Re: GCC optimizes integer overflow: bug or feature?

[Richard B. Kreckel]

Marcin Dalecki wrote:

> > But the same applies to floating point numbers. There, the situation 
> > is even better, because nowadays I can rely on a float or double 
> > being the representation defined in IEEE 754 because there is such 
> > overwhelming hardware support.
>
>
> You better don't. Really! Please just realize for example the impact 
> of the (in)famous 80 bit internal (over)precision of a
> very common IEEE 754 implementation...


Well, I know. Which is why I used the term "storage representation" down 
the rest of the paragraph that you decided not to quote. :)

> > > However it's a quite common mistake to forget how "bad" floats 
> > > "model" real numbers.
> >
> >
> > And it's quite a common mistake to forget how "bad" finite ints 
> > "model" integer numbers.
>
>
> No it isn't. Most people don't think in terms of infinite arithmetics 
> when programming.
> And I hold up that the difference between finite and infinite is 
> actually quite a fundamental
> concept.


Let us agree to disagree.

> However quite a lot of people expect the floating arithmetics rouding 
> to give them
> well behaved results.


I don't know what you mean by "well behaved" in this context, but I am 
beginning to suspect something: what you were trying to express with 
that algebra argument was a mere gut feeling of yours that there is a 
fundamental need to distinguish between discreet and continous domains. 
Let me just say that "algebra" has not much to do with that distinction 
and is off-topic. Anyway, you are entitled to feel that way or the 
other, but I cannot agree.

Leaving such gut feelings aside, I maintain: GCC has been careful not to 
violate a standard (IEEE 754) that was not binding, but assumed by many 
of its users, unless told otherwise (using -ffast-math). Now it is 
violating another standard (LIA-1) that is not binding, but assumed by 
many of its users, unless explicitly told to follow that standard (with 
-fwrapv). That is A Bad Idea.

Cheers
-richy.

--
Richard B. Kreckel
<http://www.ginac.de/~kreckel/>