Re: Problem with optimizations

[johnscpg@googlemail.com]

On Apr 28, 8:15 pm, Olivier Scalbert <olivier.scalb...@algosyn.com>
wrote:
> Albrecht Käfer wrote:
> > (see below) schrieb::
> >> There is no else part for the following if:
>
> >>             if Is_Solved(Cube) then
> >>                 New_Line;
> >>                 Put("Solved !"); New_Line;
>
> >>                 for i in 1..depth-1 loop
> >>                     Ada.Integer_Text_IO.Put(Integer(Moves(i)));
> >>                 end loop;
> >>                 Result := True;
> >>             end if;
>
> >> So the procedure can exit with Result undefined. Since the program is
> >> incorrect, varying optimisation, or implementation details, is likely to
> >> give varying results. Some may co-incidentally be the results you expect.
>
> > Shouldn't that, you know, create a warning or something?
>
> > Albrecht
>
> That is what I was thinking, but with:
>
> gnatmake -f -O3 -W -gnatp rubikmain
>
> it detects the bug (and also an other one in the same function)
>
> Shame on me !

I also learned this recently using gfortran.
You need the -O flag along with the -W to
get the gcc compiler to report uninitialized
variables.  As you've shown, it spots possible
uninitialized variables in  if-then-else
blocks or for-loops.  It mostly speculates - its
usually wrong, but I like it. In gfortran
you can limit the -W with: -O -Wuninitialized.
Don't know about gnatmake, but
gnatmake  accepts it.

cheers
jonathan

(This will be my 3rd and final attempt to post this -
hope 3 of these posts don't eventually appear,
but I can't seem to get gmail to let me through.
It sure likes spam tho' ;)