Safety and power
Stavros Macrakis · ACM SIGSOFT Software Engineering Notes · 1982
Although the "question of safety vs. power in programming languages" is most interesting, Glass's letter [SEN 6:5 (Oct 81)] addresses only a peripheral issue, namely some (seemingly arbitrary) areas in which certain programming languages have required unsafe solutions.In order to focus discussion, let me propose a definition of safety: a practice is "safe" if it relies only on the language's semantics. Clearly, unsafe practices are in general not portable between compilers or machines. There appear to be two primary motivations for unsafe practices: efficiency and extra-linguistic operations.