2012/2/28 Paul Eggert <eggert@cs.ucla.edu>

> > There is something wrong around lib/strftime.c:946
> >           if (negative_number)
> >             u_number_value = - u_number_value;
> >
> > u_number_value being unsigned, this is wrong.
>
> I don't see anything wrong there.  The value of
> -X is well-defined if X is unsigned int; it's
> equivalent to ~X + 1.  Can you supply a test case
> illustrating the actual bug?
>
> Please bear in mind that the code is already
> tested to LP64 hosts.  Any bug related to this
> particular code would also be a bug on LP64 hosts,
> no?  So, it's not likely that any bug here is specific
> to Windows 64.  If there is a bug, it should be
> describable and fixable independently of Windows 64.


I don't see things that way. Some things that would pass on LP64 because
ints are 64 bits may turn into a crash because of sign extension on Windows
64
promoting 32bits ints to 64bits ints. So I'm definitely more confident when
ints are used
without any abuse of any kind. I don't see the point with using an unsigned
int to negate
it.

-- 
Fabrice