[Lazarus] Large Integer errors

[Dave Coventry]

Thanks Howard.

If the result is never going to be more than maxint anyway, then I
don't have to do anything at all, that will be perfectly acceptable to
me as it would already be beyond what is usable.

What would screw me up would be if I just lose the topmost bit and it
all resets to zero again.

2009/2/14 Howard Page-Clark <hdpc at talktalk.net>:
> On Sat, 14 Feb 2009 17:56:50 +0000
> Dave Coventry <dgcoventry at gmail.com> wrote:
>
>> Yes, great, thanks.
>>
>> Would there be any error or would Lazarus just continue as if nothing
>> had happened?
>>
>>
>> 2009/2/14 Howard Page-Clark <hdpc at talktalk.net>:
>> > On Sat, 14 Feb 2009 12:30:55 +0200
>> > Dave Coventry <dgcoventry at gmail.com> wrote:
>> >
>> >> function mycalc( num1,num2: real):integer;
>> >> begin
>> >>   Result:=trunc(num1*num2);
>> >> end;
>> >>
>> >> How can I protect from the product of the two numbers being too
>> >> big?
>> >>
>> >> If it's too big I want to discard it anyway.
>> >
>> > function mycalc( num1,num2: real):integer;
>> >  begin
>> >   Result:=trunc(num1*num2);
>> >  if result > maxint then result := 0;
>> >  end;
>> >
>> > Would this do what you want?
>
> Actually the code I fired off is useless - since result is an integer
> it is never going to be more than maxint! What you need is
>
> function mycalc( num1,num2: real):integer;
> var tmp: int64;
>  begin
>   tmp :=trunc(num1*num2);
>   if (tmp <= maxint) then
>    result := tmp
>  else result := 0;
>  end;
>
> Errors will only be flagged up if you compile with -Co
> or tick the appropriate overflow check box in the compiler options
> dialog, Code tab, "Checks" panel.
>
> ----
> Howard Page-Clark <hdpc at talktalk.net>
> _______________________________________________
> Lazarus mailing list
> Lazarus at lazarus.freepascal.org
> http://www.lazarus.freepascal.org/mailman/listinfo/lazarus
>