[Libreoffice] "safe" numeric casts with explicit min/max

On Thu, Nov 24, 2011 at 11:25:30AM +0100, Lionel Elie Mamane wrote:
> On Thu, Nov 24, 2011 at 10:56:57AM +0100, Lionel Elie Mamane wrote:

> > I understand we don't want to touch that in stable branch, but anyone
> > opposing me pushing something like that in master, and similar for
> > llimit_cast and limit_cast?

> > // assumptions:
> > // \forall typename Type:
> > //     ( (::std::numeric_limits< Type >::min() == 0) XOR ( ::std::numeric_limits< Type >::min() 
> > < 0  && ::std::numeric_limits< Type >::is_signed ))
> > // &&  (::std::numeric_limits< Type >::max() > 0 )

> > template< typename ReturnType, typename Type >
> > inline ReturnType ulimit_cast( Type nValue, ReturnType nMax )
> > (...)
> >   else if ( ! ::std::numeric_limits< ReturnType >::is_bounded ||
> >             ::std::numeric_limits< ReturnType >::max() > ::std::numeric_limits< Type >::max() )

> That should be:

>   else if ( ! ::std::numeric_limits< ReturnType >::is_bounded ||
>             nMax > ::std::numeric_limits< Type >::max() )

And even then it is buggy, because I was thinking only of the case
where nMax is ::std::numeric_limits< ReturnType >::max(), but it can
be an arbitrary (possibly negative!) value.

That thing is devilishly hard to get right... I think I have a correct
version now, but I'm going to unittest the hell out of it before I
commit, that's for sure!

-- 
Lionel