Re: non trivial constant folding

Mihai Christodorescu <mihai@cs.wisc.edu> wrote:
> As far as I know, expression rewriting for optimization purposes is
> NP-complete. There are some compilers that perform some expression
> optimizations, but not really extensive, as you have noticed yourself.
> There is also another problem: the optimized expression can behave
> differently than the original expression, due to operations being executed
> in different order/with different operands. Consider:
> "MAXINT + x - 1"
> If x is 1, then evaluating MAXINT + x will overflow.


> If you optimize the expression to be:
> "the_value_of_MAXINT_minus_1 + x"
> (i.e. you evaluate MAXINT - 1 at compile time), then if x is 1, the
> expression will not overflow. Of course, these are contrived examples, but a
> compiler has to work for all cases, not most of the time (OK, at least in
> theory).


But what you suggest is perfectly fine in C. An overflow produces
undefined behavior and one of the allowed results is to not actually
overflow. What is not allowed is converting an expression from one
which cannot overflow to one which would.


    Michael

Return to the comp.compilers page.
Search the comp.compilers archives again.