Re: Possible Bug in bitlib under Windows?

[KHMan <keinhong@...>]

Ralph Hempel wrote:
> Andrew Gorges wrote:
>
> > print(bit.band(3, 0xFFFFFFFF))
> > -- displays 0
>
> What does bit.band( 3, 1 ) result in?
>
> Are we seeing the result of truncation and endianness issues
> of what we assume is a 32 bit value that is actually a 64
> bit value when compiled?

Just got a copy of Lfw 5.1.4.20. The problem is probably in 
TOBIT() in lbitlib.c. It looks rather complex, I haven't tried 
picking it apart yet...

Using bit.cast to test TOBIT() by itself:

> print(bit.cast(0x7fffffff))
2147483647
> print(bit.cast(0xffffffff))
-2147483648

Which is supposed to be -1 with 32-bit ints. 0xffffffff is 
correctly represented in Lua, since:

> print(0xFFFFFFFF)
4294967295

So, something wrong is happening when TOBIT() tries to convert the 
double to an integer.

BTW, I think the LfW page should also point to the Google project 
for regular downloads, since it is much faster.

--
Cheers,
Kein-Hong Man (esq.)
Kuala Lumpur, Malaysia