Re: non trivial constant folding

Mihai Christodorescu <mihai@cs.wisc.edu> wrote:
> 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.


As our moderator has already noted, whether changing exception
behavior is actually *wrong* depends on the definition of the
language. A number of languages say that the result in such a case is
ill-defined -- that whether or not it causes an exception is not
predictable, because left-to-right evaluation is not guaranteed.


Of course, there is a practical issue to contend with: users who do
not understand the language specification well enough to realize this.
They may still complain.


>...Of course, these are contrived examples, but a
>compiler has to work for all cases, not most of the time...


Except when the language spec says it doesn't have to.
--
When failure is not an option, success | Henry Spencer henry@spsystems.net
can get expensive. -- Peter Stibrany | (aka henry@zoo.toronto.edu)

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