Uniform syntax

[certik]

Extracted from #49. Author: Van Snyder

Reference: Correspondence preceding 1986 Albuquerque meeting.

Uniform referential syntax to array elements, updaters and structure components increases
software mutability, and therefore decreases maintenance costs. (The alternative is to follow
Parnas’s advice to encapsulate everything in procedures — which provides uniform referential
syntax with higher labor expense and lower performance). To move in that direction, it ought to
be allowed to access a structure component using the syntax component(structure). This should
be classed as a generic updater. To avoid a conflict with an array of the same name as component an array ought to be considered to be a generic updater. Access to type-bound procedures or
procedure pointer components with PASS OBJ could require the structure to be first, or to be
in the same place in the reference as the corresponding dummy argument. For symmetry with
this syntax, array components should be referenced as component(structure,subscript,. . . ). A
small step in the direction of uniform syntax is to allow type-bound functions to be referenced
without parentheses enclosing the empty argument list. An anti-symmetric change to allow a
scalar to be referenced with an empty subscript list would make it impossible to know, and
impossible for the syntax to depend upon, whether a reference to an entity of a derived-type
object is a reference to a component or a type-bound function or updater. To complete the
ability to convert an entity between an array, structure component and updater, an integer
interval type, using the same constructor as a subscript triplet, is needed to allow an updater
reference with the same syntax as an array section reference.