EWD249 - 60

the interrupted program!)

Ten years later, when ALGOL 60 emerged, the scene changed and we did not talk any more about closed subroutines: we called them "procedures" instead. They remained to be appreciated by the programmer as a very handy means for shortening the program text, and more and more programmers started to use them for the purpose of structuring, so that program adaptation to foreseen changes in problem specification could be confined to the replacement of one or more procedure bodies, or to a procedure call with some actual parameters changed. But the main novelty was the concept of the local variables.

This was reflected in two important aspects. The first one was the concept of "scope", i.e. the idea that not all variables are homogeneously accessible all through the program: local variables of a procedure are inaccessible from outside the procedure body because outside it they are irrelevant, What local variables a procedure needs to do its private task is its private concern, is no concern of the calling main program and the fact that the main program can (and must!) be conceived independently of these local variables is judiciously reflected. We may have some misgivings about the specific scope rules, as embodied in ALGOL 60, but we should appreciate them as a very significant step in the right direction.

The second aspect of the novelty was given by the fact that procedures could be used recursively, more precisely, that a procedure was allowed to call itself, either directly or indirectly. The virtue of this facility has been the subject of many hot debates; as far as I can see the discussion has died down. The argument against recursive procedures was always an efficiency argument: non-re-entrant code could be executed so much more efficiently. But with the advent of multiprogramming another need for flexible storage allocation has emerged. And if there are still machines in which non-re-entrant code can be executed much more efficiently, i.e. in which the use of recursive routines is punished by too heavy a penalty, then I would venture the opinion that the structure of such a machine should now be called somewhat old-fashioned. The recursive procedure, however, forced upon us the recognition of the difference between its (static) text and its (dynamic) activation -its "incarnation" as it has been called. The procedure text is one thing; the set of local variables it operates upon this time is quite another matter.