Re: [AD] C99 isms in test.c

[David Schwingenschlögl <indeterminatus_ml@xxxxxxxxxx>]

Hi all,

comments below.

Evert Glebbeek wrote:
> > >     ct = 0;
> > >  
> > >     do {
> > > -      z = fixmul(x,y);
> > > +      z += fixmul(x,y);
> > >        x += 1317;
> > >        y += 7143;
> > >        ct++;
> > >       
> > Why are these changes needed? Don't x and y change every iteration?
> >     
>
> Yes, but z is never used. A clever compiler will notice this and remove the 
> `redundant' calculation.
> I haven't actually checked if it does this in this case, but I've seen it 
> doing that in similar cases and I thought it best to err on the side of 
> caution.
>
> Evert
>   
While I don't want to argue about the relevance of this piece of code, 
I'd rather like to suggest to comment on the changes. I don't know how 
you guys handle commenting, but still, the x += and y+= lines of code do 
not comment on themselves and were added with some assumptions about the 
compiler in mind - something that would nicely go along with something 
like "Both x and y have to be changed so that a clever optimizer does 
not optimize away z".

Just my two cents ;-)

Yours,
 David.