For example, the following statement or some variation of it can be found in many programs: x = x + 1 This means "find the number stored in the variable , add 1 to it, and store the result of the addition in the variable ." As simple as this seems, it may have an inefficiency, in that the location of variable has to be looked up twice if the
compiler does not recognize that two parts of the expression are identical: might be a reference to some array element or other complexity. In comparison, here is the augmented assignment version: x += 1 With this version, there is no excuse for a compiler failing to generate code that looks up the location of variable just once, and modifies it in place, if of course the machine code supports such a sequence. For instance, if x is a simple variable, the
machine code sequence might be something like Load x Add 1 Store x and the same code would be generated for both forms. But if there is a special op code, it might be MDM x,1 meaning "Modify Memory" by adding 1 to x, and an optimizing compiler would generate the same code for both forms. Some machine codes offer INC and DEC operations (to add or subtract one), others might allow constants other than one. More generally, the form is x
?= expression where the stands for some operator (not always ), and there may be no special op codes to help. There is still the possibility that if is a complicated entity the compiler will be encouraged to avoid duplication in accessing , and of course, if is a lengthy name, there will be less typing required. This last was the basis of the similar feature in the
ALGOL compilers offered via the
Burroughs B6700 systems, using the tilde symbol to stand for the variable being assigned to, so that would become and so forth. This is more general than just x:=~ + 1; Producing optimum code would remain the province of the compiler. ==Semantics==