Rechenberg

So-called software metrics, which quantify different properties of software products and processes, are used to measure software products. The software metric introduced by Rechenberg in 1986 defines differential parameters for the complexity of a software development and then adds these to a value known as overall complexity.

Rechenberg developed this relationship because, in his view, the approaches of Halstead and McCabe took insufficient account of the complexity of software. The methodology is generally regarded as very flexible due to the freely selectable weights for executable instructions. The consideration of most diverse aspects for example the nesting depth of instructions is to be regarded as advantageous. On the other hand, the time-consuming computation of the individual quantities is disadvantageous. By the complexity measure of computing mountain the aspects of the readability, the understanding as well as the changeability of software are considered.

Total complexity: CC = SC + EC + DC

The nesting depth of alternatives and loops is generally considered by this method by introducing an exponentially increasing factor f, which assumes typical - previously determined - values with increasing nesting depth.

f = 1, 1.5, 2.25, 3.38, 5.06 ...

Instruction complexity SC: Is the sum of the instruction complexities of all instructions. The individual weighting factors are e.g. for a value assignment = 1, Goto = 5, procedure call = 1+ parameter number, While/For control structures = 3, ...

Example: SC(if b then S) = 2 + f x 1 = 3.5

Expression complexity EC: Is the sum of the expression complexities of all expressions. First, the complexity of each part of the expression under consideration is explicitly weighted. Thereby the nesting depth of the expressions is evaluated by the weighting factors. The weighting factors define for +/- =1, MOD=3, indexing=2, AND/ OR=3, MUL/DIV=2 as well as dereferencing=2.

Example: EC(2 x(i+1)) = 2+f x EC(i+1) = 2+fx1 = 3.5

Data complexity DC: Is the sum of the data complexities of all identifiers. First, all identifiers are weighted according to the distance between declaration and usage. As weighting factors local variables receive the value 1, formal parameters the value 2 and global variables the value 3. From the sum of the weighting factors then finally the data complexity( DC) results.