This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: g++ optimization issue??


On Fri, 2004-12-17 at 08:19 -0500, Robert Dewar wrote:
> ColinLaney wrote:
> > Thanks for the link. Looks like you've been discussing this quite a bit.
> > I guess it boils down to the philosophical question whether the compiler
> > should *ever* generate code that give ambiguous results even if the
> > problem's source is elsewhere. 
> 
> I am not sure what you mean by ambiguous results.
I meant that in the same binary I get different results for the same
code. In my example code this has to do with excess precision - yes,
*but* it could have been avoided, say by simply doing a fildl every
time. Like I said earlier I understand that IEEE representation issues
cannot be fixed but why can something like excess precision not be
minded?

>  In any case the
> only real absolute responsibility of the compiler is to generate
> correct code as defined in the standard.
Fair enough.

> 


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]