the spawn, then advancing the spawn implementation so as to feature code-on-demand capabilities, necessary for mobile scenarios.
